Total colour management in textiles
Total colour management in textiles Edited by John H. Xin
Cambridge England
Published by Woodhead Publishing Limited in association with The Textile Institute Woodhead Publishing Limited Abington Hall, Abington Cambridge CB1 6AH, England www.woodheadpublishing.com Published in North America by CRC Press LLC 6000 Broken Sound Parkway, NW Suite 300, Boca Raton, FL 33487, USA First published 2006, Woodhead Publishing Ltd and CRC Press LLC © 2006, Woodhead Publishing Ltd The authors have asserted their moral rights. This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. Reasonable efforts have been made to publish reliable data and information, but the authors and the publishers cannot assume responsibility for the validity of all materials. Neither the authors nor the publishers, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording, or by any information storage or retrieval system, without permission in writing from Woodhead Publishing Limited. The consent of Woodhead Publishing Limited does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. Library of Congress Cataloging in Publication Data A catalog record for this book is available from the Library of Congress. Woodhead Publishing ISBN-13: 978-1-85573-923-9 (book) Woodhead Publishing ISBN-10: 1-85573-923-2 (book) Woodhead Publishing ISBN-13: 978-1-84569-108-0 (e-book) Woodhead Publishing ISBN-10: 1-84569-108-3 (e-book) CRC Press ISBN-10: 0-8493-9207-1 CRC Press order number: WP9207 The publishers’ policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured from pulp which is processed using acid-free and elementary chlorine-free practices. Furthermore, the publishers ensure that the text paper and cover board used have met acceptable environmental accreditation standards. Typeset by SNP Best-set Typesetter Ltd., Hong Kong Printed by TJ International Limited, Padstow, Cornwall, England
Contents
Contributor contact details
ix
Introduction
1
J H Xin, The Hong Kong Polytechnic University, Hong Kong
Part I Measuring colour
5
1
Colour perception
7
1.1 1.2 1.3 1.4 1.5 1.6
Introduction The nature of colour The physical basis of colour The human colour vision system Colour perception References
7 7 8 10 17 20
2
Colour description/specification systems
22
B Rigg, University of Leeds, UK
2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
S Westland and V Cheung, University of Leeds, UK
Introduction Basic facts Additive and subtractive mixing The CIE system of colour specification Calculation of tristimulus values from Rl values measured at 10 or 20 nm intervals Relationships between tristimulus values and colour appearance Usefulness and limitations of the CIE system Colour order systems Colour specifiers
22 24 26 27 36 36 38 40 41
vi
Contents
2.10 2.11
Future trends References
42 42
Instrumental colour measurement
44
3
P J Clarke, The Tintometer Ltd, UK
3.1 3.2 3.3 3.4 3.5 3.6 3.7 4
Introduction Types of colour measurement Colour measuring instrumentation Inter-instrument agreement and traceability Future trends Sources of further information and advice References
44 44 47 53 54 55 56
Colour quality evaluation
57
M R Luo, University of Leeds, UK
4.1 4.2 4.3 4.4 4.5 4.6
Introduction Colour difference formulae Metamerism Colour constancy Conclusions and future trends References
57 58 68 69 73 74
5
A practical guide to visual evaluation of textile samples
76
K Butts, Datacolor, USA
5.1 5.2 5.3 5.4 5.5 5.6 5.7
Introduction The components of colour perception Industrial guidelines for visual colour assessment Practical application of visual colour assessment methods Future trends Sources of further information References
76 77 83 86 90 92 92
Part II Managing colour
95
6
97
Colour simulation of textiles H Shen and J H Xin, The Hong Kong Polytechnic University, Hong Kong
6.1 6.2 6.3
Introduction Characterisation of colour displays Colour mapping for two-dimensional texture image
97 98 99
6.4 6.5 6.6 6.7 7
Contents vii Texture effect on visual colour difference evaluation Colour synthesis for three-dimensional objects Future trends and further information References
103 108 113 115
Effective colour communication from mind to market
117
G Littlewood, Datacolor, UK
7.1 7.2 7.3
7.10 7.11
Introduction The ‘fast fashion’ concept and its effect on colour Colour palette development as part of the whole product development process Review of existing ‘manual’ communication methods between design and production and why things go wrong Best practice in communicating between design and production – human and technological considerations Creating the standard Colour approval – where is it done? Colour approval – how is it done? Electronic colour communication programmes – associated considerations and options Electronic tracking and reporting packages Future trends and conclusion
133 134 134
8
Controlling colourant formulation
136
7.4 7.5 7.6 7.7 7.8 7.9
117 118 120 121 124 127 130 131
J H Xin, The Hong Kong Polytechnic University, Hong Kong
8.1 8.2 8.3 8.4 8.5 8.6
Introduction Colourant recipe formulation Improvement of the formulation accuracy A case study for matching a target using a commercial colour recipe formulation system Sources of further information and future trends References
136 137 147 152 156 157
9
Controlling digital colour printing on textiles
160
J R Campbell, Iowa State University, USA
9.1 9.2 9.3
Introduction Characteristics and variables of digital ink jet printing (DIJP) Design potential and limitations of digital textile printing
160 168 176
viii
Contents
9.4 9.5 9.6 9.7 9.8
Role of end output: artist and industry approaches Ensuring accuracy and uniformity Future trends Sources of further information and advice References
178 179 188 189 190
Colour management across the supply chain
191
10
R Lawn, Consultant, UK
10.1 10.2 10.3 10.4 10.5 10.6
Introduction Colour supply chains Supply chain colour process requirements Future trends Conclusions Further reading
191 191 194 205 209 209
11
Quality assurance management for coloured goods
210
M S Ball, Consultant, UK
11.1 11.2 11.3 11.4 11.5 11.6 11.7
Reproduction of colour Instrumental or computer recipe prediction Colour variation evaluation and monitoring Colour performance Future trends Notes and references Sources of further information
210 213 216 220 225 226 227
Index
229
Contributor contact details
(* = main contact)
Introduction
Chapter 2
Professor John Xin Institute of Textiles and Clothing The Hong Kong Polytechnic University Hung Hom Kowloon Hong Kong Tel: +852 2766 6474
Professor Bryan Rigg Hill View House Hillington Road White Grit Minsterly Shropshire SY5 0JL UK Email:
[email protected]
Email:
[email protected]
Chapter 1 Professor Stephen Westland* and Dr Vien Cheung Centre for Colour Design Technology School of Design, University of Leeds Leeds LS2 9JT UK Email:
[email protected] Email:
[email protected] Tel: +44 (0)113 343 3752 Fax: +44 (0)113 343 3704
Chapter 3 Dr Peter Clarke The Tintometer Ltd Waterloo Road Salisbury SP1 2JY UK Tel: +44 (0)1722 327242 Fax: +44 (0)1722 412322 Email:
[email protected]
Chapter 4 Professor M. Ronnier Luo Department of Colour and Polymer Chemistry University of Leeds ix
Contributor contact details
Leeds LS2 9JT UK
Tel: +44 (0)161 923 0254/07831 615985 Email:
[email protected]
Tel: +44 (0)113 343 2763 Fax: +44 (0)113 343 2947 Email:
[email protected]
Chapter 8
Chapter 5 Mr Kenneth Ray Butts 7638 Sedgebrook Drive Stanley, NC, 28164 USA Tel: +1 704 827 0990 Email:
[email protected]
Chapter 6 Dr Hui-liang Shen and Professor John Xin* Institute of Textiles and Clothing The Hong Kong Polytechnic University Hung Hom Kowloon Hong Kong Tel: +852 2766 6474 Email:
[email protected]
Chapter 7 Mr Glenn Littlewood Datacolor 6 St George’s Court Dairyhouse Lane Broadheath Altrincham Cheshire WA14 5UA UK
Professor John Xin Institute of Textiles and Clothing The Hong Kong Polytechnic University Hung Hom Kowloon Hong Kong Tel: +852 2766 6474 Email:
[email protected]
Chapter 9 Mr J. R. Campbell Research Fellow, Centre for Advanced Textiles Glasgow School of Art 167 Renfrew St Glasgow, G3 6RQ Scotland, UK and Associate Professor – Textiles and Clothing Iowa State University 1078 Lebaron Hall Ames, IA 50010 USA Tel: +1 515 294 0945 Email:
[email protected]
Contributor contact details
Chapter 10
Chapter 11
Mr Richard Lawn The White House 59 New Street Shrewsbury SY3 8JQ Shropshire UK
Mr Malcolm Ball PO Box 327 Macclesfield SK10 5XX UK
Email:
[email protected]
Tel: +44 (0)1625 572284 Email:
[email protected]
xi
Introduction J H X I N, The Hong Kong Polytechnic University, Hong Kong
This book is intended to provide a basic yet comprehensive understanding of many aspects of colour management with industrial applications in mind. It is suitable for individuals from a variety of disciplines and organisation levels in colour management, colour quality monitoring and evaluation, textiles, fashion, design, graphic printing, desktop publishing, dyestuff, paint, ink, plastics, cosmetics, food and beverages, etc. The list includes colourists, colour quality assessors, colour co-ordinators, graphic designers, merchandisers, product development specialists, academics or anyone who uses colour in their work. It is also suitable for university students in textiles, fashion, design, fine arts, or any other colour-related courses. This book has two parts: Part I – Measuring colour and Part II – Managing colour. There are a total of 11 chapters, Chapters 1–5 in Part I and Chapters 6–11 in Part II. Chapters are written by experts in their relevant fields covering subjects from colour science, colour quality evaluation, colour simulation, colour communication and colourant formulation, to colour management in the supply chain and colour assurance systems. A number of figures are reproduced additionally in colour in a four-page colour section. Chapter 1 covers colour perception, which explains the nature of colour, the physical basis of colour, the human colour vision system and colour perception, as ‘colour exists only in the mind; it is a perceptual response to light that enters the eye either directly from self-luminous light sources or, indirectly, from light reflected by illuminated objects’. Chapter 2 describes colour specification systems which include the CIE system of colour specification, colour order systems and colour specifiers. ‘The colour of an object depends on many factors, such as lighting, size of sample, and background and surrounding colours. Much more importantly, colour is a subjective phenomenon and depends on the observer. The measurement of subjective phenomena, such as colour, taste and smell, is obviously more difficult than that of objective phenomena such as mass, length and time.’ ‘When assessing the usefulness of the colour specification system,
Total colour management in textiles
the reader should carefully consider how far the system enables us to deal with real problems with respect to colour and how far the system fails to deal with the subjective nature of colour.’ Chapter 3 focuses on instrumental colour measurement. ‘Colour measurements are essentially measurements of light shining through an object or light reflected from an object.’ The measurements in terms of diffuse and regular reflectance and transmittance are explained and so are the optical configurations and standard or recommended geometries for making these measurements. Chapter 4 explains colour quality evaluation. It is well understood in the colour-related industries that ‘the most important use of the measurement data is undoubtedly for colour quality control in terms of colour difference (DE) between a pair of samples.’ In this chapter, the colour difference formulae CIELAB, CMC (l:c), CIE94(KL:KC:KH), and the most recent CIEDE2000 (KL:KC:KH) are discussed in detail. These are followed by the set-up of industrial colour tolerance, observer uncertainty, the phenomenon of metamerism, colour constancy, as well as the calculation of the colour inconstancy index. Chapter 5 is a practical guide to visual evaluation of textile samples. Visual colour evaluation is ‘the most straightforward method of assessing the quality of one colour versus another.’ Visual colour evaluation must be standardised as much as possible so as to minimise the subjectivity that exists in individuals. This chapter discusses various aspects of the visual colour evaluation process including the illuminant, the object, and the observer. Industrial guidelines for visual colour assessment including AATCC and ASTM D1729 methods are listed and explained. In the latter part of the chapter, the practical application of visual colour assessment methods is discussed in detail. Chapter 6 ‘Colour simulation of textiles’ is the first chapter of Part II: ‘Managing colour’. This chapter discusses the accurate colour simulation on display devices including solid colour and colour samples with texture structures, such as textile fabrics. The colour synthesis technique for threedimensional textile products, which is based on a physical vision model, is also presented. Colour mapping algorithms for two- and three-dimensional textile fabric and textile products are introduced. The texture effect on visual colour difference evaluation is also investigated. Chapter 7 discusses effective ways of colour communication. The importance of colour within today’s retail environment is examined, and historic and current practices for colour communication between the specifier and the supply chain are critically reviewed. This is followed by an analysis of current ‘best practice’ in the chain together with associated impacts and benefits. It concludes with a proposed forecast of how colour could be managed in the future.
Introduction
Chapter 8 is on controlling colourant formulation. ‘Computer colourant formulation has been widely applied, especially when supplying coloured articles to companies with global sourcing practice thanks to the formidable advance of computers, especially personal computers. This greatly improves the lead-time for colour matching, especially when experienced colourists are not available. It has become a necessity for a modern dyehouse to install a computer colourant formulation system.’ This chapter explains the widely adopted Kubelka–Munk theory and the practices for computer colourant formulation of textile materials. It also discusses recipe correction methods, ways of improving formulation accuracy, and the use of artificial neural networks in colourant formulation. Chapter 9 discusses digital colour printing on textiles. ‘If the same print design were digitally printed on a variety of different printers using the same type of ink sets and fabric, the color results would vary widely.’ ‘Factors such as environmental conditions, ink properties and print head construction can cause results to vary from day to day using exactly the same printer and inks. This can be very troublesome and frustrating for print providers.’ Therefore, this chapter focuses on the multifaceted factors that users of digital textile printing technology must address in order to control the application of colour to digitally printed fabrics. Chapter 10 concentrates on colour management across the supply chain. It explains the structure of a typical colour critical supply chain in the apparel industry. ‘This industry is chosen as an example because colour is so important there and garment supply chains are unusually complex, but the general points apply to plastic components, supply chains for the automotive or electronic industries, for example, and to many other supply chains.’ Requirements of the supply chain colour process, methods of colour communication throughout the supply chain versus those requirements, the role of the application service provider (ASP) model, best practice for supply chain colour management, and future trends are discussed in detail. Chapter 11 is on quality assurance management for coloured goods. It discusses quality assurance issues in the reproduction of colour, which includes traditional identification of colour attributes and recipe/reproduction forecasting, the use of instrumental or computer recipe prediction, colour variation evaluation and monitoring, and colour performance which includes fitness for purpose and testing methodology. The editor hopes that this book, through the comprehensive discussion of some essential elements, will provide useful guidance for total colour management in most colour-related industries.
Part I
Measuring colour
1 Colour perception S W E S T L A N D A N D V C H E U N G, University of Leeds, UK
1.1
Introduction
Colour exists only in the mind; it is a perceptual response to light that enters the eye either directly from self-luminous light sources or, indirectly, from light reflected by illuminated objects. The nature of light and the spectral reflectance properties of objects are therefore described in the first part of this chapter. The second part of the chapter is concerned with the physiology and functional properties of the retina in the human eye. Light that enters the eye is sampled by three classes of light-sensitive cells in the retina known as cones. In order to understand colour, it is necessary to appreciate that the effective spectral sensitivities of these cones are not static; rather, they change with the illumination conditions and the responses of spatially neighbouring cells, to name but two factors. Furthermore, the three classes of signals from the cones are processed by the neural pathways that lead from the retina to various areas of the cortex in the brain. Although our understanding of colour processing in the human visual system is sufficient to allow us to predict when two spectrally dissimilar objects will be a visual match, it does not allow us to make reliable predictions of colour appearance. In this chapter, three current problems for the science of colour vision are described: colour contrast, colour constancy and colour appearance.
1.2
The nature of colour
Light is a form of energy. Specifically, it is that part of the spectrum of electromagnetic radiation that our eyes are sensitive to. Radio waves and X-rays, as well as ultraviolet and infrared radiation, are all part of the spectrum of electromagnetic radiation but the human visual system is only capable of sensing a very narrow band of wavelengths in the approximate range 360–780 nm (a nanometer is 10-9 metres). The light from any source can be usefully described in terms of the relative power emitted at each wavelength in the visible spectrum. Figure 1.1 shows the wavelengths of the
Total colour management in textiles Gamma rays
X-rays
Ultraviolet rays
Infrared rays
Radar
Broadcast bands
AC circuits
Visible light
400 nm
Higher energy
Wavelength
700 nm
Lower energy
1.1 The electromagnetic spectrum. Only radiation in the range 360– 780 nm is visible to the human eye.
visible spectrum and the colours with which we normally associate the wavelengths. However, Newton was famously aware that ‘the rays are not coloured’. By this phrase Newton meant that light is not intrinsically coloured; short-wave light, for example, has no intrinsic property by which it is blue but, rather, it may induce in us the sensation of blueness. Under some circumstances, however, short-wave light may appear black or some colour other than blue. It is therefore clear that colour cannot be understood without a study of the properties of the human visual system, since colour exists only in the brain. The spectral power distribution of daylight varies with geographical position, atmospheric conditions, and with the time of day and year but the set of daylight power distributions is very similar to that emitted by a blackbody1 heated at different temperatures (Judd et al., 1964). For many light sources it is useful to refer to the temperature (usually expressed in Kelvin) of the blackbody whose radiation most closely resembles that of the light source. This temperature is called the correlated colour temperature (Sinclair, 1997). The radiation of north sky daylight on a cloudy day has a correlated colour temperature of about 6500 K, whereas the light from a tungsten filament bulb has relatively more power at the long wavelengths, which gives it a much lower correlated colour temperature.
1.3
The physical basis of colour
When light strikes an object, some light is always reflected from the surface, at the boundary between the object and air, because of the change in refractive index as the light passes from air to a more dense medium. This surface reflectance has the same relative spectral power distribution as the illumi1 A blackbody is a hollow heated chamber with a small hole; as the blackbody is heated, the spectral power distribution of the light emitted from the hole varies.
Colour perception
nating source and may be diffuse or specular in nature. Diffuse reflectance, where the light is dispersed in many different directions, occurs when the surface is rough, whereas smooth glass-like surfaces give rise to specular surface reflectance where the angle of reflection is equal to (but with opposite sign) the angle of incidence of the illumination. The light that is not reflected at the surface enters the body of the object, where further interactions take place. If the material is transparent, some light will pass through the material and emerge at the other side. The most common processes that reduce transparency are absorption and scattering. Absorption is a process whereby light is removed by an interaction with the molecules of the object at an electronic level. Most objects are coloured because this absorption process is more efficient at certain wavelengths than at others, in a way that depends upon the properties of the molecules (Zollinger, 1999). Scattering is a kind of reflection that occurs when particles (or air bubbles) are present in the material. The amount and directional nature of the scattering depends upon the size of the particles and their refractive indices (relative to the medium in which they are contained). Many opaque (non-transparent) white materials are manufactured by adding particles of a white pigment such as titanium dioxide, which has a particularly high refractive index. Translucency is a visual phenomenon that can give materials a milky or cloudy appearance and occurs when the material is partially transparent but exhibits scattering. Further details about the physics of light and its interaction with materials is provided by Nassau (1983) and Tilley (2000) or, for an explanation at the level of quantum electrodynamics, Feynman (1990). The proportion of light reflected by a sample can be measured using a reflectance spectrophotometer and represents the (physical) colour fingerprint of the sample. A spectrophotometer typically measures the proportion (sometimes expressed as a percentage) of light reflected by the object at each of several equally spaced wavelength intervals. Commercially avail able instruments typically measure at 31 wavelength intervals centred at 400 nm, 410 nm, 420 nm, . . . , 690 nm, and 700 nm.2 Most reflectance spectra are smooth functions of wavelength so that it is reasonable to measure the reflectance at wavelength intervals of 5 nm or even 10 nm with little loss of information (Maloney, 1986). For non-fluorescent materials, the spectral reflectance factors are independent of the intensity or spectral distribution of the light source that is used by the spectrophotometer. That is to say, if 2 Many instruments extend the measurements to wavelengths shorter than 400 nm and/or longer than 700 nm. Although the spectral sensitivity of the visual system is usually given as 360–780 nm, at the very short and long wavelengths in this range we are not very sensitive and therefore whether an instrument extends to, say, 760 nm or 780 nm is not usually of great practical importance.
10
Total colour management in textiles
a given object reflects 50% of the light at a given wavelength, this is independent of whether the incident illumination contains 100 or 1000 units of power at that wavelength. The spectral reflectance factor is obtained by comparing the intensity of the reflected light for an object at a given wavelength with the intensity of the light reflected by a perfect Lambertian diffuser.3
1.4
The human colour vision system
The light that is reflected by objects or emitted by light sources enters the eye, where it may be absorbed by visual pigments in the photoreceptors, or cones, contained within the retina. The spectral sensitivities of the pigments in the three cone classes play a significant role in the nature of our colour perception. However, colour perception can only be fully understood if the processes that take place in the nervous system that transmits the retinal signal to the occipital lobe of the brain’s cortex are studied. A brief review of physiological processes that are important for colour vision is given in this section.
1.4.1 The human eye The eye is an approximately spherically shaped organ that contains an aperture and a light-sensitive inner lining called the retina. The aperture is at the front of the eyeball and allows light to enter where it can be focused by the lens onto the retina. The front of the eye is covered by a clear layer of tissue known as the cornea through which light must pass before it can enter the eyeball. The main function of the cornea is to protect the eye from injury; however, it also acts to refract the light so that it is focused appropriately at the retina4. About two-thirds of the focusing of light by the eye is carried out by the cornea (Meek, 2002). the lens – a crystalline structure that is suspended by the ciliary muscles (see Fig. 1.2) – being responsible for the remainder. The shape of the lens can be changed as it is squeezed by the ciliary muscles as a way of focusing a sharp image of the scene on the retinal layer that coats the inner surface of the eye. The iris can change size, so that the area changes from about 50 mm2 in dark conditions to about 10 mm2 in 3 Practically, since perfect Lambertian diffusers are difficult to manufacture and maintain, a white tile with known (the reflectance of the white tile, which is referred to as a secondary standard, is known relative to the perfect diffuser) spectral reflectance is used. 4 The cornea focuses light by being highly curved (more curved than the rest of the eye), possessing a smooth optical surface, and by being highly transparent.
Colour perception
11
Ciliary muscle Retina Lens
Fovea
Cornea Aqueous humour
Optic nerve Vitreous humour
1.2 Schematic diagram of the human eye.
bright sunlight. Although the pupil area can change by as much as a factor of 10 in response to the light intensity, in fact the range of illuminances in which the visual system operates covers many orders of magnitude (e.g. 10 lux in a darkened room to 100 000 lux in brilliant outdoor sunlight). Therefore, the change in pupil size can only play a minor role in the adaptation of the visual system to changes in light intensity. Light is focused onto the retina which includes specialised cells, known as rods and cones, that contain photopigments that undergo a chemical transformation when light of an appropriate wavelength is absorbed. In the rods, the photopigment is based upon the compound known as rhodopsin, and the events that occur when rhodopsin absorbs light have been studied by psychophysical, biochemical, physiological and, most recently, molecular techniques (Bowmaker, 2002). Rhodopsin consists of a protein called opsin and another molecule called 11-cis vitamin A aldehyde (also known as retinal). When rhodopsin absorbs light, the rhodopsin, molecule decomposes into opsin and vitamin A. Once a photon has been absorbed by rhodopsin, there is a change from the 11-cis isomer to the all-trans isomer as the terminal chain connected to opsin rotates. The protein then undergoes a series of transformations to eventually produce all-trans retinal and opsin. The bleached photopigment loses its colour and no longer responds to light. Regeneration of the coloured photopigment is very slow and takes
12
Total colour management in textiles
several minutes. Further details about the visual pigments are available in the literature (e.g. Bowmaker, 2002). The photopigments that are found in the rods and cones have sensitivity functions (see Fig. 1.3) that are approximately bell-shaped functions of wavelength (Stockman, MacLeod and Johnson, 1993). Under normal levels of illuminance (referred to as photopic vision) the responses of the rods become saturated and therefore normal colour vision is mediated by the responses of the cones. However, the rods perform an extremely useful function because at low levels of illuminance (referred to as scotopic vision) the cones do not respond at all and thus night vision is mediated by the rods. The three classes of cone are referred to as being short-, medium- and long-wavelength sensitive in relation to the wavelength of the peak sensitivity of the photopigments. The abbreviated terms L-, M-, and S-cone classes are often used. The spectral sensitivities of the extracted pigments have been measured and found to peak at 420 nm, 530 nm and 560 nm but the effective sensitivities of the cone classes are sometimes given as 440 nm, 545 nm and 565 nm because of absorption by the macula pigment in the retina and by the lens itself. Rushton (1965) emphasised that, when a photopigment molecule absorbs light, the effect is the same, no matter what the wavelength of the absorbed light might be (Wandell, 1995). Thus, even though a quantum of light at 400 nm possesses more energy than a quantum at 700 nm, the sequence of chemical reactions in response to absorption of a 400-nm quantum is identical to the response sequence to a 700-nm quantum. This important property
1
Spectral sensitivities
0.8
L M
S
0.6
0.4
0.2
0 400
500
600
700
Wavelength, nm
1.3 Spectral sensitivities of L (red), M (green) and S (blue) human cone classes (Stockman, MacLeod and Johnson, 1993).
Colour perception
13
is known as the principle of univariance; a photopigment makes a singlevariable response to the incoming light. The photopigment maps all spectral lights (whether single wavelength or broadband) into a single-variable output, the rate of absorption, and thus confounds the independent vari ables of intensity and wavelength. Consequently, under scotopic viewing conditions, when vision is mediated by a single photopigment in the rods, we cannot discriminate between lights of different spectral compositions that are equally bright. Scotopic vision is monochromatic; we simply see shades of grey. Note, however, that even under scotopic conditions the rod photopigment does not respond equally well to all wavelengths, the sensitivity being greatest for the middle wavelengths in the visible spectrum. The response of each cone class can be estimated by the integration over the visible spectrum of the spectral sensitivity of the cone’s photopigment and the spectral distribution of the energy in the colour signal that enters the eye. If we represent the reflectance P(l), the light source E(l) and the cone spectral sensitivities FL(l), FM(l) and FS(l) each by discrete functions at a number of wavelengths, then the cone responses can be computed by the following simple equations: L = ∑ E ( λ ) P ( λ ) Φ L ( λ ), M = ∑ E ( λ ) P ( λ ) Φ M ( λ ), S = ∑ E ( λ ) P ( λ ) Φ S ( λ ),
1.4.2 Colour vision It can be misleading to think about the spectral sensitivities of the cones as being static functions of wavelength. It is well known that processes of light adaptation take place in the retina (Wandell, 1995) and the visual system not only adapts to different light levels, it also adapts to colour (Hurvich, 1981). The effective sensitivity of the S cones, for example, will be reduced in the presence of short-wavelength radiation. These adaptive processes may well partly explain the phenomenon of colour constancy whereby it is believed that surfaces tend to maintain their approximate daylight appearance when viewed under a wide range of light sources. Colour constancy is surprising, at least at first consideration, since when the spectral properties of the illumination are changed (such as when we take an object from one room to another) the spectral distribution of the light that reaches the eye from an illuminated object can change quite markedly and yet the colour of the object remains almost constant. A further important process that occurs in the retina is that the responses of the cones are combined with each other to produce opponent signals (Kaiser and Boynton, 1996; Hurvich, 1981). The responses of the L and M
14
Total colour management in textiles L L+M
L–M M
Luminance
R–G
B–Y
S
S – (L + M )
1.4 Schematic diagram to show how the cone responses may be combined to generate a luminance channel and two opponent chromatic channels.
cones are additively combined to produce a luminance signal and subtracted from each other to produce a red–green signal. A blue–yellow signal is generated by subtracting the S cone signal from the sum of the L and M signals. Possible anatomical linkages are schematically represented by Fig. 1.4. Mathematically, we can express the opponent processing by equations such as: Lum = L + M, ORG = L - M, and OBY = S - (L + M) where Lum, ORG and OBY represent the opponent signals and L, M and S represent the responses of the cones5. The principles of additive mixing and the choice of additive primaries may be explained by the principle of univariance and the spectral sensitivities of the cones (Westland, 2002). Other phenomena of colour vision, however, require an understanding of opponent processing to be explained. For example, although additive colour mixing and trichromacy lead to the view that there should be three 5 Note, however, that this is only one possible way of combining the cone responses. There still remains uncertainty, for example, as to whether the short-wave sensitive cones contribute to the luminance signal.
Colour perception
15
colour primaries, psychologically it has been long known that there are four colour primaries, red, green, yellow and blue (Wandell, 1995; Hurvich, 1981). The facts that redness and greenness are opposite sensations (that do not occur at the same time) and that similarly yellowness and blueness are opposite sensations, stem from the opponent processing in the retina. Anatomically, the opponent processing is carried out in the retina as the cone responses are combined, first by horizontal and bipolar cells, and finally by retinal ganglion cells (Bowmaker, 2002). The retinal signals leave the retina and pass along the optic nerve away from the eye into the brain. The fibres from the two eyes meet and cross over at an anatomical structure known as the chiasma. The signals from the left-hand part of the visual field from both eyes project to the right hemisphere of the brain and those from the right-hand part of the visual field are processed in the left hemisphere. The retinal ganglion fibres terminate at the lateral geniculate nuclei or LGN (one LGN lies in the right hemisphere of the brain whereas the other lies in the left hemisphere), where a certain amount of further processing takes place, before long fibres known as optic radiations take the signals to an area in the lower rear (occipital) part of the cerebral cortex which is called the visual cortex.
1.4.3 Spatial vision It is important to understand that colour vision and spatial vision are inextricably linked. Invariably, when we observe a colour stimulus it is in the context of a surround and/or background (Hurlbert, 2002). In Fig. 1.5, for example, two physically identical grey-square stimuli do not have the same appearance because of their different backgrounds. The spatial processing
1.5 The two small grey squares are physically identical but do not have the same appearance because they are displayed against different backgrounds.
16
Total colour management in textiles
+
Excitation
+
–
Top view
–
+ –
–
Inhibition Side view
1.6 Three different representations of the spatially opponent receptive fields of a retinal ganglion cell.
that takes place in the human visual system therefore impacts upon how colour perception operates. That many retinal ganglion cells are spectrally opponent has already been introduced in the context of colour opponency. However, most opponent cells are also spatially opponent (Bartleson, 1984). The concept of spatial opponency is best explained in terms of receptive fields. The receptive field of a cell is the part of the visual field that influences the response of that cell. The receptive fields of retinal ganglion cells are quite large and this is because many cone cells contribute to the response of each ganglion cell. A given ganglion cell may be excited to respond by some cone cells and inhibited by others6. Typically, the pattern of excitatory and inhibitory connections of cones to a ganglion cell may cause the ganglion cell to be excited by an increase in light intensity in the centre of its receptive field but inhibited by an increase in light intensity at the periphery of its receptive field (Fig. 1.6). The spatial opponency of many retinal ganglion cells is undoubtedly the physiological basis for simultaneous-contrast effects such as that illustrated by Fig. 1.5. It also explains why the visual system is more sensitive to contrast in a scene rather than to absolute light intensity. 6 If a cone cell excites the ganglion cell then it means that a strong response of the cone cell causes a strong response of the ganglion cell. Conversely, cone cells that inhibit the ganglion cells cause the ganglion-cell response to reduce as the conecell response increases.
Colour perception
17
There is overwhelming evidence that the visual system performs a multiresolution analysis (in parallel) of images (Cambell and Robson, 1968; Wandell, 1995). What this means is that the visual system encodes multiple representations of an image for different colour directions (e.g. luminance, red–green, blue–yellow), spatial frequencies and temporal frequencies. The spatial-frequency response of the visual system as a whole can be quantified by the contrast-sensitivity function (CSF) (Campbell and Green, 1965). The CSF records the sensitivity of the three colour channels to contrast for gratings of varying spatial frequency. Measurements of CSF for the luminance channel show, for example, that the visual system is more sensitive to pattern (corresponding to a spatial frequency of about six cycles per degree of visual angle) than it is to spatially uniform fields7. For further details of the retinal and cortical processes and the interactions between colour vision, spatial vision and motion readers are directed to other sources (Wandell, 1995).
1.5
Colour perception
Colour perception for humans is three-dimensional, a fact that almost certainly stems from the existence in the retina of three different classes of light-receptive cells. Three terms, or numbers, are necessary and sufficient to define a colour stimulus for the visual system under standard conditions (Westland and Ripamonti, 2004). For example, we might describe a colour by its hue, its colourfulness and its brightness or we might specify a stimulus by the XYZ tristimulus values of the CIE system. It is important to note, however, that colour is only one aspect of total appearance for surfaces in the world and many other phenomena contribute to overall appearance, including gloss and texture. Furthermore, the complex nature of colour perception means that it is impossible to predict even the approximate colour appearance of a patch in a scene without specifying the surrounding colours and the state of adaptation of the eye.
1.5.1 Colour contrast The three-dimensional nature of colour may be explained by trichromacy and the existence of the L-, M-, and S-cone types but the notion of spatial colour contrast is essential to a modern understanding of colour perception. Figure 1.5 illustrates an example of lightness contrast, but the same type of 7 The luminance CSF is said to be band-pass in shape; that is the luminance channel has maximum sensitivity to contrast at about 6 cyc/deg and reduced sensitivity for increasing or decreasing spatial frequencies. The chromatic CSFs have been rather less-well studied but are believed to be low-pass.
18
Total colour management in textiles
phenomenon occurs for colour stimuli. Thus, a yellow patch viewed on a green background will appear more reddish than the same physical yellow patch on a red background. The simultaneous contrast effect illustrated in Fig. 1.5 can also be demonstrated temporally as successive contrast or afterimages. We often notice these contrast effects when we are presented with contrived displays such as Fig. 1.5, but both spatial and temporal contrasts are continually operative in our visual field during normal visual tasks. The colour appearance of a patch in a scene is dependent upon the context (both spatial and temporal) in which the patch is observed. A thorough review of spatial and contrast aspects of colour vision is provided by Hurlbert (2002). For certain stimuli, the colour of a patch takes on the hue of the surrounding background rather than contrasting with it and this effect is known as colour assimilation (Ripamonti and Gerbino, 2001). Since the effects of contrast and assimilation are not yet entirely understood, it is difficult to predict colour appearance and this issue is discussed in the following two sections.
1.5.2 Colour constancy Spatial colour contrast effects can cause otherwise physically identical colour stimuli not to have the same colour appearance. Similarly, physically disparate stimuli under certain circumstances can have the same appearance. This latter phenomenon is one consequence of colour constancy, one of the great mysteries of colour vision. How is it that our colour perception of surfaces in the world remains approximately constant as we move from a bright scene to a dimly lit scene and even from a room illuminated by one coloured light source to a light source of another colour? The phenomenon of colour constancy is a critical functional component of our colour vision because it enables us to discount changes in the light source and so recognise objects in the world by their colour. It is surprising, nonetheless, because the changes in the colour signal (the spectral power distribution that results when an object is illuminated by a light source), which occur when we change the illumination, can be substantial and therefore would not seem trivial to discount. For example, the absolute intensity of the light reflected by a black object outdoors on a sunny day may be greater than the intensity of the light reflected by a white object in a dimly lit room, but the objects appear black and white, respectively, under both illumination conditions. A number of approaches have been proposed for the way in which the visual system can discount the effect of the illumination (Finlayson, Drew and Funt, 1994; Hurlbert, 2002; Foster, 2003). The retinal process of chromatic adaptation may play a role in enabling colour constancy, since a consequence of this process is that the spatial-average output of a cone class
Colour perception
19
is likely to be similar regardless of the colour of the illumination. However, some researchers argue that chromatic adaptation and colour constancy are entirely different phenomena (Brill and West, 1986). Chromatic adaptation requires several seconds to occur whereas colour constancy tends to take place immediately (Land, 1986). Secondly, chromatic adaptation can occur for simple stimuli, whereas colour constancy works best for so-called complex images where there are many different surfaces in the scene (Hurlbert, 2002). Since it seems clear that chromatic adaptation cannot entirely account for the phenomenon of colour constancy, some computational models try to recover the spectral reflectance of the surfaces in a scene from the triplets of cone responses that they elicit (Maloney and Wandell, 1986). Other approaches propose that colour constancy may be achieved by using the information contained in highlights produced by specular reflection from glossy surfaces (Hurlbert, 2002). The highlights contain the same relative spectral power distribution as the light source and therefore may provide a clue to the colour of the light source. However, most successful approaches to explain colour constancy incorporate the fact that the colour appearance of a patch in a scene is relative to that of other patches in the scene (Land, 1986).
1.5.3 Colour appearance The phenomena of colour contrast and colour constancy are both aspects of colour appearance. For many industrial applications colour appearance is strictly not an issue. For example, if a manufacturer wishes to determine whether two batches of paint are the same colour, he could measure the CIE tristimulus values for a sample of each of the two paints. If the two samples have the same CIE XYZ values, the manufacturer would be able to state with confidence that, for the average observer8, the two samples would look the same colour as each other. However, this match would be conditional upon a given illumination, viewing distance, sample size, background, etc. A spectral match (where the spectral reflectance factors of the two paints were identical) would give greater confidence that the two batches of paint would look the same, but such a match would still be conditional on the paint samples being viewed against the same background as each other. Neither the spectral reflectance values nor the CIE tristimulus values, however, would give much useful information as to what colour the 8 Strictly, the manufacturer would only be able to claim a match for the CIE standard observer. Since the CIE standard observer was the result of visual assessments made by a very small number of people, it is unlikely that the CIE standard observer is equal to the average observer.
20
Total colour management in textiles
paint samples would actually appear to be for any given observer under any set of conditions. It has been shown that colour appearance is a complex phenomenon, and a deep understanding of the properties and function of the human visual system is required in order to appreciate it. Some advances in predicting colour appearance have been made. For example, the CIELAB system can be described as a colour-appearance model although it is not a particularly effective one. So, for example, a neutral achromatic object will have CIELAB coordinates (a* = b* = 0) that do not change with the illuminant. Furthermore, recent progress has been made in the form of sophisticated colourappearance models such as CIECAM97s (Li, Luo and Hunt, 2000) and CMCCAM2000 (Li et al., 2002). Unfortunately, whilst such models are useful for certain practical and well-defined situations, further research is required before such models even begin to be able to account for the majority of visual phenomena. Ultimately, our perception of colour may forever remain a private experience.
1.6
References
Bartleson, C.J. (1984). Mechanisms of vision. In Optical Radiation Measurements, Vol 5, Bartleson, C.J. and Grum, F. (eds.). Academic Press. Bowmaker, J. (2002). The retina. In Signals and Perception: The Fundamentals of Human Sensation, Roberts, D. (ed.). Palgrave MacMillan. Brill, M.H. and West, G. (1986). Chromatic adaptation and color constancy: a possible dichotomy. Color Research and Application, 11 (3), 196–204. Campbell, F.W. and Green, D.G. (1965). Optical and retinal factors affecting visual resolution. Journal of Physiology, 181, 576–583. Campbell, F.W. and Robson, J.G. (1968). Application of Fourier analysis to the visibility of gratings. Journal of Physiology, 197, 551–566. Feynman, R.P. (1990). Q.E.D.: The Strange Theory of Light and Matter. Penguin Books Ltd. Finlayson, G.D., Drew, M.S. and Funt, B.V. (1994). Color constancy – generalized diagonal transforms suffice. Journal of the Optical Society of America A, 11 (11), 3011–3019. Foster, D.H. (2003). Does colour constancy exist? Trends in Cognitive Sciences, 7 (10), 439–443. Hurlbert, A. (2002). Colour in context: contrast and constancy. In Signals and Perception: The Fundamentals of Human Sensation, Roberts, D. (ed.). Palgrave MacMillan. Hurvich, L.M. (1981). Color Vision. Sinauer Associates Incorporated. Judd, D.B., MacAdam, D.L. and Wyszecki, G.W. (1964). Spectral distribution of typical daylight as a function of correlated colour temperature. Journal of the Optical Society of America, 54, 1031–1040. Kaiser, P.K. and Boynton, R.M. (1996). Human Color Vision. Optical Society of America.
Colour perception
21
Land, E.H. (1986). Recent advances in retinex theory. Vision Research, 26 (1), 7–21. Li, C.J., Luo, M.R. and Hunt, R.W.G. (2000). Revision of the CIECAM97s model. Color Research and Application, 25 (4), 260–266. Li, C.J., Luo, M.R., Rigg, B. and Hunt, R.W.G. (2002). CMC 2000 chromatic adaptation transform: CMCCAT2000. Color Research and Application, 27 (1), 49–58. Maloney, L.T. (1986). Evaluation of linear models of surface spectral reflectance with small numbers of parameters. Journal of the Optical Society of America, 3, 1673–1683. Maloney, L.T. and Wandell, B.A. (1986). Color constancy: a method for recovering surface spectral reflectance. Journal of the Optical Society of America A, 3 (1), 29–33. Meek, K. (2002). The cornea. In Signals and Perception: The Fundamentals of Human Sensation, Roberts, D. (ed.). Palgrave MacMillan. Nassau, K. (1983). The Physics and Chemistry of Color: The Fifteen Causes of Color. John Wiley & Sons. Ripamonti, C. and Gerbino, W. (2001). Classical and inverted White’s effect. Perception, 30 (4), 467–488. Rushton, W.A.H. (1965). Visual adaptation. Proceedings of the Royal Society of London (B), 162, 20–46. Sinclair, R.S. (1997). Light, light sources and light interactions. In Colour Physics for Industry, McDonald, R. (ed.). Society of Dyers and Colourists. Stockman, A., MacLeod, D.I.A. and Johnson, N.E. (1993). Spectral sensitivities of the human cones. Journal of the Optical Society of America A, 10 (12), 2491–2521. Tilley, R. (2000). Colour and the Optical Properties of Materials. John Wiley & Sons. Wandell, B.A. (1995). Foundations of Vision. Sinauer Associates Incorporated. Westland, S. (2002). Colour science. In Signals and Perception: The Fundamentals of Human Sensation, Roberts, D. (ed.). Palgrave MacMillan. Westland, S. and Ripamonti, C. (2004). Computational Colour Science Using MATLAB. John Wiley and Sons. Zollinger, H. (1999). Color: A Multidisciplinary Approach. John Wiley & Sons.
2 Colour description/specification systems B R I G G, University of Leeds, UK
2.1
Introduction
Colour has always been important, but especially so in the modern world. We only have to look around us to see the variety of colours produced on textiles, painted surfaces, paper and plastics. In most cases the colour is an important factor in the production of the material and is often vital to the commercial success of a product. Textiles often sell as much on colour and design as on quality of materials. In our homes, colours of carpets, curtains and paints are carefully chosen to produce a pleasing effect. The same colours may be represented as colour prints, for example in mail order catalogues, or on computer screens for design work or in e-commerce. In some cases, such as foods, we use the colour to judge the quality of the material, while in other cases, such as packaging, the colour is important in attracting customers. In almost all applications it is difficult to produce exactly the colour required and to ensure that successive batches are the same colour as each other. Raw textile materials such as cotton and wool vary from batch to batch as do dyes and pigments, despite strenuous efforts by producers to try to ensure good quality control. There are normally two main considerations for the producer of coloured products (or computer images). Firstly, whether the colour is reasonably close to that desired. Secondly, whether repeat samples or batches match each other. For example, successive batches of paint, or arrays of clothing items displayed together should match very closely. In this case, it is the differences in colour between samples or within a sample that matter. Two alternative methods are available to assess the correctness of the colours. For many years, only visual assessment was available. This worked fairly well, but there were problems. It is difficult to remember colours accurately and difficult to describe differences in colour. Experienced colourists have a good idea as to the size of difference which might be tolerated, but there is no objective way to settle differences of opinion between say a dyer who considers that a match is close enough, and a customer of the opposite opinion. 22
Colour description/specification systems
23
2.1.1 The nature of measurement In all branches of science and engineering, measurement plays an important part. In commerce materials are usually bought and sold by weight or volume. Without standardised systems for measuring mass, length and time, modern life would be very difficult. It is obvious that a standard system for measuring and specifying colour is equally desirable. However, there are important differences between colour and, for example, length. The colour of an object depends on many factors, such as lighting, size of sample, and background and surrounding colours. Much more importantly, colour is a subjective phenomenon and depends on the observer. The measurement of subjective phenomena, such as colour, taste and smell, is obviously more difficult than that of objective phenomena such as mass, length and time. Provided that good instruments are used carefully, we can be confident that, for objective measurements, the results will be correct. With colour the situation is completely different. If a colour does not look to be correct, it is not correct, no matter what any instrument indicates. The final arbiter is the human eye. One problem with regard to colour measurement is that the human eye is readily available and particularly sensitive to colour; millions of different colours can be distinguished. Any measurement that is less reliable than the unaided eye will be of limited value. Another consideration is that, with colour, we are often concerned with differences in colour rather than with colour itself. For many purposes the exact colour is not as important as uniformity of colour. When buying a blue shirt, the exact shade of blue is unlikely to matter, but any appreciable difference between the collar, sleeves and other parts would not be acceptable. Small differences between different shirts on display at the same time would give an impression of carelessness on the part of the manufacturer. Similarly, when buying paint, a customer will often accept any colour within reasonable range of the desired colour, but would be much less tolerant of even a small difference between the paint from two different cans. With colour, we should never forget that the final objective is to produce something that is pleasing or satisfactory to the eye. If the colour looks wrong, it is wrong. When assessing the usefulness of the colour specification system to be described in this chapter, the reader should carefully consider how far the system enables us to deal with real problems with respect to colour and how far the system fails to deal with the subjective nature of colour. For the present we will simply examine how a system of colour specification can be set up. We will return to the question of how far such a system fulfils our needs at the end of the chapter. In considering the appearance of an object, factors such as texture and gloss are important, as well as colour. However, in this chapter these factors will be ignored.
24
Total colour management in textiles
Almost all modern colour measurement is based on the CIE system of colour specification. The initials come from the French title of the international committee (Commission Internationale de l¢Eclairage) who set up the system in 1931. Although additions have been made since, the basic structure and principles are unchanged and the system is widely used. The reader should bear this in mind when determining how useful such a system is likely to be, while we consider the development of the CIE system in the next few sections. It should also be emphasised at this point that the system is empirical, i.e. it is based on experimental observations rather than on theories of colour vision. When discussing colour in general, we could be considering coloured lights, coloured solutions or coloured surfaces such as painted surfaces, plastics and textiles. In most practical situations we are concerned with coloured surfaces, although, as we shall see, the properties of coloured lights are used in the specification of the colour of surfaces. It is important to realise that the colour of an object depends on the light source used to illuminate its surface, the particular observer who views it, as well as the properties of the surface itself. The nature of the surface is the most important factor. A piece of white paper will look white under all normal light sources when viewed by any observer with normal colour vision. (This statement is not completely true, for example small areas of ‘white’ paper viewed as part of a pattern formed by a variety of colours could look to be a different colour. In trying to understand the principles involved in colour measurement, it is probably best to ignore such exceptions for the moment and adopt a simplified view of colour. In practical applications, however, such factors must be considered whenever appropriate.) Because our white paper does remain white, there is a tendency to think of colour as a property of the surface. In considering the measurement of colour, the light source and the observer cannot be ignored; we should think about surface, light source and observer.
2.2
Basic facts
It would be very difficult to design a system of colour measurement that attempted to describe the colours that we see. We simply have to think how we might describe a colour. Colour names such as red, yellow, green and blue are reasonably well understood, but names such rose, salmon and cerise are less well standardised. What might be called rose by one person could be called pink by another. Even for red there is a considerable range of colours which any one person would accept as red. This is not necessarily the same range as for a second person. Strictly speaking, this discussion concerns the names which we give to colours rather than what people actually see. The latter is much more difficult to pin down, but there is no doubt
Colour description/specification systems
25
that different people do see colours differently. Even for one person, the appearance of a particular coloured surface can change as the circumstances (e.g. surrounding colour) change. The CIE system basically attempts to tell us how a colour might be reproduced (by a mixture of three primary light sources) rather than described. The amounts of the three primaries required to match a particular colour provide a numerical specification of that colour. A different colour would require different amounts of the primaries and hence the specification would be different. It turns out that, in many applications, this is all that is required. As we will see later, some idea of the colour seen can be deduced from CIE colour specifications; furthermore, we would never attempt to reproduce a colour by actually mixing the CIE primaries. It is well known that colour is three-dimensional. This is apparent in various ways. Colour atlases such as the Munsell atlas arrange colours using three scales (hue, value, and chroma in the case of the Munsell system). We can match a wide range of colours using a mixture of three dyes, and the concentrations of three specific dyes required on a particular substrate to match a colour could be used to specify that colour. We would need to specify the light source under which the colour was seen, but the three concentrations would give a numerical specification of the colour; the specification would be different for different colours and it would be possible for someone else to reproduce the colour. Unfortunately, dyes or pigments tend to be impure and the precise colour depends on the method of manufacture. Even repeat batches from the same plant will not match exactly, and properties of mixtures are not completely predictable. Coloured lights are much easier to define and reproduce. Imagine a red light obtained by isolating the wavelength 700 nm from the spectrum. All laboratories in the world capable of measuring wavelength accurately (an objective physical measurement) could produce the same red colour. A green colour corresponding to 546.1 nm could be produced even more easily. A mercury lamp emits light at only four wavelengths in the visible region (404.7 nm, 435.8 nm, 546.1 nm and 577.8 nm). By filtering out the other three, the required green wavelength could be obtained. The wavelengths 404.7 nm and 435.8 nm could be obtained in a similar manner. Small variations in the operating conditions have no significant effect on the wavelengths emitted by a mercury lamp. Hence, three primary light sources could be defined simply as appropriate wavelengths and be easily reproduced. Mixtures of three coloured lights can be produced in various ways, but the simplest is to imagine three spotlights shining on the same area on a white screen. The colour produced would be a mixture of the three colours being mixed and it is possible to produce a wide range of colours by varying the amounts of the three primaries. The colours to be matched could include surface colours illuminated by a particular light source.
26
Total colour management in textiles
2.3
Additive and subtractive mixing
Most of us are familiar with the colours produced by mixing paints or coloured solutions. The results depend on the exact colours mixed, but usually red + blue = purple red + yellow = orange yellow + blue = green while red + yellow + blue in the correct proportions = grey or black. To people accustomed to mixing dyes or pigments, the colours produced by mixing lights might be surprising in some cases. For example, a blue light mixed with a yellow light might well give white, while red and green lights could be mixed to give a yellow. Quite obviously there is something fundamentally different about mixing dyes (and pigments) and mixing coloured lights. The latter is an example of additive mixing, while dye or paint mixture is an example of a subtractive mixing. Since the CIE system is based on additive mixtures of lights, these must be considered further. Additive mixtures occur when two or more coloured lights are shone at the same time so that we see the two lights together. Consider red and green lights shone onto a white screen. The screen will reflect almost all the light incident upon it, and the mixture of red plus green in the same appropriate proportions will reach our eye. If the colour seen (yellow) is surprising, this is simply because we are not used to mixing colours in this way. Note that there is no way in which the two colours interact with each other. If the red and green are single wavelengths, both wavelengths reach our eye and are not interfered with in any way by each other. We see the red wavelength plus the green wavelength and hence call the mixture an additive mixture. Subtractive mixtures occur much more commonly but usually involve much more complicated processes. The simplest case occurs when we shine light through two coloured glass filters placed one behind the other. The fraction of light transmitted by the combination of filters is simply the product of the fractions transmitted by the filters independently. The most important examples of subtractive mixtures occur when we mix paints, or when we dye with a mixture of dyes. The results are often predictable on the basis of everyday experience, but the details of the process are very complicated (see Chapter 8). When we see the result of a subtractive mixture, we generally see light reflected at many different wavelengths. For example, a mixture of yellow and blue dyes will absorb some light at all
Colour description/specification systems
27
wavelengths, but mainly the blue and red wavelengths are absorbed and we see the remaining light, probably giving a green appearance. A yellow surface absorbs the short (blue and violet) wavelengths strongly and reflects most of the green, yellow and red wavelengths. The yellow section of the visible spectrum is very short, and what we see is basically an additive mixture of green and red, resulting in the yellow appearance.
2.3.1 Properties of additive mixtures of light As indicated earlier, coloured lights are easy to define and hence are suitable for use as primaries in a system of colour specification. The properties of additive mixtures of coloured light have been studied for many years1–5 and those that are particularly relevant to their use in systems of colour specification are considered here. We can match a wide range of colours using an additive mixture of, say, red, green and blue primaries. Suppose our primaries are single wavelengths and we use them to match white light consisting of a mixture of all the wavelengths in the visible region. Although the mixture is physically quite different from the white light, by careful adjustment of the amounts of the primaries, the white light can be matched, i.e. we can produce, with the mixture, a white that looks identical to the white light. If we change the colour or lightness of the surround, our white colours change in appearance. However, over a wide range of conditions, the match holds, i.e. if the colours change, they both do so by the same amount. Hence we can deal with colours without considering their spectral composition, at least in many applications. Furthermore, if colour 1 matches colour 2, and colour 3 matches colour 4, then an additive mixture of colour 1 plus colour 3 will match colour 2 plus colour 4. This is vital when we consider that normal colours are additive mixtures of all the wavelengths in the visible spectrum. We need to consider the effect of the additive mixture of all the wavelengths. It cannot be stressed too strongly that modern colorimetry is based on the properties of additive mixtures of coloured lights, and that these properties have been determined by experiment. The main properties were established well over a century ago. Subsequent work has confirmed that the simple properties described above are indeed true, but has defined much more closely the range of experimental conditions under which the simple laws hold.6–8 Theories of colour vision must attempt to explain such laws, and the exceptions.
2.4
The CIE system of colour specification
If we were to select and define three particular primaries (R), (B) and (G), the amounts of these required to match any colour could be used to specify
28
Total colour management in textiles
the colour and be called tristimulus values. Each different colour would require different amounts, the tristimulus values would therefore define the colour and with practice we could deduce the appearance of the colour from the tristimulus values. However, different results would be obtained by anyone using a different set of primaries. It can be shown that sets of tristimulus values obtained using one set of primaries can be converted to the corresponding results that would have been obtained using a second set, provided that the amounts of one set of primaries required to match each primary of the second set of primaries in turn are known. Hence we could either insist that one set of primaries is always used, or allow the use of different sets, but insist that the results are converted to those which would have been obtained using a standard set. Even using very pure colours for our set of primaries, there would still be some very pure colours that could not be matched. For example, a very pure cyan (blue–green) might be more saturated than the colours obtained by mixing the blue and green primaries. Adding the third primary (red) would produce even less saturated mixtures. A possible solution in this case would be to add some of the red primary to the pure cyan colour, and then match the resultant colour using the blue and green primaries. Experimentally, it has been found that, if this procedure is followed, all colours can be matched using one set of primaries, the only restriction in the choice of primaries being that it must not be possible to match any one of the primaries using a mixture of the other two. Negative values are undesirable. It would be easy to omit the minus sign or fail to notice it. By careful choice of primaries it is possible to reduce the incidence of negative tristimulus values. The best primaries are red, green and blue spectrum colours. Mixtures of these give the widest possible range of colours. However, there is no set of real primary colours that can be used to match all colours using positive amounts of the primaries, i.e. there is no set of real primaries that will eliminate negative tristimulus values entirely. Since it is possible to calculate tristimulus values for one set of primaries from those obtained using a second set, there is no need to restrict ourselves to a set of real primary colours. Purely imaginary primaries can be used; it is only necessary that these have been defined in terms of the three real primaries being used to actually produce a match. This is not just a hypothetical possibility. Negative tristimulus values would be a nuisance in practice and, in the CIE system, imaginary primaries are indeed used so as to avoid negative values. If we produced a visual match, the amounts of the primaries required could be noted, and the results converted to the equivalent values for a standard set of primaries. Such a procedure is perfectly possible, the main problem lying in the precision and accuracy achievable. The results would vary from one observer to another because of differences between eyes.
Colour description/specification systems
29
Even for one observer, repeat measurements would not be very satisfactory. Under the controlled conditions (usually one eye, small field of view and low level of illumination) necessary in such an instrumental arrangement, it is impossible to achieve the precision of unaided eyes under normal conditions, e.g. when judging whether an appreciable difference exists between two adjacent panels on a car body under good daylight. Some of the problems could be overcome by using more than three primaries9, but such an arrangement has found little use in industrial applications. The matches in the instrumental arrangement being considered are physically very different, and visual judgements are particularly unreliable for this type of match. The CIE realised even before 1931 that this was likely to be the case and decided that the system adopted should allow calculation of tristimulus values from measured reflectance values. The system was set up in 1931 because the required information (basically infomation defining a standard observer) became available at that time. It must be stressed that the CIE system incorporates the features described earlier, i.e. a colour is to be specified by the amounts of the (X), (Y), and (Z) primaries required, in an additive mixture, to match it. The ‘standard observer’ data is added to the framework already described. It is possible to calculate tristimulus values (i.e. the amounts of three primaries which, if additively mixed, would match a colour) of a sample specified. The CIE had to define standard primaries, standard light sources and a standard observer, together with standard observing and viewing conditions.
2.4.1 Standard primaries As discussed earlier, the CIE had considerable latitude in selecting three primaries. For the moment, the main property to note is that all real colours can be matched using positive amounts of the chosen primaries (X), (Y) and (Z). These were defined by the following equations: Cl1 ∫ 0.73467(X) + 0.26533(Y) + 0.00000(Z)
[2.1]
Cl2 ∫ 0.27376(X) + 0.71741(Y) + 0.00883(Z)
[2.2]
Cl3 ∫ 0.16658(X) + 0.00886(Y) + 0.82456(Z)
[2.3]
SE ∫ 0.33333(X) + 0.33333(Y) + 0.33333(Z)
[2.4]
where l1 = 700 nm, l2 = 546.1 nm, l3 = 435.8 nm and SE = equal-energy stimulus, i.e. a stimulus having equal amounts of energy at all wavelengths through the visible spectrum. The relative sensitivity of the eye to light of different wavelength had been determined previously, and a particular curve (denoted by Vl) adopted
30
Total colour management in textiles
as standard by the CIE in 1924. The CIE 1931 standard colorimetric system was made consistent with the 1924 Vl curve by choosing primaries such that the y¯l curve (see Section 2.4.3) was identical to the Vl curve.
2.4.2 Standard light sources and standard illuminants The percentage (or fraction) of light at different wavelengths reflected from a coloured surface is independent of the amount of light incident on the surface, but the amount of light reflected is the amount incident multiplied by the fraction reflected. The amount of light reflected, and hence the appearance, depends on the light source. In practice, we use many different light sources, particularly various phases of daylight, and various types of fluorescent tube and tungsten light. However, if we wish to check that a paint sample is the correct colour, it would probably be satisfactory to check using one form of daylight, tungsten light and a fluorescent tube. Any sample that was satisfactory under all three conditions would almost certainly be satisfactory if seen under any other light source. Hence it was only necessary for the CIE to specify a small number of light sources rather than all possible sources. For standardisation purposes the minimum number is desirable. In 1931 fluorescent tubes were unimportant and the CIE specified three standard illuminants as follows. (The CIE distinguishes between sources and illuminants. A source refers to a physical emitter of light such as the sun or a lamp, while an illuminant refers to a specified spectral energy distribution. Thus an illuminant can readily be specified, but may not be realisable in practice. In calculating tristimulus values from reflectance values, the tabulated energy distribution is used, but may be different from the actual distribution of the light source in the spectrophotometer.) CIE standard illuminant A The spectral energy distribution of Illuminant A (plot of E vs. wavelength) represents a black body radiator at an absolute temperature of 2856 K. Source A can be realised by a gas-filled coiled tungsten filament lamp operating at a correlated colour temperature of 2856 K. The energy distributions of source A and illuminant A can be very close if a calibrated lamp is used. CIE standard illuminants B and C Illuminants B and C correspond to different phases of daylight; the former is intended to represent direct sunlight with a correlated colour temperature of 4874 K and the latter to represent average daylight with a correlated
Colour description/specification systems
31
colour temperature of 6774 K. The CIE gave details of how sources B and C may be obtained in the laboratory.12 However, neither corresponds very closely to real daylight, particularly in the near UV region. The differences between the illuminants, e.g. A and C should be compared with the differences between the expected reflectance curves for different colours. Since the amount of light of any one wavelength reaching the eye is proportional to the energy of the source multiplied by the reflectance factor ElRl, it is clear that the amounts of any given wavelength of light reaching the eye from a single surface illuminated by two different sources can be quite different. Thus, considering all wavelengths of the visible spectrum, the tri stimulus values for a surface under two different illuminants may well be very different, even though, after allowing time for adaptation, the colours seen under the two sources may look very similar.
2.4.3 Standard observer-colour matching functions As explained above, we can calculate, for any light source, how much light is reflected at each wavelength throughout the visible region. The tristimulus values of any one wavelength are the amounts of the three chosen primaries required to match the light of the particular wavelength. If we know these tristimulus values we can calculate the tristimulus values for the sample. The amounts required depend on the observer and results for an average (or ‘standard observer’) are required. The values were actually determined experimentally as follows. Wright10 and Guild11 used visual tristimulus colourimeters in which onehalf of the field of view consisted of a mixture of (R), (G) and (B) primaries, while the colour in the other half was light of a single wavelength. To produce a match experimentally, it was necessary to add some of (R), (G) or (B) to the wavelength to be matched. This was quite possible experimentally, as discussed earlier, and the resultant tristimulus values for the wavelength included at least one negative value. Each used a somewhat different technique, and in particular different primaries were used. Both considered each wavelength throughout the visible spectrum and averaged results from a number of observers. The results differed from one observer to another (as expected), but when the average results from the two experiments were converted to a common set of primaries, the agreement was considered to be satisfactory. The results were expressed as the tristimulus values for an equal energy spectrum, i.e. using primaries (R), (G) and (B) the results were expressed as the amounts r¯ , g¯ and b¯ required to match one unit of energy of each wavelength throughout the visible region. Since (R), (G) and (B) were real primaries, some of the values were negative. The CIE adopted three unreal primaries (X), (Y) and (Z) and the colour matching functions in terms of these primaries are denoted by x¯, y¯ and z¯
32
Total colour management in textiles
and are always positive. This ensures that tristimulus values for all real colours are always positive. The values are called the CIE colour-matching functions and they define the colour-matching properties of the CIE 1931 Standard Colorimetric Observer. Note that all values are positive. This results from the choice of primaries, as does the fact that z¯ is zero for almost half of the spectrum.
2.4.4 Standard illumination and viewing conditions The CIE specified that opaque samples should be illuminated at 45° from the normal to the specimen surface and viewed at an angle close to the normal; alternatively, the specimen should be illuminated at an angle close to the normal and viewed at an angle 45° to the normal.
2.4.5 Calculation of tristimulus values The values of x¯l, y¯l and z¯l are the amounts of the CIE primaries (X), (Y) and (Z), respectively, needed to match one unit of energy of the wavelength l. For any sample, the amount of energy reflected from the surface, i.e. El multiplied by Rl. Hence the amounts of each of the CIE primaries needed to match this is ElRl multiplied by x¯l, y¯l, or z¯l, respectively. Adding the products for all the wavelengths gives us the amounts required to match the colour. These amounts are called tristimulus values and denoted by X, Y, and Z. Thus, for example, X is calculated from X = ÂEl x¯l Rl where the greek letter sigma means ‘sum’ as in many other mathematical equations. So far, the question of units has been avoided. Using the CIE system as described we could calculate the tristimulus values for a sample. However, the units used for El and x¯l, y¯l, and z¯l are arbitrary. The values of El for one wavelength relative to another are correct, but the absolute values have not been specified. Similarly, the x¯l values for one wavelength are correct relative to other x¯l values at other wavelengths, and correct relative to the y¯ and z¯l values at the same wavelength; however, the absolute size of the values is arbitrary. For opaque samples (object colours), the usual practice is to normalise the tristimulus values by calculating using (2.5): X = ∑ Eλ xλ Rλ
Y = ∑ Eλ yλ Rλ
Z = ∑ Eλ zλ Rλ
∑E y ∑E y ∑E y λ
λ
λ
λ
λ
λ
[2.5]
Thus if Rl is expressed as a percentage, Y runs from zero (for a sample which reflects no light) to 100 (for a sample which diffusely reflects all the
Colour description/specification systems
33
Table 2.1 Tristimulus values for a sample reflecting 100% of the incident light based on the CIE 1964 supplementary standard observer and weighting factors at 10 nm intervals Illuminant
X
Y
Z
A C D65
111.14 97.29 94.81
100.00 100.00 100.00
35.20 116.15 107.30
light incident on it) and is independent of any units used. The ranges for X and Z depend on the illuminant. For a sample reflecting all the light incident upon it, i.e. Rl = 100% at all wavelengths, the X, Y and Z values for the sample under illuminants A, B, C and D65 are given in Table 2.1. It can be seen that the Z tristimulus values, in particular, vary greatly with the illuminant. The fact that the tristimulus values of a sample take no account of the intensity of light incident on the sample causes no problems in normal practice. If, for example, we produce a sample with the same tristimulus values as a target (for a particular illuminant), the two will match for all normal levels of illumination by the chosen illuminant. The actual appearance of object colours is almost independent of the level of illumination. A piece of white paper looks white in weak sunlight. If the intensity of the sunlight increases, we recognise this, but the paper does not look any lighter. A medium grey is recognised as such whether viewed in daylight on a very dark day or on a very bright day. This is true even though the amount of light reflected by the grey sample on a bright day would probably be much more than that reflected by the white paper on a dark day. To a very large extent, the appearance of object colours seems to be judged relative to the light source. We separate the properties of the sample from those of the source. This applies to changes in the distribution of light throughout the spectrum as well as to changes in the intensity of the source. Our white paper will still look white under tungsten light even though the distribution reaching the eye is vastly different from that reaching the eye when the paper is seen in daylight. In summary, the CIE defined a standard observer, standard illuminants, standard illuminating and viewing geometries and a particular set of primaries (X), (Y) and (Z). The tristimulus values of a particular colour are the amounts of (X), (Y) and (Z) required to match the colour under standard conditions and are referred to as X, Y and Z. In practice, X, Y and Z for coloured materials
34
Total colour management in textiles
are calculated from measured reflectance values together with x¯, y¯, and z¯ and El for the chosen standard observers and illuminants. For the standard observer, an additive mixture of X units of the (X) primary together with Y units of the (Y) primary and Z units of the (Z) primary would look the same as the sample illuminated by the appropriate standard illuminant.
2.4.6 Additions to the CIE system since 1931 Since the CIE system of colour specification was adopted in 1931, the basic system has remained unchanged, but experience has led to additions being made. D illuminants Illuminants B and C were intended to represent different phases of daylight. Later measurements13 have shown that neither represents any common phase of daylight at all closely, particularly in the near UV region. This latter point is particularly important when considering samples treated with fluorescent brightening agents. Illuminant D65 is based on measurements of the total daylight (i.e. sun plus sky) in a number of countries. Except for times near sunrise and sunset, the relative spectral energy distribution generally corresponds to correlated colour temperatures between 6000 and 7000 K. If we consider illumination by only part of the sky (e.g. a portion of blue sky from a north-facing window or direct evening sunlight from a west-facing window), the correlated colour temperature and energy distribution can be quite different. Judd et al.13 showed that the different relative energy distributions could be represented quite closely by a series of curves dependent only on the correlated colour temperature of the particular form of daylight. Based on Judd’s work, the CIE has defined a series of D illuminants ranging from correlated colour temperatures of 4000 to 25 000 K. In the interests of standardisation, the CIE recommends that D65 should be used whenever possible. There seems to be no doubt that the D illuminants represent a substantial improvement over illuminants B and C. Illuminant B has rarely been used and in recent years illuminant C has been almost completely replaced by the D illuminants, D65 (for textiles) and D50 (for graphic arts), in particular. However, there is a major problem with the D illuminants in that there is no satisfactory way of obtaining, say, D65 in the laboratory. Problems occur with metameric pairs. A pair that are a close match according to tristimulus values calculated for D65 might well be seen to be a poor match when viewed under a so-called daylight source in a viewing cabinet. However, the match might well look better if inspected under real daylight.
Colour description/specification systems
35
1964 supplementary standard observer (10°) The original 1931 CIE standard observer was based on experiments using a 2° field of view. This is a much narrower field of view than that normally used for critical colour appraisal. In addition, a few problems were encountered using the 1931 observer.14 New colour matching experiments were therefore carried out by Stiles and Burch15 and by Speranskaya.16 The experiments were similar to those by Wright and Guild for the 1931 standard observer in that each wavelength in turn was visually matched using an additive mixture of the primary light sources. The main difference was the much wider field of view (10°). As indicated earlier, the use of wide fields with metameric matches causes problems. These were largely overcome by ignoring the centre 2° of the field of view. The two sets of experimental results were combined and used to define the CIE 1964 Supplementary Standard Colorimetric Observer, often referred to as the 10° Observer. The subscript 10l is used to distinguish the 10° data from the original 1931 2° standard observer data. It is recommended that the 1964 observer is used whenever a more accurate correlation with visual colour matching of fields greater than 4° is required. However, there is probably little to choose between the two standard observers for some applications. Standard illuminating and viewing conditions The original CIE recommendation was that the sample should be illuminated at 45° to the surface and the light viewed normally, i.e. at right angles to the surface. This mode can be represented 45/0. It was assumed that the opposite mode (0/45) would give the same result, but this is not the case if the incident light is polarized17. Four possible sets of conditions are now recommended. These are 45/0, 0/45, d/0 and 0/d. In the third case the sample is illuminated by diffuse light while in the last case the light reflected at all angles is collected (using an integrating sphere, as in many spectrophotometers).
2.4.7 Standard of reflectance factor The CIE recommends that reflectance measurements should be made relative to a perfect diffuser, i.e. a sample that diffusely reflects all the light incident upon it. No such surface exists, but working standards of known spectral reflectance factors are normally used, allowing the correct results to be obtained. (For example, if the working standard reflects 98% of the light of a particular wavelength, all values measured relative to the working standard need to be multiplied by 100/98.) In practice, instrument manufacturers supply calibrated white tiles with their instruments. Using these, corrected R values are obtained automatically.
36
Total colour management in textiles
2.5
Calculation of tristimulus values from Rl values measured at 10 or 20 nm intervals
The equations given above for calculating tristimulus values from reflectance values may look awesome to a beginner, but involve only simple additions and multiplications, and in any case these days are invariably carried out using computers generally attached to a spectrophotometer. However, there are problems. The CIE standards20 tabulate El, x¯l, etc. at 5 nm intervals over the full visible region. Many spectrophotometers measure Rl at 10 or 20 nm intervals, often over a restricted wavelength range, e.g. 400 nm, 420 nm, . . . , 700 nm. At one time it was considered sufficient simply to use the approptiate CIE values at the measured wavelengths, and to add the values at, e.g. 720 nm, 740 nm etc. to the values at 700 nm. It was gradually realised18,19 that large and completely unnecessary errors could be introduced in this way. Substantial work has been carried out in attempts to minimise these errors. Stearns has described a method that allows appropriate adjustments to the El, x¯l, etc. tables to be calculated for any wavelength interval. The results from using these adjusted tables are consistent with those obtained using 5 nm interval tables and interpolated R values where necessary, and give considerably improved results with no extra effort on the part of the user. However, even very small differences can cause problems, for example when checking computer programs, or when comparing results for the same sample measured by a supplier and his customer, and it is highly desirable that everyone uses the same set of values. In an attempt to promote uniformity of practice, the American Society for Testing and Materials (ASTM) has produced 10 nm and 20 nm tables for commonly used illuminants and for both standard observers. ASTM also give a method for calculating appropriate values for other illuminants. Some problems still remain. For most users the values to be used are built in to the software supplied by the instrument supplier. However, manufacturers are understandably reluctant to change software. Improved sets of values or improved methods would lead to different results even for identical sets of reflectance values. Users can check whether appropriate sets are being used by inputting R values of 100 at all wavelengths and checking the results.
2.6
Relationships between tristimulus values and colour appearance
It is difficult to relate, in a simple way, the tristimulus values of a sample to the colour appearance. One reason is that the colour depends not only on the stimulus, but also on surrounding colours and the state of adaptation of the eye. Even ignoring such factors, the three-dimensional nature of colour
Colour description/specification systems
37
makes it difficult to determine relationships and it is usual to simplify any relationship involving colour by considering only one or two dimensions at a time. It was indicated earlier that, in choosing primaries for the CIE system, the wide range of possibilities had been used to produce certain desirable features in the final system. A consequence of ensuring that the y¯l curve corresponded to the Vl curve was that the Y tristimulus value should roughly represent the lightness of a sample, i.e. the higher the Y value, the lighter the sample appears. The scale is far from uniform and caution should be exercised when comparing the lightness of quite different colours such as reds and greys. Nonetheless, in general it will be found that, the higher the Y value, the lighter a sample will look. Thus if Y = 80, we can be sure that the sample will appear light, while if Y = 3, the sample will look dark.
2.6.1 Chromaticity diagrams To represent the other two dimensions of colour, it was usual to first define ‘chromaticity coordinates’ (x, y and z) and then plot y against x: x = X(X + Y + Z) y = Y(X + Y + Z) and z = Z(X + Y + Z)
[2.6]
From (2.6) it follows that x + y + z = 1 for all colours; it is therefore only necessary to quote two of the chromaticity coordinates and these can be plotted on a normal two-dimensional graph. It can also be shown that X and Z can easily be calculated from x, y and Y; hence the latter set is an acceptable form of specification, and consideration of Y values and plots of y against x should cover all possible colours. A plot of y against x is called a chromaticity diagram and, at one time, was widely used. It is described in detail in many textbooks on colour. It was stated earlier that the Y scale is far from uniform. The same applies to the x y diagram; equal distances in the diagram do not correspond to equal visual differences. For a fixed difference in x and y, the difference seen would be much smaller for a pair of green samples than for pairs of blue or grey samples. This problem will be considered further in Chapter 4. A further disadvantage of the x y diagram is that colours measured under different standard illuminants, e.g. Illuminant D65 and Illuminant A are represented by completely different points in the diagram. It has been
38
Total colour management in textiles
emphasised that colour is three-dimensional. Thus any two-dimensional plot cannot represent colour completely. In the case of the chromaticity diagram, it is simplest to regard the missing factor as the Y tristimulus value. In general, any one point on any chromaticity diagram corresponds to a range of colours differing in lightness, and this should always be kept in mind when trying to visualise the colours corresponding to particular chromaticity coordinates.
2.7
Usefulness and limitations of the CIE system
In many ways the CIE system of colour specification has been remarkably successful. Almost all important applications of colour measurement use the CIE system. The basic system has survived unchanged for over 70 years. The additions made since 1931 have led to improvements in some respects, but have not changed the basic principles of the system in any way. It is unlikely that any major changes will be made in the foreseeable future. On the other hand, we must consider the limitations of the system. These stem basically from the limited objectives of the system rather than from a failure to meet the objectives. The CIE tristimulus values for a sample are related to the colour of the sample, but ignore other important features such as surface texture, gloss, sheen, etc. Thus a gloss paint sample and a matt paint sample might have the same tristimulus values, but obviously will not look the same. Whether the colours of the two samples will look the same depends critically on the geometrical arrangements for illuminating and viewing the samples. Only if the instrumental geometry of illumination and viewing conditions are close to that used visually will the colours seen be close. An instrument will always average out the light reflected from the area being measured (typically a 2 cm diameter circle). In judging a colour visually some sort of averaging takes place, but the observer is always conscious of any nonuniformity over the area viewed. Thus a matt paint surface, a woven textile surface and a pile fabric will always look different from one another, but their measured tristimulus values could be the same. Ignoring all features other than colour, the tristimulus values for a sample give only a limited amount of information. Basically, the tristimulus values tell us the amounts of three imaginary primaries, which if additively mixed will give the same colour as a surface illuminated by a standard source and viewed by a standard observer using one of the standard geometries. It follows that the mixture of the CIE primaries would be unlikely to match the surface if the latter was illuminated by a different source or if the ‘match’ was viewed by an individual observer or if a different illuminating or viewing geometry was used. Control can be exercised (to some extent) over the source and the geometry. If these are important, we must try to
Colour description/specification systems
39
ensure that the instrumental conditions correspond as closely as possible to those to be used when viewing the object visually. The only choice as far as the standard observer is concerned is whether to use the 1931 (2°) observer or the 1964 (10°) observer. Neither is likely to correspond closely to any individual observer. However, the standard observers may well correspond reasonably closely with the average judgement of real observers, bearing in mind that, in many applications, the product is mass produced and will be seen by many different observers. A full specification of a colour requires X, Y and Z values (or equivalent sets such as x, y and Y or L*, a*, and b*) for several different illuminants. The results are still only valid for the standard observer and could be unsatisfactory for a real observer. This should not be a problem in practice since we usually require colours to be acceptable to a large number of potential customers; the standard observer is probably a better guide to the population in general than any one observer. However, situations often do arise where one particular individual (e.g. a head dyer or a buyer for a chain store) inspects material and problems will occur when that individual’s colour vision is appreciably different from that of the standard observer. Problems will be much more severe for highly metameric pairs of samples. Strictly speaking, the tristimulus values tell us nothing about the colour of a sample although, as discussed above, with experience we can make a reasonable estimate of the colour from either X, Y and Z or x, y and Y values. In such cases it is essential that the illuminant used for the measurements is known. Chromaticity coordinates of x = 0.314 and y = 0.331 correspond to a neutral colour if derived from measurements under illuminant D65, but if derived from measurements under illuminant A will correspond to a blue colour. In many applications the aim is to match a particular target, which might be defined by a set of tristimulus values. If we produce a sample and wish to test whether this matches the target, it is essential that sample and target measurements correspond to exactly the same conditions (illuminant, standard observer, illuminating and viewing geometries, and, in practice, the same instrument). If, for example, the sample is a really good match to the target, but the tristimulus values are to be measured using different standard observers for the sample and target, the resulting tristimulus values and chromaticity coordinates would be appreciably different. Again, the sample and target might have different surface structures, such as those of matt paints, gloss paints or pile carpets. In these cases the tristimulus values could be identical but the surfaces would look appreciably different. Whenever possible, sample and target should have the same surface structure. It is usually important that the colours within one batch and between repeat batches of material match closely. In these cases the samples will be
40
Total colour management in textiles
of the same material, and the same dyes or pigments will have been used. It would be natural to measure all the samples using the same instrument and (if a spectrophotometer was used) to calculate the tristimulus values for the same standard observer and standard illuminant. Under these conditions, if the tristimulus values for a sample are very close to those for the standard, the sample will be a close visual match to the standard for any normal observer viewing under a light source roughly equivalent to the standard illuminant used to calculate the tristimulus values. (Exceptions can occur, for example, with metallic paints, where the appearance depends very much on the illuminating and viewing geometries.) If the variation of appearance with, say, viewing angle is different for the sample and for the standard, they may not match visually, although according to the instrumental results (obtained with a different viewing angle) they should be a good match. In many ways the chief limitation of the CIE system is its non-uniformity. Equal changes in x, y or Y do not correspond to the same perceived difference. Many attempts have been made to provide a more uniform system. In each case the basic approach has been to start with the tristimulus values or chromaticity coordinates from the CIE system, and to transform these in some way to give a more uniform system. The end result is a colourdifference formula which, for a pair of samples, gives a number that is intended to be proportional to the difference seen (see Chapter 4).
2.8
Colour order systems
It is often useful to have real physical samples to represent colours. A designer can use such a sample to see the effect of the colour alongside other colours and, when satisfied, to pass the sample to a dyer to represent the colour required. The sample may be a piece of coloured paper or dyed fabric, etc. The main advantage of this is that designer and dyer can both see the colour. One limitation is that such samples are not stable – the colour changes with time, particularly if the sample is handled often and becomes soiled. If both designer and dyer have sets of the same collection, there is no guarantee that the two sets are identical, even when first purchased. A further problem is that, even with collections of more than a thousand samples, it is often found that the required colour is not present in the collection. This problem can be minimised by using a colour order system, in which the samples are arranged in an ordered fashion, enabling interpolation to be made between adjacent samples. Such systems have other advantages: they can be used for designers and students to understand colours and the relationships between them. The most common such system is the Munsell system.
Colour description/specification systems
41
In the Munsell system the samples are arranged along three scales, hue, value and chroma. Hue is the attribute by which we distinguish between red, yellow, green, etc. In the Munsell book, the samples are arranged so that all the samples on one page are the same hue as judged visually. Thus on a page of orange colours none should look to be redder or yellower than the rest. There are 40 such pages, 2.5YR, 5YR, 7.5YR, 10YR, 2.5Y, and so on, round to 2.5YR again. The Munsell value is roughly what is commonly called lightness. On one page of the Munsell Atlas, all the samples on one row have the same value, i.e. none appears lighter or darker than the rest. Rows higher up the page are lighter. The value scale runs from 0 (for black) to 10 (for white). The remaining scale, chroma, represents colourfulness or distance from neutral. The samples on one row vary in Munsell Chroma and range from neutral (grey) to the most colourful (or saturated) colour that can be produced without changing the hue or value. The chroma scale runs from 0 (for neutral) to a maximum of up to 14 depending on the samples that can physically be produced. One particular feature of this system is that, for each scale, the differences between neighbouring samples are visually all the same, as far as possible. Since the book was first produced, alterations have been made to improve the spacing. Unusually, the samples are also represented by their CIE specifications. Great care is taken to try to ensure that the samples are stable, and that each set is the same as all the others. Sets are available in matt and gloss paint. Other colour order systems all involve three dimensions, but the scales are somewhat different. All have the advantage that the colours can be seen, but the accuracy of colour representation is somewhat limited. The Natural Colour System (NCS) is based on the concept that there are four unique hues: red, yellow, green and blue and that these together with white and black make six basic colours. Colours are represented by the relative amounts of these basic colours perceived to be present. Thus a pure orange might be perceived as 50% each of red and yellow, while a dark brown might be 50% black and 25% red and yellow. Again, the samples on one page of the NCS Atlas are the same hue, but the other samples on the page are represented in a equilateral triangular arrangement, black to white again forming the vertical axis, with the purest possible representation of the hue at the other apex of the triangle.
2.9
Colour specifiers
Various sets of specifiers are available commercially, often with a good selection of colours, but are generally not arranged according to any particular pattern. It is particularly useful if the samples are on the same type
42
Total colour management in textiles
of material as the samples to be produced. It is easier to match a wool sample to another wool sample rather than to a gloss paint sample.
2.10 Future trends The CIE system is well established and it is unlikely that any major changes will be made in the foreseeable future. Some additions are possible. The D illuminants are good representations of daylight, but are almost impossible to reproduce accurately in colour measuring instruments and colour matching cabinets. A major problem lies in the relatively large amount of UV radiation. This leads to problems with materials treated with brightening agents (almost all textiles and papers). Many of these are generally seen indoors, when daylight has passed through glass, hence reducing the UV content. There has been a suggestion that a standard illuminant representing indoor daylight would be more applicable in many cases and easier to reproduce in practice. Details of the procedures for calculating tristimulus values from reflectance values will be tidied up. With respect to colour order systems, an enormous amount of effort has been put into systems such as Munsell and NCS and, while alternative arrangements are possible, it is unlikely that the effort required will be undertaken. Alternative colour specifier systems will quite possibly be produced for particular applications.
2.11 References 1. Grassman, H. (1853). Zur theorie der Farbenmischung. Poggendorf’s Annalen Physik Chemie 89, 69–84. 2. Grassman, H. (1853). Zur theorie der Farbenmischung. Phil. Mag. 7 (4), 254. 3. Blottiau, F. (1947). Les défauts d’additivité de la colorimétric thrichromatique. Rev. D’Opt. 26, 193. 4. Trezona, P.W. (1953). Additivity of colour equations. Proc. Phys. Soc. (London), B66, 548. 5. Trezona, P.W. (1954). Additivity of colour equations. Proc. Phys. Soc. (London), B67, 513. 6. Brindley, G.S. (1953). The effects on colour vision of adaptation to very bright lights. J. Physiol., 122(28), 332–350. 7. Trezona, P.W. (1972). Colour metrics, 36, Soesterberg: AIC/Holland. 8. Stiles, W.S. and Wyszecki, G. (1973). Rod intrusion in large-field color matching. Acta Chromatica, 3, 155–163. 9. Donaldson, R. (1947). A colorimeter with six matching stimuli. Proc. Phys. Soc. (London), 59, 554–560. 10. Wright, W.D. (1928–1929). A re-determination of the trichometric co-efficients of the spectral colours. Trans. Opt. Soc. (London), 30, 141–164. 11. Guild, J. (1931). The colorimetric properties of the spectrum. Phil. Trans. Roy. Soc. (London), A 230, 149–187.
Colour description/specification systems
43
12. Davis, R. and Gibson, K.S. (1931). Miscellaneous Publication 114. Washington, DC: National Bureau of Standards. 13. Judd, D.B., MacAdam, D.L. and Wyszecki, G. (1964). Spectral distribution of typical daylight as a function of correlated color temperature. J. Opt. Soc. Am. 54, 1031–1040. 14. Judd, D.B. (1949). A comparison of direct colorimetry of titanium pigments with their indirect colorimetry based on spectrophotometry and a standard observer. J. Opt. Soc. Am. 39, 945. 15. Stiles, W.S. and Burch, J.M. (1958). NPL colour-matching investigation: final report. Optica Acta, 6, 1–26. 16. Speronskaya, N.I. (1959). Determination of spectrum color co-ordinates for twenty-seven normal observers. Optics and Spectroscopy 7, 424–428 (USSR), English translation. 17. Johnson, S. et al. (1963). JSDC, 79, 731. 18. Foster, W.H. Jr et al. (1970). Weights for calculation of tristimulus values from 16 reflectance values. Colour Eng. 8, 35. 19. Stearns, E.I. (1975). Weights for calculation of tristimulus values. Clemson Rev. Ind. Managem. Tex. Sci. 14, 79–113. 20. Supplement No. 2 to CIE Publication No. 15 Colorimetry (E1.3.1) (TC1.3) Paris: Bureau Central de la CIE, 1978. 21. Newhall, S.M., Nickerson, D. and Judd, D.B. (1943). Final report of the O. S. A. Subcommittee on Spacing of the Munsell Colors. J. Opt. Soc. Amer. 33, 385.
3 Instrumental colour measurement P J C L A R K E, The Tintometer Ltd, UK
3.1
Introduction
Colour measurement instrumentation is very varied. It varies from large top of the range scanning spectrophotometers, maybe coupled with reflectance accessories through bench-top instruments, to hand-held small portable instruments. The instrumentation may be set up to make a variety of different colour measurements or to only make measurements on one particular colour scale. Colour measurements are essentially measurements of visible light shining through an object or visible light reflected from an object. There are many optical configurations to achieve the measurement of transmitted and reflected colour. There are standard or recommended geometries for making these measurements and nomenclature for describing them and the variations that are possible.1 The choice of measurement geometry usually depends on the properties of the artefact to be measured, but may be due to historical use within a particular industry or product area. There are colour measuring instruments available that can measure transmitted colour, reflected colour or both.
3.2
Types of colour measurement
Colour measurement divides into two areas, reflectance and transmittance. Each of these can be further divided into diffuse and regular. Regular means the light travels through undeviated or is reflected off the sample in a mirror-like way without change of frequency, and diffuse means that the light is scattered as it is reflected by, or transmitted through, a sample, again without change of frequency of the light. The total reflectance or transmittance is the sum of the regular and diffuse parts. The geometry of illumination and detection with respect to the sample is used to define the measurement geometry. Reflectance and transmittance are defined in terms of the ratio of the incident light to the reflected or transmitted light. These 44
Instrumental colour measurement
45
are special cases of the more general definitions of radiance or luminance factor, which are defined in terms of measurements made for a specific illumination and detection geometry in comparison to the perfect reflecting diffuser identically illuminated.
3.2.1 Regular reflectance Regular reflectance is the regularly reflected or specular component of the illumination, i.e. light that is reflected in a mirror-like way off a surface at the same angle and in the same plane as the illumination beam. It usually has the same spectral profile as the illuminating source. When measuring colour with an integrating sphere, this component is either included or excluded from the measurement. It is not usual when making colour measurements to measure this component separately, as it has the same spectral profile as the illumination. Regular reflectance is usually only measured on its own when the sample has a very high regular reflectance, such as a mirror.
3.2.2 Diffuse reflectance Diffuse reflectance is probably the measurement that most people think of when referring to colour measurement. This is measurement of light scattered from a surface and is most commonly measured using an instrument incorporating an integrating sphere. Quite often, the measurement referred to as diffuse reflectance by some users is not strictly the diffuse reflectance, but includes the regular component and is actually the total reflectance. The simplest measurement geometry is illumination at the sample normal or near normal and detection over the whole hemisphere, but excluding the regular or specular component. The instrumentation could also be set up the other way, due to Helmholtz reciprocity and the reversibility of light paths, with diffuse illumination and detection at the normal or near normal. For reflectance measurements, the trade-off is between light level and the ratio of the sphere surface area to total port area. The more ports and/or the bigger the ports in a sphere, the larger the sphere needs to be. The CIE has recommended that the total area of ports in a sphere should be less than 10% of the total surface area.2
3.2.3 Regular transmittance Regular transmittance is light that is undeviated as it passes through a sample. The sample may attenuate (absorb) the light but the direction is not changed. This is the normal type of light transmission such as looking through a clear pane of glass.
46
Total colour management in textiles
3.2.4 Diffuse transmission Light that does not pass through a sample in a regular way is diffuse transmission, i.e. the light direction has been changed. An example of this is the covers to lamps that spread the illumination over a larger area than would be possible without them.
3.2.5 Radiance factor and transmission factor Radiance factor and transmission factor are the fundamentals of the reflectance and transmittance measurement geometries, as the others can be made up from a series of radiance or transmittance factor measurements. At its simplest, the radiance factor refers to illumination of a sample in a specific direction, over a specific angular range, with detection at another specified direction and angular range. The most common colour measurement geometry of this type is illumination at the normal (0°) to a sample with detection at 45° to the sample normal (or vice versa).
3.2.6 Geometries of measurement There are endless ways of arranging the optical system to make colour measurements. The colour of a sample varies, depending on the way it is measured. A sample may reflect differently depending on the illumination angle, whether the sample is translucent, the state of polarisation of the illumination, the detection angular range, the way the detector measures the reflected light, etc. For example, illuminating a sample from a particular direction may not give the same result as illuminating it from all directions equally (diffusely). For measurements made on the same sample on different instruments to be comparable they have to be made in a consistent way with regard to geometry, and also with respect to other parameters such as traceability, orientation, polarization, etc. To this end the CIE1 recommends the use of three reflectance measurement geometries: specular included, specular excluded and 0°/45° and their reciprocal geometries. They have recommended geometric arrangements for directional, conic and integrating sphere illumination or influx angles, and detection or efflux angles. The beam cone sizes, and their tolerances are defined along with descriptive nomenclature for the measurement geometry. Use of these geometries is not mandatory but it makes sharing and comparing measurements with others a possibility.
3.2.7 Sample-induced effects Some samples will introduce other effects that make colour measurement difficult in a conventional way. These effects included translucency – the
Instrumental colour measurement
47
sideways spreading of light; fluorescence – emission of light at a different wavelength from the incident light; metallics – metal flakes within a surface coating that have mirror like properties, but are orientated parallel to the sample surface; interferometric effects – the colour is different depending on the illumination and viewing conditions. All of these are difficult to measure in a conventional manner with an integrating sphere or by using a single angle geometry. There are instruments available that can alter the angles of illumination and detection so that they can measure many combinations to build an overall picture of the sample.
3.3
Colour measuring instrumentation
There are many reasons for wanting to measure colour, but it is important to have a colour measuring instrument that will measure what is required with the necessary accuracy. Firstly, the user needs to determine what the instrument specification is for the task. This may mean defining the spectral range, the data interval, the bandwidth, the colorimetric data required, the measurement geometry, the measurement area, the accuracy, the precision, the traceability, reference artefacts, software, export of data options, etc. There may be a specification to meet in terms of measurement geometry, measurement area/instrument port size. Usually, the type of samples that are being measured will define the most suitable measurement geometry, but standards and common usage within certain industries will also have an influence. The CIE recommend measurement geometries1 and parameters for port sizes and other geometric considerations.2 ASTM also recommends geometries for colour measurement, especially for single colour scales.3 Special effect/appearance materials will need careful consideration as to the choice of a suitable measurement instrument. Precision is a measure of how well an instrument repeats a measurement. Accuracy is a measure of how well the instrument will make measurements compared to particular standard reference materials, which should be traceable to a national standards laboratory (see later section for more details). The instrument may come with a calibration artefact traceable to a particular national measurement institute. There may be differences in scales between national measurement institutes so the users need to be aware of the traceability route of their instruments and any with which they are comparing measurements.
3.3.1 Single-scale instruments These are instruments that are primarily for measuring a specific colour measurement scale, in order to ascertain where the sample lies on the scale. These types of instruments are used in medicine, chemistry, manufacturing
48
Total colour management in textiles
control, food processing, etc. Examples of scales and the regulatory body defining the scale are: edible oils – American Oil Chemists’ Society (AOCS), honey – mm Pfund, maple syrup – USDA, Vermont Department of Agriculture, and Canadian; fruit juice such as apple, pear, white grape – International Fruit Juice Union (IFU), platinum–cobalt/hazon – water quality, solvent quality, beer – European Brewing Convention (EBC), colour – Gardner, Saybolt, ASTM colour, etc. These types of instruments are usually associated with measurements of a specific product such as petroleum, honey, beer, oil and chlorophyll. These are designed to assess the quality of a particular product by linking colour to purity, refinement level, impurity indicator, desirability, etc. Many of these instruments measure the colour of a liquid or require the product, e.g. waxes, to be measured in a liquid form. The most common instruments for this type of measurement are those from Tintometer Ltd such as the Comparator 2000/3000, PFX195, PFX880 and PFX995. Many of these instruments incorporate calibrated colour standards or come with calibrated reference materials for the particular scale of interest, and can be operated as a stand-alone instrument or as a computer-controlled instrument.
3.3.2 Visual instruments Visual instruments are usually of a comparative nature. They allow a sample to be viewed under the same conditions as a reference artefact or artefacts and the user determines if they are a match. The most common of these would be a simple light box, which usually has neutral grey walls, and a choice of light sources designed to simulate recommended sources. The reference and the test samples can be viewed side by side. This is how the textile industry used to do much of its colour matching. Alternative instruments are those such as The Tintometer model F, for visual assessment of transmitting samples using the Lovibond RYBN scale.4 A light source and white diffuser provide the illumination, with the viewing being done through an eyepiece. A bipartite field is presented to the user. The test sample is placed in one side of the field and the user moves neutral, red, blue and yellow glass filters into the other half of the field to match the colour. The filters used will indicate the colour in terms of Lovibond units. Other instruments act as comparators and glass discs of values of the single colour scales mentioned above can be rotated into place to match the sample.
3.3.3 Hand-held/portable instruments Hand-held instruments come in a variety of types, from small spectrophotometers to much simpler colourimeters. Generally, these are small instruments, measuring a small area. They have the advantage that they can be
Instrumental colour measurement
49
taken to the object to be measured instead of bringing the sample to the instrument. This means that finished products such as cars, vehicle interiors, large objects and interior decoration can be measured in situ without the need to remove samples for testing. The disadvantage of these types of instruments is that there are compromises in measurement performance in order to achieve portability. Sphere-based instruments will generally have a smaller size sphere than the benchtop equivalent, functions available may be more limited, and data resolution is likely to be lower. If the user has a held-hand colorimeter, it may be limited to certain illuminants and observer combinations for colorimetric data. However, many have the capability to store a large number of measurements, for downloading later to a computer. The measurement uncertainty from one of these instruments is almost certainly higher than a larger benchtop instrument because of these factors. In some cases the measurement geometry available is not truly one of the recommended CIE geometries, but a compromise between these and realising a practical result. So long as the user is aware of the limitation, this type of instrument is more than adequate for a lot of colour measurements. Example instruments are Minolta spectrophotometers and colorimeters; some of these can be used to obtain displays as well as other measurements. Tintometer and X-Rite have a range of portable 0°/45° and integrating sphere instruments. Aventes produces the Ava-mouse, which plugs into a computer and has a similar shape and size to a computer mouse, but gives 45°/0° data.
3.3.4 Multiangle instruments Multiangle instruments have a lot of the features of other instrumentation, but offer several fixed angle geometries, rather than just 0°/45° or a spherebased geometry. They are an attempt to measure appearance or special effect samples. By looking at their properties at several angles, the user can pick out the best angles for meaningful results on the sample types being investigated. These are a cheap alternative to goniometric measurements. One popular instrument of this type is the X-Rite MA68, measuring at five detection angles, and variants which have found favour with the automotive industry as they can provide useful results on metallic paints.
3.3.5 Benchtop instruments Most of the colour measuring instrument manufacturers sell more instruments of this type than any other. For reflectance measurements, these instruments generally incorporate diode array detectors in their optical systems. The light source is generally a xenon flash tube. Because of these
50
Total colour management in textiles
features, measurements can be made very rapidly. Some are quite sophisticated in that they have double gratings and arrays in them, one viewing the sphere wall or reference and the other the sample, enabling them to compensate for the light source. Others achieve this by using one array with moveable optics and two xenon flashes, one for viewing the sample and the other for viewing the sphere wall or reference. The geometry of these instruments is usually in the alternative configuration from scanning instruments. Bench-top instruments usually have diffuse or directional illumination with detection at the normal or near normal, e.g. d/8° geometry. Bench-top instruments are mainly set up to perform comparisons of one sample against a reference, for example, in a dye house for matching a production sample to the required reference colour. The software allows for easy comparison of samples and references. Tolerances on the closeness of colour match can be set to the user’s requirements. For reflectance instruments the data resolution may vary from 5 nm at the top end through 10 nm to 20 nm. The spectral range available is also variable. All spectral instruments will include the range 400 nm to 700 nm, but the better instruments will offer wavelengths outside this range, down to 360 nm and up to 780 nm. For transmittance instruments measurements in the range 200 nm to 800 nm plus will generally be available with at least 1 nm resolution. Spectral range and resolution are important for the type of samples to be measured. Those with sharp or rapidly changing features will require more point measurements to avoid features being smoothed out or missed. Bench-top reflectance instruments are usually supplied with a white reference standard. The user needs to be aware as to where it is traceable to. The instrument should also come with a good black reference that approximates to zero – a black trap for integrating sphere geometries and a black glass for 0°/45° geometry. Typical manufacturers are Datacolor, GretagMacbeth, X-Rite, Tintometer and Analytik.
3.3.6 Scanning instruments Scanning instruments are the most complex of the colour measuring instruments available. They require the user to have a good knowledge of the type of measurements they wish to take. The user is able to have control of, and set, a large range of parameters, and may be able to set some parameters such that the measurement made is nonsense or highly distorted, for example, a high scanning speed coupled with a long measurement time on a sample with a lot of spectral structure. What scanning instruments do allow is the expert user to change parameters such as bandwidth, scan speed and wavelength interval, to suit the type of sample being measured. Scanning
Instrumental colour measurement
51
instruments are slower than other types of instrumentation, but they offer wider spectral ranges, such as into the UV and NIR, which may not be useful for colour measurement but may be useful for other applications or research. These types of instruments have the potential to offer the highest accuracy and precision of all the commercially available instruments. They are available in double beam and single beam options, the double beam generally giving more stable results as they compensate for fluctuations in the light source. These instruments are primarily designed for regular transmittance measurements; examples are available from Perkin-Elmer, Varian, Shimadzu, Thermo (Unicam) and Hitachi, among others. For diffuse reflectance and diffuse transmittance measurements, it is usually possible to buy a standard or custom reflectance accessory, either from the instrument manufacturer or from a specialist supplier such as Labsphere. These accessories either replace the regular transmittance accessory or can be fixed into place on the transmittance accessory or fit inside the transmitting sample compartment. The results obtained are usually just the spectral data. The processing of the data to correct its traceability and calculation of the colorimetric data may not be possible with the instrument manufacturer’s software, but may need to be done outside of the instrument software or by purchasing special software.
3.3.7 Goniometric instruments Goniometric instruments are the most exciting instruments and more will become available in the future. These instruments allow light to be put onto a sample at a wide variety of angles, and then detected at a wide variety of angles, quite often over the hemisphere. This makes this type of instrument very useful when measuring special effect samples, as the best parameters in terms of illumination and detection can be chosen by the user. These instruments are very versatile, but also require a lot of specialist knowledge to operate them. Some of this type of instrument are self-built and operated by national standards laboratories. Such instruments included the STARR at NIST and the reference reflectometer at the National Physical Laboratory (NPL). The operation of these instruments can be quite time consuming. Faster commercial instruments are available from Murakami in Japan,5 and Tintometer have developed a gonio apparent spectrophotometer (GASP), the first one being for NPL, UK.6
3.3.8 Camera Cameras are a growing area for colour measurement. A camera is used to view a sample with the red (R), green (G) and blue (B) values being
52
Total colour management in textiles
evaluated and used to generate colorimetric data. The advantage of a camera is that an image can be taken and colorimetry conducted throughout the image on a pixel-by-pixel basis. This relies on a good light source approximating to a standard illuminant and some good camera characterisation to map the R, G and B values to X, Y and Z colour functions. The advantage of this is that small sample areas, or samples that are nonuniform, can be measured. There are systems available that can do this: basically a lightbox with light sources for one or more standard illuminants. The illumination can be arranged to approximate to a standard geometry. The sample is placed in the light box and then viewed by the camera. Usually a test target, such as one of the Macbeth colour checker charts, is used to calibrate and characterise the camera. Associated software will control the calibration and measure the sample under test. These systems are well suited to measuring products that are variable and consist of multiple similar items, for example foodstuff such as cereals. The camera can be combined with software to pick out individual samples, and makes it possible to discern the variation in the product. One of the systems available – DigiEye,7 in addition uses algorithms based on colour inconstancy indexes to compute a synthesised reflectance spectrum for the test sample, which has a stable colour across several standard illuminants. Texture and gloss measurements can also be made by using only one of the pair of lamps for each illuminant, resulting in a directional illumination. This provides a shadow, and software can analyse the pattern of light and dark pixels to provide texture and gloss information. Another camera system is Tintometer’s CAM 500 system for colorimetry and product colour quality control.
3.3.9 Custom instruments Custom instruments include any system that does not readily fall into the other categories. It could be a home-built system, such as at a national standards laboratory, a commercial system with user specified enhancements or it might be a system for measuring a specific product on-line. These types of systems are usually put together from a variety of standard components, but integrated with custom software, etc.
3.3.10 Fluorimeter This instrument is specifically designed to get round the problems of measurements on fluorescent samples without using standard illumination sources. True fluorimeters are able to illuminate and detect independently, by using two monochromator systems, or filters in such a way that the fluorescent component can be separated from the normal reflectance or trans-
Instrumental colour measurement
53
mittance component. The overall colour for any observer and illuminant combination can then be calculated
3.4
Inter-instrument agreement and traceability
Agreement and traceability are important issues for the user who has several colour measuring instruments or who wishes to share or compare measurements with another user’s instrument. If the user has several identical instruments from the same manufacturer purchased at the same time, they should agree with each other. If they are different models or have been purchased over a longer time period, the probability of good agreement is much less. The instruments to be compared should have the same geometric specification for there to be a chance of agreement. Then the two instruments need to be traceable to the same measurement scale. Traceability refers to showing an unbroken link from a scale disseminated by a national measurement institute, such as NPL in the UK, NIST in the USA, BAM in Germany, etc., to the scale the instrument is using, with an appropriate uncertainty statement. The scales disseminated by the national measurement institutes are the most accurate, with the lowest uncertainties. As the scale is disseminated downwards, each level adds a further degree of uncertainty. For two instruments from the same manufacturer, the scale to which they are traceable is likely to be the same. Most manufacturers have a master instrument to which all the instruments they produce are compared, and this instrument in turn is traceable to a national measurement institute. However, there are, at the time of writing, differences between national measurement institute scales of reflectance, and scales from national measurement institutes can change with time. National measurement institutes compare their scales with each other, work to understand the causes of differences and then try to minimise these differences in their scales. The best way to check that instruments are traceable to the same scale is to use an independent check, rather than relying on the scale the manufacturer has used. The easiest way to do this is to purchase some calibrated colour transfer standards from a national measurement institute or from an ISO 17025 accredited laboratory. Suitable accredited laboratories are those that are UKAS or equivalently accredited. The standards can then be measured using your own instruments and the results compared to the calibration values for the standards. If they do not agree within the combination of the uncertainty of the measurement and the uncertainty of the calibration of the standard, appropriate corrections can be calculated to correct the measurements on your instruments. Another issue that raises itself here is the different way in which each manufacturer’s instrument communicates the measurement results. This is
54
Total colour management in textiles
usually in some proprietary way that is incompatible with anyone else’s way. The NPL, The Society of Dyers and Colourists and The University of Leeds with others, mainly in the UK, are working with instrument manufacturers and users to define a format for the exchange of colorimetric data. This format is HTML based. This means that the data files produced are totally text based. The format has been defined and is currently being proposed as an ISO standard. The aim is for instrument manufacturers and software writers to incorporate the format into instruments and into their control software in order to make it easy to import and export measurement results so that measurements can be compared. This, along with suitable traceability, makes it possible for a particular colour to be specified numerically for matching, without needing to send a sample of the colour for measurement, and without being restricted to one manufacturer in the choice of instrument for measuring the samples.
3.5
Future trends
Trends are towards smaller instruments with more elements to their diode arrays, providing better spectral measurement resolution. Coupled with this, instrumentation of all types will become cheaper in real terms and offer more features to the user as advances in electronics lead to cost savings on electronic parts. Better and faster computers mean that instruments can be much more basic in terms of onboard processing with only the raw measurements being taken and passed directly to computer software for very fast manipulation before presentation as the measurement result. Transfer of data between instruments and other colour evaluation software will increase, as users are able to afford a variety of instruments from different manufacturers, but want to process results with the same piece of software. Software will also allow better portability of databases of samples built up over many years. The issue of traceability will become easier to handle as the demand increases for colour and spectral data standards to be transmitted electronically rather than with physical samples. Electronic transmission demands that instruments are calibrated to the same source scale for the data to be meaningful. Manufacturers’ software will allow data for the instrument calibration standards to be defined for several traceable sources. The improved measurement software will also offer better export of data and supporting information, and for exchanging measurement data. More and faster goniometric instruments will become available as they become relatively cheaper and as people attempt to measure and quantify samples by appearance effects. Software for controlling these instruments and providing data manipulation and analysis of the measurement results will allow 3D visualisation of the results.
Instrumental colour measurement
55
Cameras will be used increasingly for colour measurement as their resolution increases greatly and the need to do separate r, g and b measurements using filters in front of the camera becomes eliminated by cameras incorporating a matrix of r, g and b sensitive pixels to measure colours directly.
3.6
Sources of further information and advice
Accredited laboratories NPL ORM Club Society of Dyers and Colourists Colour group GB
www.ukas.org.uk www.npl.co.uk/opticalradiation/orm www.sdc.org.uk www.colour.org.uk
3.6.1 Instrument manufacturers websites Avantes Analytik UK Bentham BYK-Gardner Carl Zeiss Cecil Instruments Datacolor Digieye D R Lange GretagMacbeth Hitachi Hunterlab International Light Jovin-Yvon Kirstol Labsphere Murakami Konica Minolta Ocean Optics Perkin-Elmer Photo Research Spectro Solutions Shimadzu Thermo The Tintometer Ltd Tricor Systems Varian Verivide X-rite
www.avantes.com www.analytik.co.uk www.bentham.co.uk www.bykgardner.com www.zeiss.co.uk www.cecilinstruments.com www.datacolor.com www.digieyeplc.com www.drlange.co.uk www.gretagmacbeth.com www.hitachi-hta.com www.hunterlab.com www.intl-light.com www.jyhoriba.co.uk www.kirstol.co.uk www.labsphere.com www.colourmeasure.com www.konicaminolta.com www.oceanoptics.com www.perkinelmer.com www.photoresearch.com www.spectrosolutions.ch www.shimadzu.com www.thermo.com www.tintometer.com www.tricor-systems.com www.varian.com www.verivide.com www.xrite.com
56
Total colour management in textiles
3.7
References
1. CIE 15:2004 Colorimetry, CIE Central Bureau, Austria (2004). 2. CIE 130:1998, Practical methods for the measurement of reflectance and transmittance, CIE central bureau, Austria (1998). 3. ASTM international standards on color and appearance measurement, seventh edition 2004, ASTM, PA, USA. 4. www.tintometer.com. 5. www.colourmeasure.com. 6. A new goniospectrophotometer for measuring gonio-apparent materials, Pointer, M.R., Barnes, N.J., Clarke P.J. and Shaw, M.J. Coloration Technology, 121, 96–103 (2005). 7. Luo, M.R., Cui G.H. and Li C., British patent 0124683.4, Apparatus and method for measuring colour (DigiEye system), Derby University Enterprises Limited, Oct 2001.
4 Colour quality evaluation M R L U O, University of Leeds, UK
4.1
Introduction
Instrumental colour measurement systems have been widely used by colour-using industries such as textiles, coatings, plastics, graphic arts and imaging. The most important application is undoubtedly colour quality control by means of colour difference formulae, which are used to quantify colour variations between pairs of specimens. Conventionally, this task was carried out by experienced colourists, but more recently this was replaced by instrumental methods in order to reduce labour costs, save time and apply a more scientific methodology. Some typical colour quality control tasks include: • setting the magnitude of tolerance for making instrumental pass/fail decisions; • evaluating fastness grades for assessing change in colour and staining; • predicting the metameric effect between a pair of specimens; • determining the change in colour appearance of a single specimen across different illuminants. All the above tasks rely on the availability of a robust colour difference formula. This has long been eagerly sought by industry. This chapter will briefly review the development of colour difference formulae, including the CIE 2000 colour difference equation, CIEDE2000.1,2 An example will be given to illustrate the method for establishing a tolerance value for industrial applications. In addition, methods for calculating a metamerism index, relating to changes in illumination of a pair of samples, and a colour inconstancy index, recently proposed by the Colour Measurement Committee (CMC) of the Society of Dyers and Colourists (SDC), will be introduced. Finally, a summary will be given of some new developments. 57
58
Total colour management in textiles
4.2
Colour difference formulae
The development of colour difference formulae can be divided into three stages: before, during and after 1976. Over 20 separate formulae were derived prior to 1976. They can be grouped into three families: those derived to fit MacAdam ellipses3, Munsell4 data, and those transformed linearly from CIE tristimulus colour space.5 Some formulae are still in use today. Some representative formulae from each group are FMC2,6 ANLAB7 and HunterLAB,8 respectively. However, significant progress was really made after the recommendation of CIELAB and CIELUV5 in 1976.
4.2.1 CIE L*a*b* Formula (CIELAB) In 1976, the CIE recommended two uniform colour spaces: CIELAB (or CIE L*a*b*) and CIELUV (or CIE L*u*v*) for industries concerned with the subtractive mixture (surface coloration) and additive mixture of colour (e.g. TV), respectively. Although the agreement of these two formulae with the then available experimental data was generally not good, they worked at least equally as well as any of the alternatives. CIELAB has been more widely used than CIELUV, especially in the surface colour industries and so only this is detailed here. L* = 116 f(Y/Yn) - 16
a* = 500 [f(X/Xn) - f(Y/Yn)]
[4.1]
b* = 200 [f(Y/Yn) - f(Z/Zn)] where f(I) = I1/3, for I > 0.008856 Otherwise, f(I) = 7.787 I + 16/116
where X, Y, Z and Xn, Yn, Zn are the tristimulus values of the sample and a specific reference white considered. It is common to use the tristimulus values of a CIE standard illuminant or a light source for the Xn, Yn, Zn values. Correlates of hue and chroma, given in (4.2), are defined by converting the rectangular a*, b* axes into polar coordinates. The lightness (L*), chroma (C*) and hue (hab) correlates correspond to perceived colour attributes, which are generally much easier to understand when describing colours. hab = tan-1 (b*/a*) C*ab = (a*2+ b*2)1/2
[4.2]
Colour quality evaluation
59
A three-dimensional representation of the CIELAB colour space is shown in Fig. 4.1. The neutral scale is located in the centre of the colour space. The L* values of 0 and 100 represent a reference black and white, respectively. The a* and b* values represent redness–greenness, and yellowness– blueness attributes, respectively. The C*ab scale is an open-ended scale with a zero origin. (This origin includes all colours in the neutral scale, which do not exhibit hue.) The hue angle, hab, lies between 0° and 360°. Colours are arranged following the sequence of rainbow colours. The four unitary hues (pure red, yellow, green and blue) do not lie exactly at the hue angles of 0°, 90°, 180° and 270°, respectively. Colour difference, represented by DE*ab, is given in (eqn 4.3) and is calculated as the distance between the standard and sample in the CIELAB colour space. DE*ab = (DL*2 + Da*2 + Db*2)1/2 or DE*ab = (DL*2 + DC*ab2 + DH*ab2)1/2
[4.3]
L* 100
White
Yellow (+) b*
Cab * hab
Green (–)
Red (+) a*
Blue (–)
0
Black
4.1 A three-dimensional representation of the CIELAB colour space.
60
Total colour management in textiles
where DH*ab = 2 (C*ab,1 C*ab,2)1/2sin[(hab,2 - hab,1)/2] and subscripts 1 and 2 represent the standard and sample of the pair considered, respectively. Although the CIELAB colour difference formula is by no means perfect (see later), its colour space is still the most widely used, mainly because it is relatively easy to relate colours as seen with positions on its diagram.
4.2.2 Formulae developed after CIELAB As mentioned previously, the earlier formulae were derived mainly to fit the Munsell and MacAdam data. The viewing conditions applied in these experiments are very different from those used in the surface industries such as textiles and paint. Many sets of experimental results on colour discrimination have been published since 1976 and most of them were conducted using large surface samples viewed under typical industrial viewing conditions. Of these, the important data sets in terms of numbers of observers and sample pairs, and smaller observer variations, are those accumulated by McDonald,9 Luo and Rigg,10,11 RIT-Dupont,12,13 Kim and Nobbs,14 Witt,15 Chou et al.16 and Cui et al.17 These data sets were used to develop or to verify more advanced formulae: CMC(kL : kC),18 BFD(kL : kC),19 CIE9420 and LCD.14 (In general, one or two of these data sets were used to develop each formula.) Finally, all of these data sets were used to develop the CIEDE2000 formula.1,2 These more advanced formulae have a common feature: they are all modified versions of CIELAB and have no associated colour space. Only three equations – CMC, CIE94 and CIEDE2000 – are introduced here because they have been adopted by standards organisations such as CIE and ISO. CMC(kL: kC) and JPC79 colour-difference formulae McDonald9 at J.P. Coates company accumulated a comprehensive data set. These visual results were used to derive the JPC79 formula.21 At a later stage, the formula was further studied by the members from the CMC of the SDC and it was modified to correct some anomalies. The modified formula is named CMC(kL: kC)18 and is the current ISO standard for the textile industry. The formula is given in (eqn 4.4). DECMC = [(DL*/kLSL)2 + (DC*ab/kCSC)2 + (DH*ab/SH)2]1/2 where SL = 0.040975L*ab,1/(1 + 0.01765L*ab,1)
[4.4]
Colour quality evaluation
61
unless L*ab,1 < 16 when SL = 0.511 SC = 0.638 + 0.0638 C*ab,1/(1 + 0.0131C*ab,1) SH = SC(Tf + 1 - f) f = [C*ab,14/(C*ab,14 + 1900)]1/2 T = 0.36 + | 0.4 cos(hab,1 + 35°) | unless hab,1 is between 164° and 345° when T = 0.56 + | 0.2 cos(hab,1 + 168°) | where DL*, DC* and DH* are the CIELAB lightness, chroma and hue differences (‘batch’ minus ‘standard’). The L*ab,1, C*ab,1 and hab,1 refer to the ‘standard’ of a pair of samples. The kL and kC parametric factors were included to allow different weights for lightness and chroma, respectively, to be used depending on the circumstances. The best kL and kC values have been found to be 2 and 1, respectively, for predicting the acceptability of colour differences for textiles. For predicting the perceptibility of colour differences, kL and kC should both equal 1. A constant DE according to the CMC formula (eqn 4.4) can be considered as an ellipsoid equation in CIELAB L*, C* and hue polar space with semi-major axes of kLSL, kCSC and SH, respectively. Its chromaticity ellipse points towards the achromatic axis.
CIE94(kL : kC : kH) colour-difference formula Berns et al.12,13 at the Rochester Institute of Technology (RIT) also accumulated visual assessments using glossy acrylic paint pairs. This data set is named RIT–Dupont. A colour difference formula given in (eqn 4.5), having a similar structure to that of CMC(kL : kC) but having simpler weighting functions, can fit their data very well. They believed that the CMC formula is over-complicated. (In fitting any formula to a particular set or sets of experimental data, it is always possible to obtain a better fit by making the formula more complex. It is often difficult to judge just how much complexity is justifiable.) The formula was later recommended by the CIE for field trials in 199420 and it is thus named the CIE94 colour difference formula. DE94 = [(DL*ab /kLSL)2 + (DC*ab /kCSC)2 + (DH*ab /KHSH)]1/2
[4.5]
62
Total colour management in textiles
where SL = 1 SC = 1 + 0.045C*ab,1 SH = 1 + 0.015C*ab,1 where C*ab,1 refers to the C*ab of the standard of a pair of samples. In some situations, such as calculating large magnitude colour difference, the geometric mean is suggested rather than using C*ab,1. The kL, kC and kH terms are parametric factors accounting for variation in experimental conditions such as luminance level, background, texture and separation. For all applications except for the textile industry, a value of 1 is recommended for all parametric factors. For the textile industry, the kL factor should be 2 and the kC and kH factors should be 1, i.e. CIE94(2 : 1 : 1). The parametric factors may be defined by industry groups, depending on the typical viewing conditions for that industry. CIEDE2000(KL : KC : KH) colour-difference formula After the development of the CIE94 formula, two separate colour difference equations recommended by different organisations co-existed, i.e. CMC(kL : kC) by ISO22 and CIE94 by the CIE.20 However, these formulae were derived from two main data sets: Luo & Rigg10,11 and RIT-DuPont.12,13 Figure 4.2 shows both sets of experimental ellipses – plotted with dashed lines. If the CIELAB formula agreed perfectly with the experimental results, all ellipses should be constant radius circles. Hence, the patterns shown in Fig. 4.2 indicate a poor performance by CIELAB. A clear pattern of ellipses can be seen, i.e. very small ellipses for neutral colours, increasing in size when chroma increases. All ellipses for the blue region point away from the neutral axis, whereas ellipses for all other colour regions generally point towards the neutral point. The latter phenomenon indicates that both the CMC and CIE94 formulae would badly model experimental results in the blue region because the ellipses representing both formulae all point exactly towards neutral. Detailed comparisons of these two formulae reveal there are large discrepancies in predicting lightness differences, and both have errors in predicting colour difference in the grey and blue regions.23 With this in mind, a CIE Technical Committee (TC) 1–47 on ‘Hue and Lightness Dependent Correction to Industrial Colour Difference Evaluation’ led by Alman of DuPont was formed in 1998. It was hoped that a generalised and reliable formula could be achieved. The members in this TC worked closely together using four selected experimental data sets: Luo and Rigg,10,11 RIT-Dupont,12,13 Kim and Nobbs14 and Witt.15 A formula named CIEDE2000 – see (eqn 4.6) – was then published.1,2 It includes five modifications to CIELAB: a lightness weighting
Colour quality evaluation
63
120
100
80
60
b*
40
20
–60
–40
–20
0
0
20
40
60
80
100
120
–20
–40
–60
a*
4.2 RIT-DuPont and BFD experimental chromaticity discrimination ellipses (in ) compared to the corresponding ellipses from the CIEDE2000 equation (in ).
function, SL, a chroma weighting function, SC, a hue weighting function, SH, an interactive term, RT, between chroma and hue differences for improving the performance for blue colours, and a factor, 1 + G, for re-scaling the CIELAB a* scale to improve performance with grey colours. The CIE94 and CMC formulae only included the first three corrections and so vary markedly from CIEDE2000 for chromatic differences in the blue and neutral regions. The performance of CIEDE2000 is shown in Fig. 4.2, where solid line ellipses correspond to a constant CIEDE2000 colour difference. It can be seen that these solid line ellipses fit very well to the majority of the experiment ellipses plotted with dashed lines. Melgosa and Huertas24 also later found that CIEDE2000 is more accurate than CMC at a statistical significance at 95% confidence interval for the combined data set, which includes the above four data sets. More recent experimental results have been published to verify the performance of CIEDE2000. They all reached the same conclusion that the
64
Total colour management in textiles
CMC, CIE94 and CIEDE2000 out-performed CIELAB by a large margin. However, they all gave a very similar degree of accuracy, except for the blue and near neutral colour regions, for which only CIEDE2000 gave an accurate prediction. 2 2 2 ∆L ′ ∆ C ∆ H ∆ C ∆ H ′ ′ ′ ′ + [4.6] ∆E00 = + + RT kC SC kC SC kH SH kL SL kH SH
( )
(
)
where SL = 1 +
0.015 ( L ′ − 50 )
2
20 + ( L ′ − 50 )
2
and SC = 1 + 0.045 C¢ and SH = 1 + 0.015 C¢T where T = 1 - 0.17 cos( h¢ - 30°) + 0.24 cos(2 h¢) + 0.32 cos(3 h¢ + 6°) - 0.20 cos(4 h¢ - 63°) and RT = -sin(2Dq)RC where Dq = 30 exp{-[( h¢ - 275°)/25]2} and Rc = 2
C′
7
7
C ′ + 25 7
and L¢ = L* a¢ = (1 + G)a* b¢ = b* C ′ = a ′ 2 + b′ 2 h¢ = tan-1(b¢/a¢)
Colour quality evaluation
65
where 7 C* G = 0.5 1 − 7 C * + 25 7
4.2.3 Establishing industrial colour tolerance As mentioned earlier, successful colour quality control is heavily dependent upon the use of a reliable colour difference formula. Furthermore, there is a need to set up a magnitude of tolerance to determine whether a batch is within tolerance (pass) or outside of it (fail) relative to the standard. To obtain a reliable tolerance requires visual data for assessing colour difference pairs in terms of pass or fail. Experimental A set of paint samples was taken as a worked example here. It includes a grey colour centre having CIELAB values of 51.0, 0.2 and 1.2 for L*, a* and b*, respectively. Each sample had a size of 5 ¥ 10 cm. In total, 38 pairs of samples were selected. Figure 4.3 shows the distribution of sample pairs surrounding the colour centre in the Da* Db* (left) and DL* DC* (right) diagrams. It can be seen that these pairs gave a good coverage for almost all directions. Each sample was measured with a spectrophotometer in terms of CIELAB values under CIE Illuminant D65 and using the CIE 1964 standard colorimetric observer.
(a)
(b)
–0.6 –0.4
–0.2
1.2
1
1
0.8
0.8
0.6
0.6
DC *
Db*
1.2
0.4
0.4
0.2
0.2
0
0 Da*
0.2
0.4
0.6
–0.6 –0.4 –0.2
0
0 DL*
0.2
0.4
4.3 The sample pair distribution in (a) Da* Db* and (b) DL* DC* diagrams.
0.6
66
Total colour management in textiles
Note that it is important to select the sample pairs carefully when determining tolerance. All pairs should have perceptible colour differences around the pass/fail borderline. If their magnitudes are too large, all pairs will be rejected by assessors. Similarly, all pairs will be accepted by assessors for small colour difference magnitudes. The results of these pairs cannot be used in the following data analysis. These pairs were judged by a panel of ten professional assessors in terms of ‘pass’ or ‘fail’. Each assessor performed their judgements twice for each pair. The assessment was carried out in a viewing cabinet, which included a high quality D65 simulator. Measure of fit: wrong decision The data accumulated in the above experiment are described in terms of percentage acceptance (A%), in which a batch sample is judged as pass against a standard in percentage. For example, an A% of 30 indicates that 30% of observers regard the batch as an acceptable match to the standard. The ‘wrong decision’ measure25 was used to indicate the observer accuracy, observer repeatability and colour difference equation’s performance. When investigating observer accuracy, each individual observer’s results were compared with the panel results. For a panel result of 35 A% for a batch sample, it is considered as a fail decision because 65% of observers (a majority) reject it. If an individual observer passed it, it will be counted as a wrong decision. Finally, the performance is expressed by WD%, which is the number of pairs with the wrong decision divided by the total number of sample pairs. For a perfect agreement, WD% should be zero. When examining observer repeatability, the WD% measure was used to represent the number of wrong decisions made by the two repetitions from a single observer. The WD% measure was also used to indicate the performance of a colour difference formula. An example is given in Fig. 4.4, showing two scatter diagrams which plot visual results in A% values against the DE values calculated by CIELAB(2 : 1) and CIEDE2000(2 : 1 : 1), respectively. (A lightness parameter, kL, of two was applied to both formulae.) Each diagram in Fig. 4.4 includes four quadrants (marked Q1 to Q4) divided by the 50 A% and colour tolerance (DEt) for x- and y-axes, respectively. A trend can be found in these diagrams, indicating a decrease of A% with an increase of DE. This is expected as the samples will be rejected when their colour differences increase. The wrong decision is calculated by the sum of data points in Q1 and Q3, for which the data in Q1 represent small visual differences but large DE values of a colour difference equation. The data in Q3 are also wrong decisions with large visual differences against small instrumental DE values. When calculating WD%, the DE value (x-axis) will systematically
Colour quality evaluation
67
vary from zero to a pre-defined large colour difference (say 10) with a fixed increment (say 0.1). For each small increment, a new WD% is calculated and stored. Finally, the minimum WD% value represents the tolerance value (DEt). The WD% values in Fig. 4.4a and 4.4b are 26% and 18%, respectively, indicating that CIEDE2000(2 : 1 : 1) formula out-performed CIELAB(2 : 1) formula by 8 WD% and should be recommended in this application. Note that the kL value of 2 gives a better performance than kL of 1 for each formula. Observer uncertainty
Percentage, %
As mentioned earlier, ten assessors participated in the experiment and each made their assessments twice. In total, there were 20 observations for each of the 38 pairs studied. Observer uncertainty, including accuracy and repeatability, were analysed for each observer. The results are summarised in Table 4.1 in terms of WD%.
100 90 80 70 60 50 40 30 20 10 0
Q2
Q1
Q3 0.3
Q4
DEt 0.4
0.5
0.6
CIELAB(2:1) DE
Percentage, %
(a)
100 90 80 70 60 50 40 30 20 10 0 0.2
Q2
Q1
Q4
Q3 DEt 0.3
0.4
0.5
0.6
0.7
0.8
0.9
CIEDE2000(2:1) DE (b)
4.4 Method for determining the wrong decision for (a) CIELAB(2 : 1) and (b) CIEDE2000(2 : 1 : 1) colour-difference equations.
68
Total colour management in textiles
Table 4.1 Summary of the observer uncertainty in WD% Assessor
1
2
3
4
5
6
7
8
9
10
Mean
Accuracy
29 21
26 18
29 26
21 18
34 45
32 40
29 21
13 24
37 40
32 16
27
Repeatability
29
32
16
24
37
29
29
16
32
24
28
Table 4.2 Performance (WD%) of formulae for Phase 1 results (38 pairs)
CIEDE2000
CIELAB
CIE94
CMC
kL = 1 Tolerance DEt WD%
0.45 21%
0.39 32%
0.38 32%
0.52 29%
kL = 2 Tolerance DEt WD%
0.45 18%
0.37 26%
0.36 26%
0.52 24%
Table 4.1 results show that the performances of observer accuracy and repeatability are very similar (about 28 WD%). It indicates that assessors could have around one wrong decision in three judgements. Testing different colour difference formulae The visual results obtained from two phases were used to test four colour difference formulae: CIEDE2000, CMC, CIE94 and CIELAB. Their performances are given in Table 4.2. For each formula, lighting parameters, kL, were set to 1 and 2, respectively. The results in Table 4.2 show that CIEDE2000 gave the best performance amongst those formulae studied. For all the formulae tested, a kL of 2 is more suitable than a kL of 1. Finally, CIEDE2000(2 : 1 : 1) with a colour tolerance of 0.45 was the best in this study.
4.3
Metamerism
Another application of colour difference formula is in predicting the degree of metamerism between sample pairs. Metamerism occurs when two samples match each other under one set of viewing conditions but fail to match under another. There are four types of metamerism: illuminant, observer, field size and geometric. Illuminant metamerism is the most important type of metamerism and occurs when two samples appear to match well under one illuminant, but exhibit a large mismatch under a second illuminant. Observer metamerism
Colour quality evaluation
69
Table 4.3 An example to illustrate the calculation of the metamerism index
DL
DC
DH
Reference illuminant (D65) Test illuminant (A) Difference
1.0 -1.0 -2.0
1.5 1.5 0.0
2.0 -2.0 -4.0
DEM = (22 + 02 + 42)1/2 = 4.5
occurs when a pair of samples matches for one observer, but fails to match when seen by a second. Field size metamerism arises from a satisfactory match being lost when the field size changes. Geometric metamerism occurs when a mismatch develops due to changes in the illumination and viewing geometry. The CIE International Lighting Vocabulary26 defines metamerism as a property of a pair of spectrally different colours having the same tristimulus values under a set of viewing conditions. In practice, it is impossible to achieve precisely identical tristimulus values. However, it is possible to apply some corrections to make the pair under consideration exactly match in the reference illuminant. In Table 4.3, an example is given of quantifying illuminant metamerism by applying an additive correction. The DL, DC and DH are calculated from a colour difference formula (see Section 4.2) under a reference illuminant (e.g. illuminant D65). There are subtracted from the corresponding DL, DC and DH under a test illuminant (say illuminant A). Finally, the metamerism index, DEm, is calculated using (eqn 4.7). 2
2
∆Em = ( ∆L1 − ∆L2 ) + ( ∆C1 − ∆C2 ) + ( ∆H1 − ∆H 2 )
2
[4.7]
where the subscripts 1 and 2 represent the test and reference illuminant, respectively. In the case of lightness difference, the sample is judged lighter than standard by one unit under illuminant D65, but darker by the same amount under illuminant A. Thus, the corrected lightness difference between the two illuminants is two units. The same correction is applied to chroma and hue differences. The resultant metamerism index is therefore 4.5.
4.4
Colour constancy
Another important application of the colour difference formula is in evaluating the degree of colour constancy under different illuminants.
70
Total colour management in textiles
4.4.1 Concept of colour constancy For the majority of industrial colour matching, the aim is to produce samples possessing a spectral reflectance as close as possible to that of the standard. If the reflectances are the same, the match will hold for all illuminants and observers. This type of reproduction is known as a spectral, or non-metameric, match. In many cases it is not possible to obtain a spectral match using the desired set of colourants. If the reflectances are very different, the match might be very good for one illuminant, say D65, and the match may not hold for other illuminants such as illuminant A or F11, a tri-narrow band illuminant. This type of match is known as metameric match. The degree of metamerism may be quite small or very marked, depending on the difference between the two reflectance functions in question and on the specific illuminants concerned. A metamerism index was introduced earlier in (eqn 4.7). Even if the match is only slightly metameric, however, the sample produced may well look a completely different colour under certain other illuminants, depending on the reflectance property of the reference sample. Such samples are said to be non-colour constant. As with metamerism, there may be varying degrees of colour inconstancy, i.e. the severity of colour change to be expected when moving from one illuminant to another. For real samples, visual estimates can be made by observing the sample under different light sources. Using a recipe formulation system, it is possible to calculate alternative recipes using many different sets of colourants. Using a colour inconstancy index, it is possible to determine which recipe gives the most colour constant product. R%
XYZ
Xr Yr Zr
CAT02
Xc Yc Zc
CDE
DE DL DC DH CMCCON02
4.5 The procedure to calculate the CMCCON02 colour inconstancy index.
Colour quality evaluation
71
Note that, while this problem occurs relatively infrequently (spectral matches to a standard colour are usually requested), it is most important that attention should be paid to inconstancy when it does occur. If the original standard is colour constant, all subsequent spectral matches will also be colour constant. If not, a whole series of non-colour constant, and potentially non-metameric, products will be produced.
4.4.2 The structure of a colour inconstancy index A colour inconstancy index is capable of predicting the magnitude and direction of the change in colour appearance between a sample viewed under a test illuminant (say Illuminant A) and the same sample viewed under a reference illuminant (say D65). A colour inconstancy index, named CMCCON02,27 was recommended by the Colour Measurement Committee (CMC) of the Society of Dyers and Colourists (SDC). The procedure for calculating the CMCCON02 is given in Fig. 4.5. A spectral reflectance function is first obtained by measuring a test specimen with a spectrophotometer. The tristimulus values, X, Y, Z and Xr, Yr, Zr under illuminants A and D65, respectively, are then calculated in the usual way. The CAT02 chromatic adaptation transform is employed to predict the corresponding colours, Xc, Yc, Zc, under illuminant D65 from the X, Y, Z values of the sample under Illuminant A. Finally, a suitable colour difference equation (CDE) is used to calculate a DE value, together with individual colour difference components (DL, DC and DH) between Xc, Yc, Zc and Xr, Yr, Zr. The magnitude of the DE value indicates the degree of colour inconstancy. The direction of the colour change can be expressed by the individual colour difference components. A DE value of zero indicates complete colour constancy for the specimen being tested.
4.4.3 The CAT02 chromatic adaptation transform (CAT02) The key element of CMCCON02 is the CAT02 chromatic adaptation transform. The computational procedure for CAT02 is given below: Starting data: Sample in test illuminant: X, Y, Z Adopted white in test illuminant: Xw, Yw, Zw Reference white in reference illuminant: Xwr, Ywr, Zwr Luminance of test adapting field (cd/m2): LA Transformed data to be obtained: Sample corresponding colour in reference illuminant: Xc, Yc, Zc
72
Total colour management in textiles
Step 1. For the sample, calculate: Rw X w Rwr X wr R X Gw = MCAT02 Yw , Gwr = MCAT02 Ywr , G = MCAT02 Y . Z Bw Zw Bwr Zwr B where
MCAT02
0.7328 0.4296 −0.1624 = −0.7036 1.6975 0.0061 0.0030 0.0136 0.9834
Step 2. Calculate the degree of adaptation, D:
( )
1 D = F 1 − e 3.6
− La − 42 92
where F equals 1, 0.9, and 0.8 for average-, dim- and dark-surround conditions, respectively, and where LA is the luminance of the test adapting field. If D is greater than one or less than zero, truncate it to one or zero, respectively. For calculating the CMCCON02 index for a physical sample such as textiles, it is recommended that D is set to unity, i.e. it corresponds to a complete adaptation or completely discounts the illuminant. Step3. Calculate Rc, Gc, Bc from R, G, B (similarly Rwc, Gwc, Bwc from Rw, Gw, Bw): Rc = R[D(Rwr /Rw) + 1 - D]
Gc = G[D(Gwr /Gw) + 1 - D] Bc = B[D(Bwr /Bw) + 1 - D]
Step 4. Calculate for the reference illuminant the corresponding tristimulus values for the sample, Xc, Yc, Zc: Xc Rc −1 Yc = MCAT02 Gc , Zc Bc where
Colour quality evaluation
–1 CAT02
M
73
1.096124 −0.278869 0.182745 = 0.454369 0.473533 0.072098 −0.009628 −0.005698 1.015326
Calculation in both the forward and especially the reverse mode are extremely sensitive to rounding off errors, making it essential to employ the full precision implied in MCAT02 and M-1CAT02. Some older computer-based colour measurement systems may, therefore, need to employ double precision arithmetic.
4.5
Conclusions and future trends
This chapter focused on the field of colour difference applications. Firstly, the author reviewed the development of colour difference formulae. A method for setting tolerance using colour difference formulae for industrial applications was also introduced. Finally, methods for predicting metamerism and colour constancy were described. It can be concluded that after more than three decades of development of colour difference equations, a robust colour difference formula, CIEDE2000, has been achieved. However, colour difference research is still on-going. Amongst the problems yet to be tackled are: • Almost all of the recent effort has been spent on modifications to CIELAB. This has resulted in CIEDE2000, which includes five corrections of CIELAB to fit the available experimental data sets. It is highly desirable to derive a formula based upon a new perceptually uniform colour space from a particular colour vision theory. A uniform colour space based upon a colour appearance model, like the CIE colour appearance model CIECAM02,28 could be an ideal solution. • All colour difference formulae can only be applied to a limited set of reference viewing conditions, such as those defined by the CIE.20 It would be extremely useful to derive a parametric colour difference formula capable of taking into account different viewing parameters such as illuminant, size of sample, colour difference magnitudes, separation, background and luminance level.29,30 • Almost all of the colour difference formulae were developed only to evaluate colour difference between pairs of large single objects/patches. More and more applications require predicting colour differences between pairs of pictorial images. The current formula does not include the necessary components to consider spatial variations for evaluating such images. There is, therefore, an urgent need to develop a formula for this purpose.31,32
74
Total colour management in textiles
4.6
References
1. Luo, M.R., Cui, G.H. and Rigg, B. (2001). The development of the CIE 2000 colour difference formula. Color Res. Appl., 26, 340–350. 2. CIE (2001). Technical report: improvement to industrial colour-difference evaluation. CIE Publ. No. 142. Vienna: Central Bureau of the CIE. 3. MacAdam, D.L. (1942). Visual sensitivities to color differences in daylight. J. Opt. Soc. Am., 32, 247–274. 4. Newhall, S.M., Nickerson, D. and Judd, D.B. (1943). Final report of the O.S.A. subcommittee on spacing of the Munsell colors. J. Opt. Soc. Am., 33, 385– 418. 5. CIE (2004). Colorimetry. CIE Publ. No. 15:2004, Central Bureau of the CIE, Paris, 1986. 6. Chickering, K.D. (1967). Optimization of the MacAdam-modified 1965 Friele color-difference formula. J. Opt. Soc. Am., 57, 537–541. 7. Adams, E.Q. (1942). X-Z planes in the 1931 ICI system of colorimetry. J. Opt. Soc. Am., 32, 168–173. 8. Hunter, R.S. (1958). Photoelectric color difference meter. J. Opt. Soc. Am., 48, 597–605. 9. McDonald, R. (1980). Industrial pass/fail colour matching. Part I – Preparation of visual colour-matching data. J. Soc. Dyers Col., 96, 372–376. 10. Luo, M.R. and Rigg, B. (1986). Chromaticity-discrimination ellipses for surface colours. Color Res. Appl., 11, 25–42. 11. Luo, M.R. and Rigg, B. (1987). BFD(l : c) colour difference formula, Part I – Development of the formula. J. Soc. Dyers Col., 103, 86–94. 12. Alman, D.H., Berns, R.S., Snyder, G.D. and Larsen, W.A. (1989). Performance testing of color-difference metrics using a color tolerance dataset. Col. Res. Appl., 14, 139–151. 13. Berns, R.S., Alman, D.H., Reniff, L., Snyder, G.D. and Balonon-Rosen, M.R. (1991). Visual determination of suprathreshold color-difference tolerances using probit analysis. Col. Res. Appl., 16, 297–316. 14. Kim, H. and Nobbs, J.H. (1997). New weighting functions for the weighted CIELAB colour difference formula. Proc. Colour 97 Kyoto, Vol. 1, 446–449. 15. Witt, K. (1999). Geometric relations between scales of small colour differences. Color Res. Appl., 24, 78–92. 16. Chou, W., Lin, H., Luo, M.R., Rigg, B., Westland, S. and Nobbs, J. (2001). The performance of lightness difference formula. Col. Technol., 11, 19–29. 17. Cui, G.H., Luo, M.R., Rigg, B. and Li, W. (2001). Colour-difference evaluation using CRT colours. Part I: Data gathering. Col. Res. Appl., 26, 394–402. 18. Clarke, F.J.J., McDonald, R. and Rigg, B. (1984). Modification to the JPC79 colour-difference formula. J. Soc. Dyers Col., 100, 128–132 and 281–282. 19. Luo, M.R. and Rigg, B. (1987). BFD(l : c) colour difference formula, Part II – Performance of the formula. J. Soc. Dyers Col., 103, 126–132. 20. CIE (1995). Industrial colour-difference evaluation, CIE Publ. 116, Central Bureau of the CIE, Vienna, Austria. 21. McDonald, R. (1980). Industrial pass/fail colour matching. Part III – Development of a pass/fail formula for use with instrumental measurement of colour difference J. Soc. Dyers Col., 96, 486–496.
Colour quality evaluation
75
22. ISO 105-J03 Textiles: Test for colour fastness. Part 3 Calculation of colour differences. Geneva: ISO. 23. Luo, M.R. (1999). Colour science: past, present and future. In Colour Imaging: Vision and Technology, ed. by L.W. MacDonald and M.R. Luo, John Wiley & Sons Ltd, 383–404. 24. Melgosa, M., Huertas, R. and Berns, R.S. (2004). Relative significance of the terms in the CIEDE2000 and CIE94 colour-difference formulas. J. Opt. Soc. Am., A, 21, 2269–2275. 25. McLaren, K. (1970). Colour passing – visual or instrumental? J. Soc. Dyers Col., 86, 389–392. 26. CIE (1987). CIE International Lighting Vocabulary, CIE Publ. No. 17.4. Vienna, Austria: Central Bureau of the CIE. 27. Luo, M.R., Li, C.J., Hunt, R.W.G., Rigg, B. and Smith, K.J. (2003). The CMC 2002 Colour Inconstancy Index: CMCCON02, Coloration Technol., 119, 280–285. 28. CIE (2004). CIE TC8-01 Technical report. A colour appearance model for colour management systems: CIECAM02, CIE Publ. 159. 29. CIE (1993). Technical report. Parametric effects in colour-difference evaluation. CIE Publ. No. 101. Vienna, Austria: Central Bureau of the CIE. 30. Guan, S.S. and Luo, M.R. (1999). Investigation of parametric effects using large colour differences. Color Res. Appl., 24, 356–368. 31. Zhang, X.M. and Wandell, B.A. (1996). A spatial extension of CIELAB for digital color image reproduction, Proc. 4th IS&T/SID Color Imaging Conference. 32. Hong, G. and Luo, M.R. (2001). Perceptually based colour difference for complex images, Proc. 9th Session of the Association Internationale de la Couleur (AIC Color 2001), June, Rochester, USA 618–621.
5 A practical guide to visual evaluation of textile samples K B U T T S, Datacolor, USA
5.1
Introduction
By most definitions, colour is a sensation produced by certain effects of light interactions with an object, either emitted or reflected. In the commercial sense, this sensation of colour is capable of imparting significant aesthetic value to otherwise non-coloured materials and objects. For the vast majority of consumers, colour plays a very important role in their decision-making process when purchasing goods such as clothes, home furnishings, cars and many others. The art and science of the colouration industry is to add sufficient value to the item through colour such that it will be more desirable to the consumer as something newer, more stylish, complementary and more attractive than an otherwise similar product. Since colour is a sensation in the mind of an individual, it follows that the most straightforward method of assessing the quality of one colour vs. another is simply to look at it. While one cannot ignore the validity of colour measurement instruments, one must also be reminded that instruments do not measure colour but rather factors directly relating to colour. In cases where visual and instrumental results are in disagreement, the visual assessment is the final arbiter of acceptability of a sample. We draw two conclusions from this – firstly, that as there will always be a need for visual colour assessment, when visual processes are used for colour development and quality control purposes, they must be standardised as much as possible so as to minimise the subjectivity that exists in individuals, and secondly, that when utilised, it should be the goal of instrumental analysis to emulate the visual process as closely as possible. In order for visual colour evaluation to be an effective tool in the colour development and colour quality control processes, certain parameters must be well understood and controlled. This chapter will focus, therefore, on the 76
A practical guide to visual evaluation of textile samples
77
various aspects of the visual colour evaluation process and will provide practical guidelines for ensuring consistency among observers.
5.2
The components of colour perception
The physiology of human vision is a very complex process and as such lies beyond the scope of this discussion. To understand how colour is visually assessed we must, however, first realise that colour perception is a function of both the optical and neurological systems in the observer and that, by working together, they produce the sensation in the brain we call, for example, the colour pale purple. In other words, two observers who have perfectly matched optical systems, or eyes, and who see the exact same object will not necessarily have the same sensation of colour. The visual input is processed or analysed by the brain and it is in this area that the sensation of colour is realised. In Josef Albers’ famous work The Interaction of Color he writes ‘In visual perception a color is almost never seen as it really is – as it physically is. In order to use color effectively it is necessary to recognize that color deceives continually’ (Albers, 1975). The sources of this deception can be categorised into the effects of the source of light, the object itself and the observer of the sensation of colour, each of which must be properly controlled in order to ensure effective and repeatable visual assessment of colour.
5.2.1 The illuminant, or light source The first component required for the sensation of colour is a source of light. Two terms are often used interchangeably when discussing the source of light for colour evaluation, and those terms are illuminant and light source. A ‘standard illuminant’ is defined by the American Society for Testing and Materials (ASTM International) in method E284 – ‘Standard Terminology of Appearance’ as ‘a luminous flux, speci– fied by its spectral distribution, meeting specifications adopted by a standardizing organization’ (ASTM International, 2002). In simpler terms, a standard illuminant is light energy quantified and listed in a table as a set of spectral power distribution values, though it may not be exactly represented by a light bulb that produces light. A light source is defined in the same ASTM method as ‘an object that produces light or other radiant flux, or the spectral power distribution of that light’ (ASTM International, 2002). A source may therefore be defined as a physical light bulb, manufactured with the goal of producing light energy corresponding to a particular set of standard illuminant spectral power distribution data. Based on these definitions, visual colour evaluation will focus on the effects of a light source on the colour evaluation process, while illuminant data
78
Total colour management in textiles
has application in calculation of numerical colour values and colour differences. Light source selection A variety of light sources are available to the retail and clothes market, and a number of reasons are identified for selection of a particular light source or sources that will be used in the retail store environment. The reasons include, but are not limited to, the design and layout of the store, the mood generated by the light sources and their arrangement, the energy requirements of the light sources when they are used in large quantities, the availability of external lighting through skylights or windows and doors, the costs associated with equipping a large number of stores and so on. It is unfortunate, however, that little thought may be given to the effect of the selected light source on the consumer’s perception of the colour of the goods being marketed. It is not uncommon for the colour design and development process to be performed in one light source while the end-product is displayed in a different light source or combination of light sources. The net effect in this scenario is that the colour perceived by the consumer is not the colour intended by the developer. A number of standard illuminants have been defined by ASTM International in method E308 – ‘Practice for Computing the Colors of Objects by Using the CIE System’, and the goal of light source manufac turers is to produce bulbs that closely simulate the spectral power distribution curves of these standard illuminants. For the textile industry, the most commonly used light sources include daylight – designated as standard illuminant D65 and rendered in light source form as ‘D65’; fluorescent – designated as a number of standard illuminants including F2, F7, F11, and others, and rendered in light source form as CWF, TL83, TL84, and others; incandescent – designated as standard illuminant A and rendered in light source form as ‘A’; and a number of others. Additional light sources are available for which the manufacturer provides spectral power distribution data, and include light sources such as Horizon and Sylvania FO35. The goal of the colour developer should be to select the light source or light sources for use in initial colour development and on-going colour quality control that most closely duplicate the light sources used in the retail store. By doing so, there will be some degree of assurance that the colour that is developed will be rendered as intended on the showroom floor. The light source that is selected is then communicated as the primary light source throughout the supply chain. In addition to the primary light source, a secondary light source is also specified. Specification of a secondary light source will allow for the evaluation of the colour characteristics of flare and metamerism, which are
A practical guide to visual evaluation of textile samples
79
qualitative assessments of the degree of colour change that is observed when one or more samples are visually evaluated in multiple light sources. Flare is a term used to describe a condition in which a single sample exhibits an observable colour change when it is visually evaluated in multiple light sources. Metamerism is a term used to describe a condition in which the direction or magnitude of colour difference exhibited between a pair of samples visually evaluated in one light source changes in either magnitude or direction when the pair of samples is viewed in a secondary light source. Given that elimination of either of these effects is virtually impossible, clear communication of the expected limits of flare and metamerism is essential. Light source evaluation and quality The suitability of a particular light source for visual colour evaluation is determined not only by how closely the light source matches the light source used in the retail environment, but also by the ability of the light source to reliably simulate the intended standard illuminant. Manufacturers of light cabinets and similar lighting units evaluate the spectral power distributions of the installed light sources against the spectral power distributions of the standard illuminants as well as determine the colour rendering indexes of the light sources. CIE ratings are also calculated to assess the quality of daylight simulators. These assessments give an indication of the suitability of the light sources for a particular application and should be considered when comparing and purchasing a light cabinet or lighting device. Manufacturing techniques in combination with light cabinet design and filtering serve to increase the correlation between the light sources and the standard illuminants. It must be noted, however, that design and manufacturing methods vary between light cabinet manufacturers, which introduces the potential for variability in the perceived colour when light cabinets of various manufacturers are used. This same variability may exist even in light cabinets supplied by the same manufacturer and is related to cabinet design, light source specification, cabinet maintenance and cabinet condition. Light cabinets must be regularly maintained in order to ensure long-term repeatability of visual colour evaluation results as well as reproducibility of results among multiple cabinets. Replacement of bulbs according to manufacturer specifications is a key element of light cabinet maintenance. Use of non-conforming light sources, whether in a commercial light cabinet or in a light fixture, for the purpose of visual colour evaluation will result in inconsistencies in visual colour difference evaluation when compared to evaluations performed using approved bulbs.
80
Total colour management in textiles
5.2.2 The object The second component required for the sensation of colour is the object being observed. While this may be considered understood, the nature of the object plays an important role in determining the colour perceived in the mind of the observer and it cannot be overlooked. Visual evaluation of textile samples can most accurately be described as an evaluation of the appearance of the sample rather than an evaluation of the colour of the sample. The appearance of an object may be defined as the observer’s response to the combination of both the colour and the physical charac teristics of the sample. Physical aspects of textile samples and the effect on visual evaluation The processes of weaving, knitting, preparation, dyeing and finishing, among others, impart physical characteristics to a textile sample as well as colour. Common examples of physical characteristics include non-uniform fabric construction in the form of twill and corduroy or a glossy surface finish that results from a mechanical calendering process. These physical characteristics contribute to the perceived appearance of a sample by altering the way in which the sample interacts with the light that is incident upon the sample, which in turn affects the reflected light that interacts with the observer. In most cases the contribution of physical characteristics to overall appearance can only be quantified by visual evaluation and not by instrumental evaluation. Consistency in form and function Visual sample evaluation requires that a sample of the desired colour be available to each party involved in the colour development and quality control processes. In most cases an original concept sample is selected and then duplicated for distribution to others involved in colour creation. This is typically accomplished by development of small quantities of standard material by organisations specialising in this service. These working standards are then distributed for visual as well as for instrumental colour analysis. The physical characteristics of these working standards will be almost identical, but care must be taken to ensure that the colour variability among these working standards is practically non-existent. The previously defined characteristics of flare and metamerism must also be controlled for the working standards as they may be referenced for colour development for a variety of materials and dye classes. The integrity of the standard must be controlled as, over time, its perceived colour may change due to handling, storage and fading.
A practical guide to visual evaluation of textile samples
81
Specific guidelines should be either referenced or developed with regard to the placement and positioning of samples in a light cabinet for visual evaluation so as to minimise inconsistencies in visual evaluation that are due solely to the orientation of the samples. An often overlooked factor that must be considered when performing visual colour evaluation is the temperature and moisture level of the sample. Depending upon fabric construction and dye class, the temperature and relative humidity to which the sample has been conditioned may significantly affect the interaction of the sample with incident light, resulting in noticeable colour differences. Specific guidelines for textile sample conditioning are given in ASTM International method D1776 – ‘Standard Practice for Conditioning and Testing Textiles’. This method specifies that samples should be conditioned to 21 ± 1°C and relative humidity of 65 ± 2% prior to sample testing.
5.2.3 The observer The third component required for the sensation of colour is an observer. By nature, the observer is the most difficult component of the colour sensation process to characterise, and the most subjective. While substantial research has resulted in the creation of ‘standard observers’ for use in numerical colorimetric calculations, the average observer may or may not fit the model of the ‘standard observer’. Observers are affected by a number of internal and external factors, each of which can contribute to inconsistencies in observations among observers as well as inconsistency in the repeatability of the observer’s own visual evaluation results. Colour vision testing As evidenced by the process used to develop the standard observer, it is clear that the inherent capacity to visualise colour varies among observers. It is also documented that ‘about 8% of the male and 0.5% of the female population have color-defective vision’ (Berns, 2000). For these reasons it is vital that everyone involved in the visual evaluation of colour undergoes certain tests to determine if a colour vision deficiency exists or if there are issues with discrimination of small differences between similar colours. Poor results in either type of test may indicate that the individual is not suited for visual colour evaluation. Note that, as an observer’s colour vision may change over time, tests of this nature should be periodically repeated. Two well-known tests for colour deficiencies are the Isihara Tests for Color Blindness and the Dvorine Pseudo-Isochromatic Plates. Both tests will identify the presence of a colour deficiency on the part of the observer. Tests for colour discrimination – the ability to recognise and distinguish
82
Total colour management in textiles
between samples with small colour differences – include the Farnsworth– Munsell 100 Hue Test and the HVC Color Vision Skill Test. Specific guidelines for evaluation of observers are defined in ASTM International method E1499 – ‘Standard Guide for Selection, Evaluation, and Training of Observers’. Factors affecting colour perception In addition to the genetic characteristics of the observer’s optic system that result in the previously mentioned rates of colour deficiencies, a number of other physical and psychological factors influence their ability to discern and discriminate among colours. Internal factors affecting colour vision include fatigue, desensitisation due to prolonged exposure to certain colours, emotional state and the influence of chemical substances that affect the mental and physical condition of the individual. Personal bias may also exist in an individual based on their past experience, cultural influences and personal preferences with regard to style and colour. External factors affecting colour vision include both positive and negative influences. Experience and training can improve the observer’s ability to discern and describe small colour differences between similar samples. Failure to adhere to proper methods for visual evaluation, such as improper sample positioning, failure to control the surrounding area or extraneous materials in the viewing space and external lighting influences, will result in inconsistencies in visual colour evaluation. The physical characteristics of the sample being evaluated contribute significantly to the final acceptability of a sample. These effects are either consciously or subconsciously included in the overall evaluation of the sample.
5.2.4 Sample viewing environment The source of light, the object and the observer are joined together to produce a sample viewing environment. It is not enough just to define and understand the contribution of these three components to the visual colour evaluation process because it is only when they are combined together in the viewing environment that evaluation of colour is possible. As such, the interaction of these three components and the environment in which they are placed must also be defined. Guidelines for specifying and controlling the sample viewing environment as documented in various industry methods and procedures will be examined in more detail in a later chapter. The light source environment As previously mentioned, suitable primary and secondary light sources should be selected for evaluation of sample colour, and these light sources
A practical guide to visual evaluation of textile samples
83
should be incorporated into an acceptable lighting cabinet or lighting device. The surrounding environment including the colour of the walls, the pre sence of extraneous material or other coloured samples, the clothing being worn by the observer, ambient light, etc., will determine whether or not the desired level and quality of spectral energy emitted by the light source is actually incident upon the sample. Specific guidelines must be followed with respect to the colour of the surrounding area, the intensity of the light at the level of the sample and control of external influences, including ambient or direct light from other sources such as windows and lamps. Visual colour evaluations performed using uncontrolled light sources such as light through an office window, a desk lamp, standard overhead lighting or natural outdoor light cannot be considered reliable. The object environment In addition to the criteria previously mentioned for the objects being evaluated, the preparation and positioning of the objects must be addressed. To ensure consistency and repeatability in sample evaluation, samples must be prepared and positioned in the same way each time the samples are evaluated. Sample preparation will include any mechanical or manual cleaning of the surface, pile orientation for applicable fabrics and sample conditioning as previously described. Sample positioning will include specification of the number of layers of material to use, the type and colour of any background that might be present, the relative positioning of the samples to each other, the orientation of the fabric construction, the orientation of individual yarns or fibres for materials being evaluated in yarn or fibre form and designation of the minimum sample size required for evaluation. The observer environment After confirming that all observers are qualified to perform visual colour evaluations as described previously, the observers must follow specific guidelines during the visual evaluation process. A number of standard methods exist for the purpose of defining the criteria to be followed by the observer when visually evaluating samples and are addressed in other chapters. Of primary importance are the positioning of the observer relative to the sample and proper techniques for changing light sources for sample evaluation.
5.3
Industrial guidelines for visual colour assessment
A number of professional organisations including the American Association of Textile Chemists and Colorists (AATCC) and ASTM International have
84
Total colour management in textiles
developed methods to assist the colour industry with the visual evaluation process. These tests include but are not limited to: • AATCC Evaluation Procedure 9: Visual Assessment of Color Difference of Textiles • ASTM D1729: Practice for Visual Appraisal of Colors and Color Differences of Diffusely-Illuminate Opaque Materials • ASTM D4086: Practice for Visual Evaluation of Metamerism • ASTM E308: Practice for Computing the Colors of Objects by Using the CIE System • ASTM E1499: Guide for Selection, Evaluation, and Training of Observers Additional methods have been developed for specific application of visual techniques for the assessment of sample characteristics such as colour change, staining and gloss and include: • AATCC Evaluation Procedure 1: Gray Scale for Color Change • AATCC Evaluation Procedure 2: Gray Scale for Staining • AATCC Evaluation Procedure 3:AATCC 5-Step Chromatic Transference Scale • ASTM D2616: Test Method for Evaluation of Visual Color Difference With a Gray Scale • ASTM D3134: Practice for Establishing Color and Gloss Tolerances These and other methods are published yearly as the Technical Manual of the American Association of Textile Chemists and Colorists (AATCC, 2004) and the Annual Book of ASTM Standards (ASTM International, 2002) and are available from the applicable organisation. The observer should reference each of these test methods in order to become familiar with the specific details of each method. Adherence to these methods will greatly increase the repeatability and reproducibility of visual colour evaluations.
5.3.1 AATCC Evaluation Procedure 9: Visual Assessment of Color Difference of Textiles The purpose of this evaluation procedure is to provide the user with guidelines to be used for developing a repeatable technique for visual assessment of colour and colour difference. The primary focus is on lighting conditions, sample preparation and determination of visual colour difference. Lighting conditions AATCC Evaluation Procedure 9 (AATCC EP9) provides the user with specific information regarding acceptable illuminants and their colour
A practical guide to visual evaluation of textile samples
85
temperature, which may be measured with a colour temperature meter or photometer, and the illumination level at the sample plane. Illumination level may vary from one lighting environment to another, which may affect the perceived colour, depending upon the degree of variation, and may be measured using a spectroradiometer. Also described are the recommended neutral gray specifications for the surrounding area, the arrangement of the light source relative to the sample and the observer – better known as viewing geometry, and the manner in which the viewing area should be maintained. Sample preparation AATCC EP9 provides the user with guidelines for preparing samples for visual evaluation, including sample conditioning and sample size as appropriate for the type of material being evaluated, such as woven and knitted cloths, narrow cloths, loose fibre, threads and yarns. As with any recommendation for sample presentation, the user may find that their particular presentation method may require some alteration of the method presented in this procedure if deemed necessary for accurate sample evaluation. The presentation method selected must be clearly communicated to all obser vers who will be evaluating the material in question in order to avoid misinterpretation. Determination of visual colour difference Visual colour differences between samples are determined by the observer, and AATCC EP9 defines several descriptive terms that can be used to verbalise the visual evaluation. The primary descriptors used should communicate a direction of colour difference and a magnitude of colour difference between the sample designated as the standard and another sample designated as the test specimen. As previously stated, the results of the visual evaluation will be dependent upon a number of internal and external factors that influence the observer. The goal is that all parties involved in visual evaluation of a particular sample pair will perceive approximately the same colour difference direction and magnitude.
5.3.2 ASTM D1729: Practice for Visual Appraisal of Colors and Color Differences of Diffusely-Illuminate Opaque Materials ASTM D1729 was developed to assist the observer in defining the equipment and procedures required to produce repeatable visual evaluation results. ASTM D1729 addresses the same requirements of sample illumina-
86
Total colour management in textiles
tion and geometry, sample preparation and evaluation of colour difference as previously defined in AATCC EP9.
5.3.3 ASTM D4086: Practice for Visual Evaluation of Metamerism ASTM D4086 offers a method for detecting the presence of both illuminant metamerism and observer metamerism. Illuminant metamerism has been previously defined, and observer metamerism may be defined as the circumstance in which a pair of samples are a visual match to one observer in one set of conditions but are not a visual match to another observer in the same conditions. Note that calculations for metamerism are not included in this method, though they are available in computer programs that calculate colorimetric differences. Instead, ASTM D4086 provides the user with recommendations regarding use of a grey scale to determine the degree of metamerism.
5.3.4 ASTM E308: Practice for Computing the Colors of Objects by Using the CIE System ASTM E308 provides details for calculation of colour coordinates based on the CIE system. Also included in this method are spectral tristimulus values for CIE 1931 Standard (2°) Observer and CIE 1964 Supplementary Standard (10°) Observer, relative spectral power distribution data for a range of CIE standard illuminants and tristimulus weighting factors for the same range of CIE standard illuminants. These data tables are incorporated into computer programs that calculate colorimetric coordinates and colour differences.
5.3.5 ASTM E1499: Guide for Selection, Evaluation, and Training of Observers ASTM E1499 details the criteria and tests that should be used for determining the suitability of an individual for visual evaluation of colour and colour difference. Also included in this method are techniques for training obser vers in order to improve their colour discrimination abilities. Colour vision tests and training tools such as the Dvorine Pseudo-Isochromatic Plates and the Farnsworth–Munsell 100 Hue Test are described in detail.
5.4
Practical application of visual colour assessment methods
After gaining a thorough understanding of the factors affecting visual colour evaluation, the observer must actively apply this understanding to
A practical guide to visual evaluation of textile samples
87
every observing situation. While each situation and circumstance may be different, the fundamental principles for visual colour evaluation remain the same. By clearly communicating the parameters within which samples are being visually evaluated, the observers are free to focus on the primary goals of the evaluation process, which may vary.
5.4.1 Colour vs. appearance As previously defined, colour is only one of the characteristics of a sample that may be evaluated. The overall visual effect produced by the colour of a sample and its physical characteristics is defined as the sample’s appearance. For the average observer, separation of ‘colour’ from ‘appearance’ may be an impossible task. With experience, the observer may learn techniques of sample positioning and illumination that allow for some degree of isolation of the colour characteristic from the overall appearance of a sample. It is left to the observer then to determine which of these aspects are of primary importance, either the sample’s colour or its overall appearance.
5.4.2 Initial colour development The initial colour development phase begins with selection of a standard. As previously discussed, the availability of a sufficient quantity of standard material in the desired colour for use by each observer in the colour development process can be a significant challenge. Assuming that an acceptable physical sample is available for each observer, it is then necessary for the person or organisation with final responsibility for approving the developed colours to specify the viewing parameters. These parameters are: • • • • • • • •
primary and secondary light sources viewing geometry physical form of the test sample (woven, knit, yarn, etc.) size of the test sample conditioning (temperature and humidity) requirements orientation of pile or other physical modifications number of layers for each sample or backing material orientation of the standard and test sample relative to each other.
In addition to the above items, clear guidelines must be communicated regarding the expectations for the quality of the colour match. Because of the nature of visual colour evaluation, this is a very subjective aspect of initial colour development, as it relies upon the experience of the colour developer – the ‘dyer’ or ‘supplier’ in the case of textile colour development – and their familiarity with the expectations of the ‘customer’. Specification of colorimetric tolerances generated using a spectrophotometer and appro-
88
Total colour management in textiles
priate computer software is an invaluable aid in this process as it gives the colour developer specific numerical data to use in their evaluation of the quality of the sample that has been produced. Consideration must also be given to the type of material on which the new colour is being developed and how it compares with the material used to prepare the target standard. Differences in the fibre type, construction, or dye chemistry may result in differences in total appearance in one or both of the primary or secondary light sources. For example, the supplier may not be able to match the brightness of a glossy paint chip while developing the colour on cotton yarn. In this scenario the customer must recognise the limitations of colour development on the selected material and adjust expectations accordingly. This is especially true when a colour is being developed on a variety of dissimilar materials, in which case coordination and harmony among the final materials is more important than an absolute match to the original standard.
5.4.3 Production quality control The primary concerns in production quality control are deviation from the approved target standard and consistency within and between production lots. The target standard for visual evaluation in the production environment will be either the original standard used for initial colour development, the sample approved by the customer, or the first production lot produced by the supplier. There must be agreement between customer and supplier as to which standard is used in order to eliminate any misunderstanding regarding colour quality. This is especially true when production lots from multiple sources will be combined into the same garment or when they will be displayed together in the retail store. A key to visual colour evaluation in the production environment is understanding the difference between perceptible colour difference and acceptable colour difference. The goal of the initial colour development phase is to produce a colour on the specified material that is identical in colour to the original standard. While this is theoretically possible – giving consideration to the points previously mentioned regarding dissimilar materials – it is typically impractical and requires an excessive amount of time and expense. Both parties agree then on a level of colour difference that still meets the needs of the customer and, while the test sample may be almost identical in colour, there is still a perceptible colour difference. Perceptibility, therefore, is defined as a level of colour difference that is almost non-existent. This level of colour difference is not practical in the production environment, so the customer and supplier must agree on a level of colour difference that, while potentially too excessive for initial colour development, is acceptable for bulk production. Note that the end-use of
A practical guide to visual evaluation of textile samples
89
the product should be considered when specifying the degree of colour difference accepted in production. A final consideration for production quality control is shade sorting and sequencing. While production material may be acceptable to the target standard, there may be excessive variability when comparing individual production units to each other. The degree of visible variance can be minimised by organising production units – lots, rolls, batches, etc. – in such a way as to minimise the visible colour change between the individual units. Shade sorting and sequencing (or tapering) involves the visual arrangement of production units such that, even though there may be considerable variation between the colorimetric extremes in the set of production units, the sorted and sequenced group will appear uniform. The manufacture of garments from production units delivered in their sorted and sequenced order will minimise the possibility of noticeable colour variation within the set of manufactured garments or within the individual garments.
5.4.4 Sample testing Product characteristics related to appearance such as washfastness and lightfastness are evaluated based on visual assessment methods. Specific guidelines have been developed for testing of these characteristics and are detailed in AATCC and ASTM International methods, as previously mentioned, as well as in test methods developed by the International Organization for Standardization (ISO). A number of other visual characteristics may be of importance, such as gloss, metamerism, resistance to abrasion, resistance to colour change from abrasion, etc., and appropriate test methods should be used if available.
5.4.5 Application to various textile materials Visual evaluation methods for textiles will vary depending upon the physical characteristics of the material. For those samples with significant physical variability it is critical that the visual evaluation methods be well defined and clearly communicated. The following guidelines regarding sample positioning and preparation should be considered when evaluating various textile materials. For all materials, the accuracy and repeatability of the visual evaluation will diminish as the size of the sample decreases. All materials should be clean and free of visible contamination before they are evaluated. Flat wovens and knits These types of materials are often the least difficult to evaluate as they are uniform in construction. The material should be orientated with the face of
90
Total colour management in textiles
the fabric up and the horizontal/vertical position aligned according to the manufactured direction of the fabric. Twill and knit patterns should be observed if present and the sample orientated accordingly. Sample opacity should be considered, and either a suitable number of layers to achieve opacity be specified or a common backing material be used. Pile goods such as velour, corduroy and carpet Of primary concern for pile goods is the orientation of raised fibres and yarns. Most pile goods have a natural pile direction, and through handling the pile may no longer be uniformly orientated. These types of materials should be brushed with an appropriate device prior to visual evaluation to re-orientate the pile into its natural pile direction. Also of interest when evaluating pile goods is the positioning of the sample during evaluation. When the pile direction faces the observer, the observer’s evaluation is influenced by the colour of the ends of the pile, while if the pile direction faces away from the observer, the observer’s evaluation is influenced by the colour of the sides of the pile yarns. Clear communication regarding sample positioning will eliminate any misunderstandings when evaluating these types of materials. Yarns and threads The preferred method for evaluation of yarns and threads is first to wind the sample skein on to a card in order to align the individual yarns. Care must be taken in this process to ensure that a sufficient quantity of material is used so as to completely obscure the card from view. Tension during winding must also be controlled to prevent bowing of the card and to prevent changes in appearance that are related to the application of excessive tension. Winding of wet yarn typically leads to excessive yarn tension or bowing of the sample card as the yarn dries. Loose fibre Loose fibre cannot be accurately evaluated in its natural loose form. As such, the fibre must be processed into a pad for evaluation. Loose fibre may also be compacted into a compression cell under uniform pressure, especially when the sample is to be evaluated instrumentally.
5.5
Future trends
Visual colour assessment by a skilled observer will always be a part of the colour development and quality control processes. This visual evaluation
A practical guide to visual evaluation of textile samples
91
process has traditionally required transfer of large numbers of physical samples between the organisations involved in colour development. This transfer of physical samples adds a significant amount of time to the development process as samples are shipped around the world, which ultimately translates into lengthy colour development cycles. In order to have finished goods available to consumers at the proper time, colour development must often begin months in advance of delivery deadlines. To shorten the development cycle, the transfer of physical samples must be eliminated. The question that must then be answered is how are samples to be evaluated if physical samples are not available? This problem is addressed by proper application of instrumental analysis followed by digital data communication and display of the sample colour on a calibrated monitor. Technologies currently exist to address each of these steps in digital colour development. Just as very specific guidelines must be followed to ensure proper visual assessment of textile samples, specific guidelines must also be followed to ensure accurate and repeatable instrumental analysis of textile samples. A successful programme of instrumental assessment of textile samples requires guidelines for sample size, sample orientation, the number of measurements performed on a single sample, sample conditioning and so forth. A key requirement for instrumental analysis is that the results are not only repeatable, but that they correlate well with visual colour evaluations. Extensive information on the subject of instrumental sample analysis has been presented at colour symposia developed by organisations such as AATCC and the Society of Dyers and Colourists (SDC). Test methods for successful implementation of instrumental analysis have also been developed by AATCC. Digital communication of instrumental data in the form of spectral reflectance values is facilitated by any of a number of commercially available computer software programs. These programs allow for instantaneous communication of colorimetric information between organisations anywhere around the world. Once received, the digital data can be evaluated numerically against established colorimetric tolerances, and decisions regarding acceptability can be made. Additional value in digital data is realised when the spectral reflectance values can be displayed on a calibrated monitor. Software for this purpose provides tools for characterisation of the monitor’s colour output followed by accurate representation of spectral reflectance data on images of the actual textile material. Accurate display of colour on a calibrated monitor in this way allows the observer to evaluate the quality of a sample using digital data rather than a physical sample, thereby eliminating the time required to deliver the sample and ultimately reducing the time required for colour development. These technologies have been successfully implemented by a number of retail and
92
Total colour management in textiles
clothing companies and their suppliers. Though they may never replace the need for visual evaluation of the final product, they are invaluable tools for increasing the efficiency of the colour development process.
5.6
Sources of further information
A number of organisations offer extensive information on textile visual analysis, instrumental sample analysis, testing, and industry trends in textile colour development. The following organisations should be considered as sources for reference information, as well as for symposia offerings: The American Association of Textile Chemists and Colorists (AATCC) PO Box 12215 Research Triangle Park, NC 27709, USA www.aatcc.org AATCC Review (published by AATCC) ASTM International 100 Barr Harbor Drive West Conshohocken, PA 19428-2959, USA www.astm.org The International Commission on Illumination (CIE) Kegelgasse 27 A-1030 Wien, Austria www.cie.co.at/cie/ International Organization for Standardization (ISO) 1, rue de Varembé, Caisse postale 56 CH-1211 Geneva 20, Switzerland www.iso.org The Inter-Society Color Council (ISCC) www.iscc.org The Society of Dyers and Colourists (SDC) PO Box 244, Perkin House, 82 Grattan Road, Bradford BD1 2JB, UK www.sdc.org.uk
5.7
References
Albers, J. (1975). Interaction of Color, New Haven, CT: Yale University Press. Berns, R. (2000). Principles of Color Technology, 3rd Edition, New York, NY: John Wiley & Sons, Inc.
A practical guide to visual evaluation of textile samples
93
2004 Technical Manual of the American Association of Textile Chemists and Colorists, Research Triangle Park, NC: AATCC. 2002 Annual Book of ASTM Standards Section Six, West Conshohocken, PA: ASTM International.
Part II
Managing colour
6 Colour simulation of textiles H S H E N A N D J H X I N, The Hong Kong Polytechnic University, Hong Kong
6.1
Introduction
Colour is one of the most important aspects for the textile and garment industry. In colour design and colour quality control, visualisation of solid colours on display devices has become a routine process for many companies. These solid colours are firstly measured using a spectrophotometer and are then displayed on a monitor. When properly calibrated and characterised, the monitor can display the colour of the physical sample with very high accuracy. A spectrophotometer measures the average colour in a fixed aperture, regardless of the spatial variation of the colour. As a consequence, it is possible that two textile fabrics with different visual colour appearance may have the same measurement result. In addition, when textile fabrics are heavily textured, their colours cannot be regarded as solid colours and therefore visualisation using solid colour is not suitable for heavily textured fabric samples. There is a need to map solid colours to various texture patterns and this colour mapping technique should generate a textured colour image accurately so that it is perceptually very close to the original image and/or physical textile fabric samples. Colour simulation of texture images also has other advantages such as visualisation of the yarn-dyed fabrics before they are actually produced. In addition to the two-dimensional textile fabrics, there is also a need for the colour synthesis of threedimensional textile products. For example, in a virtual exhibition of textile products or interior design, it is always desirable to change the colour of three-dimensional objects to obtain different visual effects. Image synthesis is an important research area in computer graphics, and methods including texture synthesis and colour transfer have been studied recently. The goal of image synthesis in computer graphics is to create photo-realistic images, without special consideration of colour accuracy. In textile applications, however, the colour fidelity is a very important issue. It is generally accepted that the pass and fail tolerance for textile fabrics is 97
98
Total colour management in textiles
about 0.6 to 1.0 CIELAB colour difference unit. For heavily textured colour samples, the visual tolerance may become larger due to the parametric effect of texture.1 Colour accuracy is also very important in the colour synthesis of three-dimensional products, as colour appearance is one of the major factors that affect customer appreciation. In this chapter, the colour mapping algorithm for a two-dimensional textile fabric image is first introduced, and then the texture effect on visual colour difference evaluation is investigated. This is followed by the colour synthesis technique for three-dimensional textile products, which is based upon a physical vision model. Finally, future trends in colour simulation are discussed and further information in this area is provided.
6.2
Characterisation of colour displays
In the textile industry, colour cathode-ray tubes (CRTs) are widely used for the visualisation of solid colours. To ensure that the colour displayed is the same as that of the physical sample, the CRTs must be characterised. Charac terisation of a computer-controlled CRT display is to establish the relationship between the digital signals and the output visual stimuli. The input signals are typically the digital-to-analogue converter (DAC) pixel values, denoted as RGB, and the output stimuli are the measured CIE tristimulus values such as XYZ. The gain-offset-gamma (GOG) model is the most wellknown mathematical method for characterisation, and it is also the method recommended by CIE.2,3 The characterisation consists of two stages, as shown in Fig. 6.1. Stage one is a one-dimensional non-linear transformation for each channel from RGB to the output luminance levels LRLGLB, which is known as gamma correction in the GOG model. Stage two is a linear, threedimensional transformation from luminance LRLGLB to tristimulus values XYZ using a 3 ¥ 3 matrix. To calculate the parameters in the GOG model, the luminances and chromaticities of the three primaries, white and black points, and a series of neutral scales needs to be measured. Recently, flat-panel liquid-crystal displays (LCDs) have become more and more popular in various applications and eventually they may replace all CRTs. LCDs are superior to CRTs in terms of luminance, contrast ratio, sharpness and spatial uniformity, while its main deficiency is the view angle dependencies. The characterisation of LCD is similar to that of CRT, with the exception that the linearisation is accomplished by one-dimensional look-up tables.4 RGB
Linearisation
LRLGLB
3 × 3 matrix
6.1 The two-stage characterisation of displays.
XYZ
6.3
Colour simulation of textiles
99
Colour mapping for two-dimensional texture image
Texture and colour are two important characteristics of images.5 Given an original texture image and a solid target colour, the purpose of colour mapping is to simulate a new colour texture image that is a close replica of the target image. With the technique of colour mapping, the final styles of textured textile fabrics could be visualised before they were actually produced. In the literature, Botchko et al.6 analysed the relationship between the mean spectral and standard deviation of natural objects, and proposed parametric methods for the virtual colourisation of texture image. Montag and Berns also presented a method to simulate textile colour texture image in CIELCH colour space, using a singular value decomposition technique.7 In this section, a computational model for colour mapping on a textile texture image, considering colour fidelity, is presented. Using this model, the colour accuracy in texture image simulation can be improved. Texture images were scanned in using an Epson GT-10000+ flatbed colour image scanner. The physical textile samples are cotton fabrics with different woven patterns. Samples of each of the woven patterns were dyed into several colours, including green, orange, purple, pink and turquoise using reactive dyestuffs. These textile fabrics were then scanned in a resolution that gave approximately equal visual appearances to those of the physical samples when viewed under normal viewing distance of about 25 to 30 cm. Some texture images of these textile fabrics are shown in Fig. 6.2.
6.3.1 Analysis of texture image characteristics The RGB space directly corresponds to the output of imaging devices, such as colour scanner and digital camera. Let L(l; p) refer to the spectral distribution function of light entering the imaging device at pixel position p, and sn(l) be the spectral sensitivity of the nth sensor. The output signal becomes: V pn = ÚL(l; p)sn (l)dl
6.2 Four textile fabric samples with different texture patterns.
[6.1]
100
Total colour management in textiles
It should be noted that (eqn 6.1) assumes that the output of the imaging device is proportional to the light intensity entering the sensor. The luminance channel contains mainly the texture information, and can be calculated according to: Yp = 0.299V1p + 0.587V2p + 0.114V3p
[6.2]
The histograms of the red, green, blue and luminance channels (yellow) of a typical texture image are shown in Fig. 6.3 (see also colour section). The shapes of the histograms in different channels are similar, with differences only in height and width. However, the histogram only demonstrates the global statistical distribution of each channel. To simulate a texture image at the pixel-wise scale, the spatial distribution should be analysed. The correlation between the RGB and the luminance channels can be investigated using the pixel deviations to the mean value in RGB channel and luminance channel, as shown in Fig. 6.4. The degree of correlation can be described using the correlation coefficient. For a perfect correlation, the correlation coefficient should be equal to 1.0. The correlation coefficient of the blue channel, 0.988, is the highest, followed by that of the red channel, 0.985. The correlation coefficient of the green channel, 0.915, is the lowest. This observation indicates the existence of high channel correlation of texture images.
6.3.2 Colour mapping algorithm The colour mapping algorithm assumes that only the spatial distribution of the luminance channel is known, and the purpose of the algorithm is to derive three-dimensional spatial distributions of the red, green and blue channels from a one-dimensional channel. This problem is therefore undetermined since the combination of different Vnp values can produce the same Yp value (see eqn 6.2). From the constraint of channel correlation, (b)
(a)
Pixel number
Blue
Green
Yellow Red
Intensity
6.3 (a) A typical texture image, (b) its histograms of different channels.
Colour simulation of textiles
101
a solution can be found so that the synthesised texture image is perceptually close to the target one. That is, the overall colour difference between the generated and actual image is very small. Let Sn be the user-specified target colour in RGB space. Then, the generated colour at pixel p can be simply calculated as Upn = Sn + DYp
[6.3]
where DYp is the pixel deviation to mean luminance. For instance, if the selected target colour is (100,150,200), and for a particular pixel p, DYp = 10, then, for that pixel, the new colour becomes (110,160,210). When U pn is smaller than 0 or larger than 255, it should be clipped to 0 and 255, respectively. The colour mapping algorithm according to (eqn 6.3) is simplified, as it assumes the pixel deviations in RGB channels are exactly the same as that in the luminance channel. For texture images, it is not the case (see Fig. 6.4, also in colour section). By taking the standard deviation of the colour channel into consideration, the colour mapping algorithm becomes
U p n = Sn +
( σσ ) n
[6.4]
∆Y p
Y
where sn is the standard deviation of the nth channel of the generated texture image, and sY is the standard deviation of the luminance channel. For texture images of textile fabrics, it is found that the standard deviation is always related to the mean colour.8,9 Therefore, it is possible to calculate the sn value for a given target colour using their relationship. Alternatively, the
(a)
Pixel deviation of RGB channel
(b)
60 40 20 0
–60
–40
–20
0
20
40
–20 –40
red channel green channel blue channel
–60 Pixel deviation of luminance channel
6.4 (a) Texture image, (b) relationship between pixel deviation in luminance channel and that in RGB channels.
60
102
Total colour management in textiles
value of sn can also be specified by the user if he/she wants to learn the visual appearance of the synthesised image under different texture strengths.
6.3.3 Colour mapping results and discussions Some experimental results are shown in Fig. 6.5 (see also colour section). To evaluate the colour accuracy of the algorithm, we use the mean colour of the original image as a target colour to generate a new texture image. The colour scanner was characterised and the RGB value of each pixel was then transformed into a device-independent CIEXYZ value. Then, colour differences DE*ab between the original and generated image were calculated on a pixel-wise scale, as shown in Fig. 6.6. (a)
(c)
(b)
(d)
6.5 Experimental results of colour mapping algorithm. (a) original, (b) luminance channel, (c) colour mapping using the mean colour of the original image as target colour, (d) colour mapping using another target colour.
Colour difference
3 2 1 0 0
16
32
48
64
80
Sample number
6.6 Average pixel-wise colour difference between the original and generated image.
Colour simulation of textiles
103
It was reported that the threshold for detecting the colour difference of a pair of solid colour samples is around 1.0 unit of DE*ab colour difference. Nevertheless, in the case of colour with texture, one can hardly perceive the difference between the original image and the generated one even if the mean colour difference is 1.29, as shown in this study. This result is attributable to the parametric effect of the texture.
6.4
Texture effect on visual colour difference evaluation
It is known that texture has an important effect on colour perception, as a parametric factor to the colour difference equations. However, few reports on quantitative analysis of the influence have been published. A lightness tolerance thresholds experiment was performed by Montag and Berns7 using the stimuli with a simulated texture of thread wound on a card.They found that the textured stimuli had the effect of increasing the tolerance thresholds by a factor of almost 2, when compared with uniform stimuli. In this section, the influence of texture levels on visual colour difference is evaluated by using 15 samples with different texture patterns. These textured colour samples were first generated using a colour mapping algorithm, together with five pre-determined colour centres, and then displayed on a characterised cathode ray tube (CRT) monitor. For each texture pattern, two comparison pairs (with ±5 lightness difference) were used, and thus a total of 150 (15 samples ¥ 5 colour centres ¥ 2 pairs) colour difference pairs were evaluated using the grey-scale method.
6.4.1 Texture characterisation using histogram half-width For textile fabric, the texture is quite regular, that is, the elementary woven or knitted pattern is repeated over the whole image. The texture level can be described by its coarseness index in the sense that a rhomb fabric is coarser than a plain one under the same viewing conditions. The coarseness index is related to the spatial repetition period of the local structure. A large repetition period implies a coarse texture, while a small period implies a fine one. Due to the regularity of the texture patterns of textile fabrics, the coarseness index is considered to be effective for quantitative measurement of texture level. The shape of the histogram is directly related to the coarseness of a texture, i.e. the width of coarse texture is wider than that of fine texture. Although the histogram only describes the statistical distribution of the luminance channel of the texture image, it is quite effective in characterizing the regular texture of textile fabric. The half-width WY of a histogram in the luminance channel can be used to quantify the coarseness of a texture, which is called texture strength here. Suppose the luminance YC contains a maximum pixel number V, the half-
104
Total colour management in textiles
width WY is defined as the distance between the higher luminance YR and lower luminance YL containing V/2 pixel number: WY = YR - YL
[6.5]
Figure 6.7 illustrates the definition of texture strength in a luminance channel.
6.4.2 Colour difference pairs and visual assessment Five colour centres were selected based on the findings whose colorimetric values are shown in Table 6.1. These colour centres were converted to display RGB space according to the gain-offset-gamma model.2,3 For each colour centre, 30 colour mapped texture image pairs were generated, two pairs for each of the 15 textures. A spectroradiometer was used to measure the colour difference of the textured colour pairs. The distance between the spectroradiometer and the displayed texture image was about 40 cm and the average colour of a relatively large area was measured. It is known that the human visual system is more sensitive to luminance contrast than to chromatic contrast.10 Therefore, the texture effect was investigated on a medium colour difference of about 5.0 CIELAB units in lightness direction. This range is also on the borderline of the CIE recom-
Pixel number
V
V/2 WY YC
YL
0
YR
255
Luminance
6.7 Definition of texture strength WY. Table 6.1 Colour centres used in the visual colour difference evaluation
Orange
Yellow
Grey
Green
Blue
L* a* b*
48.33 13.14 16.87
69.28 4.48 19.11
68.25 3.21 0.29
28.69 -17.83 -0.50
28.96 4.43 -9.13
Colour simulation of textiles
105
mended colour difference magnitude for applying CIE94.11 The target solid colours were fine tuned until the generated textured colour pairs had ±5.0 lightness differences when measured using a spectroradiometer. The grey-scale comparison method was adopted, considering its wide use in assessing colour change in the textile industry. Five grey-scales according to the ISO standard12 were used in the evaluation. The experiment was conducted in a completely dark room and, thus, the influence of ambient illumination was eliminated. The arrangement of grey-scales and a textured pair (or solid pair) on a CRT monitor is shown in Fig. 6.8 (see also colour section). The size of the displayed samples, including the grey-scales, was 3 inches square. All the samples displayed had no black frame and there was no dividing line between the pairs. In total, ten observers were asked to rate the colour difference using the solid colour greyscale grades. The viewing distance is the same as that of a spectroradiometer and the viewing angle is 0° to the normal of the sample. At the beginning of each assessment, the right patch of the grey-scale pair was the same as the left one (standard). When the observer clicked one grey patch on top, the right patch of the grey-scale pairs changed to that grey. Observers were asked to give a grey-scale grade that produced approximately the same colour difference as the comparison pair. If the grade of a sample pair did not equal the grade of the closest grey-scale, observers were encouraged to provide an intermediate step. In order to evaluate the texture effect, a reference experiment, using solid colour difference pairs with the same measured colour difference as the texture mapped pairs, was also conducted under the same viewing condition.
Background
Grey-scale patches
Standard
Grey-scale pair
Comparison pair
6.8 Arrangement of sample pairs and grey scales on a CRT monitor.
106
Total colour management in textiles
It is known that the reliability of the results is critical in psychophysical experiments. The observer accuracy and repeatability tests were employed to check the reliability of the results. Observer accuracy represents the average deviation between each individual and the mean visual result of a panel, while observer repeatability represents the variation of the visual assessment of a particular observer. The performance factor (PF/3)11,13 has been widely used as an indicator for the observer accuracy and repeatability and for the performance of colour difference formulae in comparison with visual results. A PF/3 combines three measures of fit: gamma factor g, co efficient of variation CV and VAB. The calculation of PF/3 is given as PF/3 = 100[(g - 1) + VAB + CV/100]/3
[6.6]
A low PF/3 value indicates a small difference between two variables. In the experiment, ten observers were asked to assess each textured and solid colour pairs twice. The observer accuracy and repeatability were 26.7 and 32.1, respectively. Considering these values are acceptable, all of the ten observers’ results were used.
6.4.3 Visual evaluation results and discussions The grey-scale results were converted to the visual difference DV according to a third-order polynomial conversion equation derived using the five grey scales. Let DV+t,W and DV-t,W be the visual difference of textured colour pairs of strength WY, with an increase and decrease of 5.0 units DE*ab in lightness scale, the average visual difference DVt,W is
Visual difference DVt,W of textured pairs
DVt,W = 12 (DV+t,W + DV-t,W)
[6.7]
5 y = –0.0283x + 4.1323 r 2 = 0.9257
4 3 2 1 0 0
20
40
60
80
Texture strength, WY
6.9 Visual difference DVt,W against texture strength WY for the orange colour centre. The y-error bars show ±1 standard deviations.
Colour simulation of textiles
107
Very high correlation was found between the visual difference DVt,W and the texture strength WY for the five colour centres used. Figure 6.9 shows the relationship between DVt,W and WY of the orange colour centre. Relationships of other colour centres are similar. Let KW be the ratio between the visual difference of a textured colour pair with strength WY and that of a solid colour pair: 1 ∆V + tw ∆V − tw Kw = + 2 ∆V + s ∆V − s
[6.8]
where DV+s and DV-s are the visual differences of the solid colour pairs with an increase and decrease of 5.0 in lightness with respect to the colour centre. The KW value deviating from 1.0 indicates a parametric effect. The relationships between KW and texture strength WY for the orange colour centre is shown in Fig. 6.10. When the half-width of the Y channel WY is very low, which indicates low texture strength, the KW value is closer to 1.0. However, when WY increases, the KW value becomes smaller, indicating a stronger parametric effect. Figures 6.9 and 6.10 clearly show that the simple linear fitting could successfully reveal the texture effect on visual difference evaluation. The slope D value of the fitting line can be used to further quantify the variation of visual difference with respect to texture strength. In Table 6.2, it is found that the D values for the five colour centres are quite close. The quantitative analysis found that every increase of 10 units in texture strength in luminance scale will cause a 0.25 decrease of visual difference, and a 0.05 decrease of KW value. Based on these fundamental quantitative results, it is possible to introduce the parametric effect of the texture effect
1 y = –0.006x + 0.8699 r 2 = 0.926
Kw value
0.8 0.6 0.4 0.2 0 0
20
40
60
80
Texture strength, WY
6.10 KW value against texture strength WY for the orange colour centre. The y-error bars show ±1 standard deviation.
108
Total colour management in textiles
Table 6.2 Slope of fitting line of visual difference DVt,W and KW value with texture strength WY
Orange
Yellow
Grey
Green
Blue
Slope of DVt,W Slope of KW
-0.0283 -0.0060
-0.0271 -0.0054
-0.0231 -0.0049
-0.0232 -0.0050
-0.0263 -0.0059
into a colour difference equation as a scale factor related to texture level, provided more attributes of colour differences are investigated. This section presents the fundamental investigation of the texture effect on the visual colour difference evaluation. It is noted that two different texture images may have the same histogram, as the histogram ignores the spatial distribution of images. Therefore, it may be desirable to use additional textural features such as coarseness, contrast, busyness and complexity to represent the visual properties of texture. It should also be pointed out that consideration of the colour difference only in the lightness direction is not sufficient for practical applications, since chroma and hue may also be influenced by the texture effect.
6.5
Colour synthesis for three-dimensional objects
In computer graphics, to create a realistic image, a complete model or description of the reflectance is needed for each of the reflective objects in the scene. It is difficult to establish a complete representation of the reflective behaviour of the surface of an object due to complex interactions between light and the surface, such as by polarisation, scattering, phosphorescence and fluorescence. For opaque surfaces, the bidirectional reflectance distribution function (BRDF),14 which is a function of the incident and the reflecting angles of light, is usually employed for colour rendering. However, considering the inconvenience of collecting BRDF data, it is often desirable to perform colour rendering based on two-dimensional images.15–17 In this section, an image-based colour synthesis technique is presented to modify the colour of three-dimensional object surfaces in a single image so that the colour appearance of the synthesised image is perceptually close to that of the target one. More precisely, given two objects I1 and I2 with different surface colours, the purpose is to first recover the intrinsic colour characteristic C1 and C2 from I1 and I2, respectively, and then map C2 on to object I1, so that the colour appearance of the synthesised new object I1 is very similar to that of I2. In this technique, the implicit geometric coefficient is calculated
Colour simulation of textiles
109
using the least-squares method. Then, the body colour of the three-dimensional objects is recovered using the high correlation between different channels. With the recovered geometric coefficient and the body colour, new images for 3D images can be synthesised. The colour synthesis technique can be applied to a variety of materials that can be described by the dichromatic reflection model.18 This technique is especially useful for textile products, because of the complicated geometry of their surfaces.
6.5.1 Dichromatic-based modelling of colour reflections In computer vision, the experiential dichromatic reflection model was first introduced by Shafer to describe light reflection.18 Figure 6.11 illustrates the interaction between illumination and object surface. When a ray of light strikes the surface of an inhomogeneous material, part of it is reflected from the interface between the object and air immediately because of the difference in the refractive indices of the object and the air. This part of the light is called the surface reflection. The surface reflection is dependent upon the orientation of the local surface that varies along the interface. When the surface is smooth, the surface reflection is very directional, and when the surface has some degree of roughness, the light rays are scattered to some extent around the angle of perfect mirror reflection. In addition to the surface reflection, part of the light penetrates the interface and enters the body of the object, where it keeps hitting the colourants and is scattered by the colourants. The spectral power distribution of the surface reflection is very similar to that of the incident light. The colourants and the medium also absorb part of the light. Part of the scattering light arrives back at the object surface and re-enters the air. This part of the light has been selec-
Incident light
Surface reflection
Body reflection Interface Medium Colourant
6.11 Reflection from an inhomogeneous surface consists of surface reflection and body reflection.
110
Total colour management in textiles
tively absorbed by the colourants and the medium. It is called body reflection. Based on the dichromatic reflection model, the spectral reflectance Rp(l) at pixel position p can be decomposed into diffuse reflectance Rb(l) and constant surface reflectance Rs: Rp(l) = apRb(l) + bpRs
[6.9]
where (ap,bp) is the implicit geometric coefficient. Similarly, the RGB output, VP, of an imaging device can be decomposed into diffuse colour, VB, and surface colour, VS: Vp = apVB + bpVS
[6.10] p
p
In (6.11), when VB and VS are known, the coefficient (a ,b ) can be resolved using the least-squares method.
6.5.2 Calculation of body colour In an imaging system, the illumination colour VS can easily be obtained by imaging a perfect white patch. Thus, one purpose of colour synthesis is to accurately recover body colour, VB, from images. From (eqn 6.10), the closer the coefficient bp is to zero, the closer V p is to VB. The cosine of the including angle qp between these two vectors was calculated and then 10% of pixels with large qp were selected. For simplicity, RB is used to represent the set of pixels with mainly body colour but little surface colour. The distribution relationship of camera responses between the green and red channel of a red plastic cup in RB is shown in Fig. 6.12. The point cloud forms a straight line with a high correlation coefficient. Let hij be the slope between the ith and jth channel. Then the chromaticity of body colour, VB, can be solved as:19 1 1 η VB = 21 1 + η21 + η31 η31
[6.11]
It is noted that (eqn 6.11) does not decide the magnitude of the body colour. For the purpose of colour synthesis, the colour with largest ap value with respect to the solved VB was selected as the ‘most diffuse’ reference colour, VBR. As the value of bp is not considered, the chromaticity of VBR may not be exactly the same as that of the solved VB. However, since VBR is chosen from set RB, the surface reflection component of VBR should also be quite small, and therefore it can be used to represent the actual diffuse colour.
Colour simulation of textiles
y = 0.2828x r 2 = 0.9748
100
Green channel
111
50
0
0
20
40
60
80
100
120
Red channel
6.12 Distribution of pixel values of green channel with respect to those of red channel in colour set RB.
6.5.3 Transform of body colour under different illuminants In the case of different illuminants, it is necessary to predict the new body colour. This problem falls into the research area of colour constancy.20,21 The aim of colour constancy is to find a linear transform from colour vector, V, under one illuminant to colour vector, V¢, under another illuminant:21 V¢ = TLV
[6.12]
where TL is a 3 ¥ 3 illuminant transform matrix. To solve TL, three or more distinct training colours can be used. With the calculated TL, the new body colours under the new illuminant can easily be calculated according to (6.12).
6.5.4 Colour synthesis results and discussion The process of colour synthesis for three-dimensional objects can be summarised as follows: 1. calculating the diffuse colours of different regions; 2. recovering the geometric coefficient using least-squares method; 3. calculating the new body and surface colours under new a illuminant if necessary; 4. performing colour synthesis.
112
Total colour management in textiles
In the experiment, the imaging device is a QImaging Retiga EXi digital 12-bits monochromatic CCD camera, together with the QImaging RGB liquid crystal colour filter, to acquire colour images. The linearity of the camera was verified by investigating the relationship between the mean reflectance and camera response values of the 20 Kodak Grayscales. To evaluate the colour accuracy of the colour synthesis technique, four plastic cups with the same shape but different colours were used. The implicit geometric coefficient (ap,bp) was calculated from one cup and then applied to another cup. The coefficient calculated from the yellow cup is shown in Fig. 6.13 (see also colour section). Table 6.3 shows the mean colour difference, DE94*, between the target object and the synthesised object when applying a weighted least-square method. It is found that, in many cases, the colour difference is very small. Large errors occur when applying the coefficients of green and blue cups to synthesise yellow cups. The reason is that the colour in the red channel of these two source cups is much smaller than in that of yellow cup, and thus a small deviation of (ap,bp) is enlarged in the colour synthesis process.
(a)
(b)
(c)
6.13 Calculated geometric coefficient ap values (b) and bp values (c) from a colour image (a). The coefficients were rescaled for display. Table 6.3 Mean colour difference DE94* of colour transfer between cups with different colour
Target cups
Red
Green
Blue
Yellow
Source cups
0.441 2.412 2.482 0.814
1.634 1.240 1.190 1.644
1.716 1.255 0.341 1.607
2.844 4.543 5.332 1.833
Red Green Blue Yellow
Colour simulation of textiles
113
A light blue filter was used in front of the lens of the camera to produce a new illuminant condition. The illuminant transform matrix, TL, was calculated using the 24 colour patches on MechBetch® ColorChecker® and is given as follows: 0.623 0.027 −0.003 TL = −0.006 0.745 0.013 0.002 − 0.0007 0.840
[6.13]
The mean colour difference, DE94*, between the actual colour and the calculated one using TL is 0.385. Colour synthesis under the new illuminant is shown in Fig. 6.14 (see also colour section). It can be found that the colour appearances of these two images are quite close. For multi-coloured objects, image segmentation should be performed before colour synthesis.19 Then, the body colour of each separated region can be calculated and colour synthesis can be applied to each region. Figure 6.15 (see also colour section) shows the colour synthesis results of some three-dimensional textile products. It is found that the colour appearances of the synthesised images and the target images are very close.
6.6
Future trends and further information
In this chapter a colour mapping algorithm for a two-dimensional texture image of textile fabric is introduced in section 6.2. In section 6.4, a colour synthesis technique for three-dimensional textile products presented. In these two algorithms, colour fidelity is emphasised, because of its impor-
(a)
(b)
(c)
6.14 Colour synthesis results under different illuminant. (a) Original image of blue cup, (b) synthesised image of red cup under new illuminant, (c) actual image of red cup acquired under new illuminant.
114
Total colour management in textiles
(a)
(b)
(a)
(b)
(c)
(c)
6.15 Synthesised image (b) is produced using the geometric information of original image (a) and the body colour of target image (c).
tance in the textile industry. Based on the colour mapping algorithm, the texture effect on visual colour difference is investigated in section 6.3. Colour simulation is an active area in computer vision and computer graphics. Recently, new algorithms for colour transfer in grey/colour image/ video have been proposed.7–9,14,17,19,22–24 Of these algorithms, some are based on the physical vision model, while others are based on the statistical characteristics of natural images. Future studies may focus on research areas such as colour fidelity, computational efficiency and reduction of user interaction. With wide applications in multimedia, colour simulation in threedimensional environments may well be the future direction. The research will, undoubtedly, greatly benefit the visual exhibition of textile garments and interior design. The visualisation and matching of solid colour on CRT has become a routine process in many textile companies. However, due to the texture pattern on textile fabrics, more efforts should be devoted to the investigation of texture effects on visual colour difference evaluation. One possible future line of research might be to extract additional texture features corresponding to visual properties of texture images. Then, texture features
Colour simulation of textiles
115
could be incorporated into colour difference formulae, based on the results of new set of psychophysical experiments.
6.7
References
1. Xin, J.H., Shen, H.L. and Lam C.C. (2005). Investigation of texture effect on visual colour difference evaluation, Color Research and Application, 30, 341–347. 2. Berns, R.S., Motta, R.J. and Gorzynski, M.E. (1993). CRT colorimetry. Part I: theory and practice. Color Research and Application, 18, 299–314. 3. Commission Internationale de l’Eclairage (CIE) (1996). The relationship between digital and colorimetric data for computer-controlled CRT displays, Publication CIE No. 122, Austria: Bureau Central de la CIE. 4. Day, E.A., Taplin, L. and Berns, R.S. (2004). Colorimetric characterization of a computer-controlled liquid crystal display. Color Research and Application, 29, 365–373. 5. Pratt, W.K. (1991). Digital Image Processing, 2nd edn. New York: John Wiley & Sons. 6. Botchko, V., Jaaskelainen, T. and Parkkinen, J. (2002). Multispectral texture model for color and highlight reproduction, First European Conference on Colour in Graphics, Image, and Vision, France, 603–607. 7. Montag, E.D. and Berns, R.S. (2000). Lightness dependencies and the effect of texture on suprathreshold lightness tolerances. Color Research and Application, 25, 241–249. 8. Shen, H.L. and Xin, J.H. (2004). Dichromatic based rendering of texture image with high colour fidelity. Journal of Imaging Science and Technology, 48, 246–250. 9. Xin, J.H. and Shen, H.L. (2004). Recolouring digital textile printing design with high fidelity. Coloration Technology, 120, 6–13. 10. Fairchild, M.D. (1998). Color Appearance Models. Reading, MA: Addison–Wesley. 11. Xin, J.H., Lam, C.C. and Luo, M.R. (2001). Investigation of parametric effects using medium colour difference pairs. Color Research and Application, 26, 376–383. 12. International Standard ISO 105-A02 (1993). Textiles – tests for colour fastness – Part A02: Grey scale for assessing change in colour. 13. Guan, S.S. and Luo, M.R. (1999). Investigation of parametric effects using small colour difference. Color Research and Application, 24, 331–343. 14. Nicodemus, F.E., Richmond, J.C., Hsia, J.J., Ginsberg, I.W. and Limperis, T. (1997). Geometric considerations and nomenclature for reflectance, US Department of Commerce, National Bureau of Standards, Monograph 160. 15. Reinhard, E., Ashikhmin, M., Gooch, B. and Shirley, P. (2001). Color transfer between images. IEEE Computer Graphics and Applications, 21, 34–41. 16. Peng, L. (1998). Dichromatic based photographic modification. Proceedings of the Sixth Color Imaging Conference, 96–99. 17. Xin, J.H. and Shen, H.L. (2004). Accurate color synthesis of three-dimensional objects in an image. Journal of the Optical Society of America A, 21, 713–722.
116
Total colour management in textiles
18. Shafer, S.A. (1985). Using color to separate reflection components. Color Research and Application, 10, 210–218. 19. Shen, H.L. and Xin, J.H. (2005). Transferring colors between three-dimensional objects. Applied Optics, 44, 1969–1976. 20. Barnard, K., Cardei, V. and Funt, B. (2002). A comparison of computational color constancy algorithms – Part I: methodology and experiments with synthesized data. IEEE Transactions on Image Processing, 11, 972–983. 21. Finlayson, G.D., Drew, M.S. and Funt, B.V. (1994). Color constancy: generalized diagonal transforms suffice. Journal of the Optical Society of America A, 11, 3011–3019. 22. Xin, J.H. and Shen, H.L. (2003). Computational model for color mapping on texture images. Journal of Electronic Imaging, 12, 697–704. 23. Welsh, T., Ashikhmin, M. and Mueller, K. (2002). Transferring color to grayscale images. ACM Transactions on Graphics, 20, 277–280. 24. Levin, A., Lischinski, D. and Weiss, Y. (2004). Colorization using optimization, ACM Transactions on Graphics, 23, 689–694.
7 Effective colour communication from mind to market G L I T T L E W O O D, Datacolor, UK
7.1
Introduction
What exactly is meant by the now ‘well-worn’ phrase ‘from mind to market’ is, potentially, open to widespread interpretation and is a term that is common to many different manufacturing sectors. However, for the most part, it is acknowledged as the process whereby the manufacturer transforms the concept or specification communicated to him by the designer into a physical commodity, usually by way of an initial prototype, which upon approval, is then manufactured in quantity and transported to the point of sale for ultimate purchase and utilisation by the end consumer. When we focus on the retail apparel sector and, specifically, on the management and communication of colour from ‘mind to market’, we are typically describing the process whereby textile designers collate their target colours and then communicate their colour requirements to the supply chain, usually in the form of a request for a physical sample or ‘labdip’. Once this ‘pre-production’ sample is approved, the manufacturer is usually requested to submit a sample from their first production run before managing ongoing production in line with the retailer’s requirements. Garment assembly and transportation to warehousing followed by final despatch to the ultimate point of sale, whether that be a ‘bricks and mortar’ retail environment or transactional website, completes this cycle. The above is an oversimplified version of events and is very specific to the management of colour within the cycle; there are many other parallel processes that take place before, during and after, which, as well as adding value to the final product, introduce additional players, complexity and, inevitably, time and cost. However, these increments, together with the colour sampling process itself, are ‘necessary evils’ and are business critical to any reputable retailer in today’s worldwide market place. This process was always challenging to manage even in the ‘good old days’ when retailers were able to source significant volumes of merchandise from manufacturers in relatively close proximity to their headquarters. Nowadays, the onset and explosion of global sourcing, which is here to stay, 117
118
Total colour management in textiles
has introduced numerous additional complications including cultures, language barriers, time zones, technical expertise, capacity issues and additional ‘players’ to name but a few. Communication is intrinsically linked to all of the above and ineffective communication, especially on colour, has severe ramifications in terms of time, cost, quality and overall profitability to today’s apparel retailers. It is precisely for this reason that the proactive and farsighted retailers in today’s market are closely and continually examining the information flow within their value-added chains, and have already taken steps to utilise the latest ‘digital’ tools available on the market to help them to communicate colour and its associated information more effectively. This chapter will examine the importance of colour within today’s retail environment and, critically, review historic and current practices for colour communication between the specifier and the supply chain. This will be followed by an analysis of current ‘best practice’ in the chain, together with the associated impact and benefits and will conclude with a proposed forecast of how colour could be managed in the future.
7.2
The ‘fast fashion’ concept and its effect on colour
‘It’s a changing world’ so they say and one major paradigm shift over recent years has been the change in how the ‘typical average consumer’ shops for clothing. It cannot be denied that, without really recognising it ourselves, our shopping patterns have changed beyond all recognition over the past 5–10 years. So what has changed? When we shop As a direct result of increased Sunday trading, extended opening hours and public holidays, we are now shopping more often than in previous years – for better or for worse, our ‘shopping’ window has enlarged considerably. How often we shop The massive explosion of the so-called ‘value sector’ in recent times, together with a significant improvement of the product offer, has made us much more value conscious and has also encouraged a more regular and increased average purchasing volume (though not necessarily an increased average financial value/spend). Where we shop The ongoing proliferation of ‘out-of-town’ shopping centres and ‘outlet malls’ has opened up a world of opportunity and choice to us. When
Effective colour communication from mind to market
119
supplemented with the now mandatory and varied eating and entertainment options, this becomes a major value-added shopping experience in stark contrast to the traditional ‘high street’ experience. Online shopping continues to grow in popularity and will continue to do so as our leisure time is increasingly at a premium – vendors are now achieving much higher levels of customer satisfaction due to improved user interface, choice, fulfilment and security, and are reaping the rewards. However, the sector with unprecedented growth has to be the supermarkets and the phenomenal rise in non-food sales, especially in clothing. Enticed by the ‘one-stop’ opportunity but predominantly the value of their offer, these operators are growing at a phenomenal rate, confirmed by the recent establishment of George at Asda as the UK’s No. 1 clothing retailer. A combination of all of the above has introduced a significant amount of additional competition in what was already a fiercely competitive market and this has been compounded by the advent and expansion of international retailers outside of their regional core markets. At the same time, and partly as a result of the above, generally we have become much more demanding and sophisticated shoppers, expecting and demanding increased choice and value. However, the most significant behavioural shift has been our average frequency of purchase, which has come down from approximately 3 months to an average of around 6 weeks. Given this, we naturally expect a new product offer every time we enter a store, and the retailer without such an offer seriously diminishes the sales prospects. Whether this phenomenon was consumer or retailer driven is subject to debate but this scenario undoubtedly gave birth to the concept of ‘fast fashion’ as we see it today. The traditional ‘four-season approach’ is long since gone and this has been recognised and confirmed by insightful and very successful players such as Zara, H&M, Top Shop, New Look and George to name but a few. However, fast fashion for the retailer has a myriad of associated considerations including design, product development, sourcing, merchandising and logistics, and all of these naturally become more challenging as the supply chain becomes more disparate. Having said that, one common strand throughout the entire process, and the most important decision influencer at the point of sale is the right colour. More and more retailers are now realising that the right colour in the right place at the right time is absolutely business critical, whether this be as a means of differentiation or even for long-term survival! This is ‘easier said than done’ and is certainly not achievable using the traditional business models employed historically. Inevitably then it begs the question ‘How is colour typically managed in retail and where are the areas for improvement?’
120
Total colour management in textiles
7.3
Colour palette development as part of the whole product development process
The starting point for the whole colour development cycle is the composition of the colour palette itself and, typically, it will involve the following generic steps: (The frequency and approach to this exercise will depend upon the product offer and business model of the individual retailers.) Initially, design teams visit the regular worldwide fashion exhibitions and shows to review the forecasted designs, themes and colour trends and collect physical representations of the same. Additionally, shopping trips to competitors or aspirational brands are undertaken and samples of popular colours and designs are either purchased or visualised. This information can then be supplemented by information from dedicated trend prediction service companies and the traditional colour atlas providers such as Pantone, NCS, Scotdic, etc. Normally, this information is then assessed and translated on to physical storyboards in order to select, group or filter the colours/themes during selection meetings, either on a departmental or on a total company basis. The composition of this initial palette will usually include a combination of fabric swatches, pieces of yarn or thread, magazine tears or other printed material and maybe samples from colour atlases. Purely from a colour perspective, the objective at this stage is to consider both individual colour and colour combinations, with additional or parallel discussions taking place on how these colours will then be translated into the selected designs and into the final merchandising offer. The initial palette is then prioritised and streamlined and then, depending on the retailer, the presentation of the palette could be aesthetically enhanced by matching the palette to a CAD printout, colour atlas samples or, occasionally dyed samples from selected suppliers. From a design perspective, for many retailers, this is normally the cutoff point in responsibility. The palette has been creatively thought out, discussed and composed, yet very little consideration, if any, has been given to the technical feasibility of the palette on production substrates. This is typically considered to fall under the auspices of the QC, product development or technical department and, probably, rightly so. This verification process is usually achieved by way of the sampling or lab-dipping process via the supply chain; however, as will be discussed in later chapters, the quality (or lack of) of the original target palette, plus the need for the design team to design and approve such samples with limited production experience, has a number of inherent disadvantages. The collation of these concept colours is usually done some 12–18 months in advance, and sometimes even longer ahead, and this is usually as a direct result of the length of the product development cycle prior to getting goods
Effective colour communication from mind to market
121
in store. Given the observations in the previous section and the fact that opinions and trends can change dramatically within such a timeframe, this is a far from an ideal situation. There is a strong demand and need for just-in-time palette selection and production commitments, together with improved in-season flexibility, and this is very do-able with the latest digital tools and processes. Some retailers are already experiencing such benefits yet there are many more who aren’t and, in order to do so, they must critically examine their existing processes as we discuss in the next chapter.
7.4
Review of existing ‘manual’ communication methods between design and production and why things go wrong
For the purpose of this chapter, in order to illustrate the worst-case scenario, the author will be using a retailer with no formalised colour management procedures or instrumental systems as a case study example. There are a large number of major retailers who do have well-established colour management programmes and this number is increasing but, at the same time, there are at least an equal number that have no programme at all. So now we have the situation where the QC or product development team have received a physical palette from design and have been instructed to translate this into physical fabric swatches (lab-dips) via the supply chain. The main objectives are to assess if the designed colour is achievable on the selected substrates, to have a reference point for ongoing production and, finally, to convert the initial palette into a more presentable format that can be used as an ongoing tool for further design, merchandising and selection discussions. The first challenge is to obtain sufficient quantities of each target colour to distribute to the supply chain in multiple locations and to a chain that typically will include numerous offshore offices, agents/vendors, garment makers, dye/mills and trim/accessory suppliers. This is normally achieved in three main ways: 1. via a verbal description – ‘give me a red’; 2. by cutting the original sample (assuming it is large enough) into many pieces and posting it out to the first recipient in the chain; 3. By matching the palette to the closest reference in the preferred colour atlas and quoting this to the supply base. Once the preferred method is determined, the colour requests are distributed with instructions on the retailer’s expectations in terms of physical performance parameters but, crucially, only with very broad-based
122
Total colour management in textiles
instructions on colour performance, i.e. ‘give me your best visual match’. A deadline for submission is also included which, again, is normally quite a short timeframe. This package of information works its way down the valueadded chain, with each player taking their piece of the original ‘target’ for reference and the dyeing, finishing and submission procedure begins. It is widely acknowledged that this is a very iterative process, necessitating multiple submissions, discussions, deliberations and confrontations, not to mention a very considerable amount of unnecessary time, cost and energy. So, why is this the case? The answer will vary from company to company and will be influenced by the method of colour target delivery; however, the major cause is undoubtedly the very subjective nature of the whole process. As will have been explained extensively elsewhere in this volume, our personal interpretation of ‘give me a red’ as a starting point will vary widely due to the many internal and external factors that affect our colour per ception. This applies equally to our assessment of the relative difference between two samples once a ‘lab-dip’ has been produced – an issue that is further compounded by a complete lack of information on the customer’s expectations in terms of a numerical tolerance for matching and the conditions and illuminants under which the colour is to be matched. Given these multiple factors and considerations, and the absence of any guidelines to manage them, it should be of little surprise that this is a very frustrating and protracted process. The provision of a physical target rather than a verbal descriptor is a step forward but is a long way short of ideal. Many retailers do not understand that the starting point for a dye-mill to begin the dyeing process (using a standard computerised match prediction system) is either to measure the original target with a spectrophotometer or to input the reflectance data for the sample. If the physical target is so small that it cannot be measured when it eventually arrives at the mill, the operator has to resort to manual and subjective colour matching by eye, the serious ramifications of which are mentioned above. Even when the target is large enough to be measured, there are still many variables present in terms of the age and reliability of the spectrophotometer, the relative differences between different makes and models, the measurement technique and sample presentation of the operator and the environment in which it is measured. Additionally, targets are occasionally supplied on non-textile substrates such as paper and plastic, which present an additional complication in terms of both feasibility and stability, particularly when the supplier is required to match under more than one illuminant. The absence of clear and comprehensive instructions for matching from the customer also applies equally to this situation.
Effective colour communication from mind to market
123
In light of the above, it is little wonder that, during this initial sampling phase, each supplier potentially has an individual, independent target (which could vary widely from the original) and this issue becomes more serious when we enter into production. Typically, the approved lab-dip becomes the ‘working standard’ for production and, in taking this standard, we have already moved quite some way from the original target during this phase. Each supplier then has their own ‘working standard’, which is compounded further still during production – a scenario that is virtually impossible to manage within a reasonable timeframe. And it gets worse . . . Inadequate targets and guidelines are compounded by a general lack of knowledge and experience of both colour and production techniques (to varying degrees) at the retailer, the offshore office, the agent/vendor and the garment maker. Each of the aforementioned has their historical and specific area of expertise but typically this does not extend to colour and dyeing and finishing. General observations include the following. • The colour approval task is non-value added, though a necessary evil, and is therefore typically delegated to junior staff members (usually in buying positions) with limited experience and colour knowledge. Staff are rarely checked for colour blindness or colour discrimination skills and there is also quite a high staff turnover rate in such positions, resulting in a lack of consistency and focus. • Colour approval is also usually a decentralised and departmental function, causing a lack of cohesion, consistency and informed comment. (The lack of such experience can apply equally to design or other commercial personnel at various stages of the chain.) • Colour approval is performed on a very subjective basis – at best using a lighting cabinet, and at worst, at the desk under variable and various viewing conditions. • On the occasions when a lab-dip does fail, feedback is given to the supplier in very non-specific terms, i.e. ‘make it bit warmer’ – again very subjective and the supplier has little chance of making sense of this. He/she has to try and interpret the requirements of the customer on a ‘best guess’ basis. As was mentioned previously, the end-result of all of the above can be summarised as follows: • multiple submissions for approval; • frequent debates or confrontations often necessitating crisis trips for resolution;
124
Total colour management in textiles
• lead-time extension and production delays in a critical path with very little margin for error; • increased costs in terms of man–hours, time and courier fees; • the inevitability that off-shade colour may be accepted in order to achieve in-store deadlines, hence a deterioration of the product and brand. So what can be done to address these issues . . . ?
7.5
Best practice in communicating between design and production – human and technological considerations
The concept of instrumental colour management is not a new phenomena – far from it! The technology in its earliest form has existed in the supply chain for over 30 years and its application in the retail industry was pioneered by Marks & Spencer in the 1980s. Since this time, the use of the technology has exploded on a worldwide basis and is now common-place in the dyeing and finishing arena as well as becoming a much more common tool for the world’s leading retailers and brands. What is relatively new is the increase in the uptake and adoption of the technology by a larger number of major worldwide retailers who are realising that they need to re-assess and change their colour business models in order to compete and survive. At the same time, we are seeing increased utilisation of the latest digital and electronic colour communication tools by the earlier technology adopters in order to meet the demands of today’s fiercely competitive and fast fashion-driven market. When looking at the worldwide apparel retail market, we see a dramatic variation in the levels of technology uptake and, whilst many proactive retailers are already working their way along the technology life cycle from introduction through the modification and maturity stages, there are a significant number of retailers and brands that have yet to jump onboard. This chapter aims to mention all of the various tools and processes that are currently available with a view to suggesting a ‘best practice model for retail’ to which current and prospective users alike can aspire. As well as looking at the individual solutions, suggestions on potential applications for this technology will be examined, together with a summary of the potential associated benefits. At the same time and equally importantly, we will cover the associated human issues because, as we will see, the latest tools are only as good as the people managing and utilising them and the processes and structures that surround them.
Effective colour communication from mind to market
125
7.5.1 Palette development Technological opportunities In addition to the very manual ‘storyboard’ process described in section 7.3, other tools employed during the palette development phase include: • traditional entry level design packages (Photoshop, Freehand, etc.); • existing CAD solutions with additional functionality and scope; • physical colour order systems. Each of the above has its own merits and limitations and they will continue to be popular with retailers as an aid to palette development. However, one potential area for improvement is the ability to move away from a purely physical form of palette development, whereby a physical (usually printed) target is used as a reference, and one that provides RGB data only, which the supply chain find difficult to translate and work with. Alternative solutions offering ‘precise-colour on a calibrated screen’ design capability have been commercially available for the past 8 years or so and include the Envision (Imagemaster) product from Datacolor. More recent product offerings include Lectra’s Colour Management Module which, when used in combination with existing design solutions, offers the user calibrated monitor and printer technology, together with the ability to design the palette electronically and output the reflectance data for the palette, which can be electronically communicated and utilised by engineered standards providers and the supply chain itself. Such systems offer considerable added value over and above basic ICC profile monitor and printer calibration, as well as the potential for seamless integration of colour data between the design and QC functions. Here is a selection of practical applications currently in use. • Palettes are created electronically on-screen using a spectrophotometer, electronic colour libraries or in-built colour creation tools. • Assess the colour on ‘flat tiles’, on fabric swatches or on full garments. • Use as a selection and merchandising tool on screen and on printer. • Print out virtual lifesize garments. • Assess the colour under different illuminants. • Match trims and accessories electronically in advance. • Print out the colours in true calibrated colour for discussion meetings. • Output the colour in reflectance data format to standards provider, internal departments or supply chain. • Archive and retrieve colour standards and associated data electronically.
126
Total colour management in textiles
Through the use of this technology, the following commercial benefits can accrue. • Palettes can be designed in a fraction of the time. • A much more flexible and creative tool is available – no need to compromise on colour due to ‘best can find’. • Better and more informed decisions can be made earlier in the process. • Feedback on colour feasibility can be gained much earlier without the time and cost associated with physical sampling. • Colours that don’t work can be identified and numbers of requested samples can be reduced. • Massive sampling costs/time for swatches and garments can be reduced. • Rationalisation of palette – colours can be shared between departments and historic standards and colour libraries can also be searched to prevent duplication and reduce cost and time. • A much more exact science is available – meaningful palettes can be delivered to QC and the supply chain. • Communication can be improved with the supply chain and approval rates improved. Human considerations and caveats One of the biggest challenges that retailers face in this particular area is making design teams aware of the effect that misinformed colour decisions have on their colleagues in production and onwards into their supply chain. Whilst it is clear that design will accrue many benefits via the adoption of the latest digital tools, they must appreciate that as much benefit, if not more, will go to their colleagues in the product development area. They must not see the completion of the creative physical palette as their responsibility cut-off point. For this to be understood and implemented, there needs to be a ‘buy-in’ at the highest level, that the collective corporate benefit is more important than that of the individual business unit. The shift to working electronically with design in this way can be a significant culture shock and realisation of this at an early stage will help address resistance and concerns further down the line. It does require an open-minded and flexible working approach for all concerned if the company is to overcome the typical (sometimes unwritten) objections of: • ‘I will always need something to touch and feel.’ • ‘Designing colours on screen is a degradation of my creative flair.’
Effective colour communication from mind to market
127
• ‘I can’t trust what I see on the monitor.’ • ‘I am not being sufficiently creative if I am re-using colours we have used before?’ And so on . . . A phased period of implementation is always suggested to ‘bridge the gap’ and to get people comfortable with a new way of working; a semi-technical person to forge closer links between design and production always helps during this transitory period. The change requires commitment from all concerned and a stark realisation that, when we talk about reducing colour associated lead-times, there is as much change required in design as there is in production. The rewards are more than worth the effort and change is essential to survive – it is good for a business to be ‘design led’ but being ‘design controlled’ is another matter completely!
7.6
Creating the standard
For companies with established colour management programmes, once the palette has been given final approval by the design team, it is typically ‘handed off’ to quality control/product development to transform this into a set of ‘engineered standards’ for onward distribution to the supply chain. But what is an engineered standard? An engineered standard is a dyed piece of fabric, usually cotton, supplied with a set of reflectance data for that particular colour mounted on a piece of card and this is a service supplied by dedicated companies or the major dyestuff suppliers. The standard should clearly state that the reflectance data must be used as the starting point for the dyeing process either via keyboard input or via electronic importing and that the physical swatch is there as a visual reference only and under no circumstances should the fabric be re-measured for matching. Instructions for matching, handling and storage are normally also provided as is a recipe for the colour (though this is more often provided upon request if a supplier is experiencing difficulties matching the shade using his normal dyestuff combinations). The starting point for this process is receipt by the standards provider of either a physical reference supplied by the customer or an electronic palette sent in via email from the aforementioned design or QC systems. Submissions are sent back to the customer again either physically or electronically until approval is given, upon which sufficient lengths of the fabric are dyed and conditioned. ‘Master’ measurements are then made either on the customer’s or standards supplier’s spectrophotometer and the resultant reflectance data is then transferred on to the printed standard cards and stored for electronic delivery.
128
Total colour management in textiles
7.6.1 Commercial benefits The use of engineered standards has a number of associated benefits in its application: • They are a supremely objective and accurate definition of the original target colour for matching. • They enable every player in the supply chain to begin colour matching at exactly the same point by instructing the use of reflectance data only as the standard. • Via dialogue with the standards supplier and by utilising their experience in colour formulation, colour targets/recipes are verified for feasibility in terms both of performance parameters and colour matching in multiple illuminants on multiple substrates. • In this way, the retailer can be assured that the target is achievable and prevent ‘best can do’ responses from the supply chain. • Greater utilisation of electronic tools to specify, manage and communicate colour during this process presents major speed, cost and quality opportunities.
7.6.2 Human considerations and caveats The development of engineered standards undoubtedly presents an additional financial, and also time, consideration to the design environment, who have traditionally employed a ‘least cost’ approach to this subject. However, as mentioned at length in previous chapters, the root cause of all the major issues associated with multiple colour submissions and major time delays resides with the inaccurate specification of the target at the beginning of the process. The additional cost and time associated with the use of engineered standards is offset many times over by the dramatic reduction in the number of supplier submissions, together with the associated lead-time, cost and quality improvements. To realise these benefits, it is absolutely essential that a ‘total process’ viewpoint is foreseen and adopted by the retailer, rather than a potentially divisive departmental approach which could block such initiatives and opportunities.
7.6.3 Delivery of the standards to the supply chain Once sufficient quantities of the engineered standards have been produced, the retailer is then challenged with the onward distribution. This will vary between retailers but would typically involve offshore sourcing offices, agents/vendors, garment makers, dye mills and trim and accessory manu-
Effective colour communication from mind to market
129
facturers. Each player at each stage will want to maintain a reference and, particularly at the agent or garment maker level, additional new players – usually dyemills who are not usually visible to the retailer – will be introduced, necessitating further copies of the standard. The most important recipient in all of this is the company that is physically dyeing the fabric, as it is the reflectance data that is the precursor to the match prediction and entire sampling process. It goes without saying that it is equally important for all associated players to refer to the same original standard throughout the sampling and production process and this should be the original standard not the first approved lab-dip. Given the not insignificant number of companies involved in the process, the associated distribution costs of physical delivery, as well as the associated time lag, are considerable. Some companies choose to accept and absorb this cost, yet there are other retailers who have been actively reviewing and amending their distribution methods in order to minimise these issues. The following options are utilised to varying degrees. 1. Suppliers are charged for each physical standard they order. 2. Electronic or digital standards (reflectance data only) are delivered to the supply chain either via email or via the retailer’s website (via password access) on a free of charge basis. Physical standards are chargeable. 3. The entire standards production and distribution service is ‘outsourced’ to the standards provider and the supply chain download or purchase standards directly. Human considerations and caveats The ‘passing-on’ of this cost from retailer to supply chain is potentially a subject of debate. However, in practical terms, it would appear that the supply chain would prefer to absorb this cost in the knowledge that they are receiving a true and fair standard that they are capable of matching, rather than a moving and unrealistic subjective target necessitating multiple submits and unnecessary time and cost. ‘Is an engineered standard all that I need to get good colour matching?’ The answer to the above question is a resounding no! Engineered standards are only as good as the instructions and customer expectations that go with them. An absolute prerequisite to any retailer beginning a colour management programme is a clear and comprehensive set of instructions, usually in the form of a dedicated colour quality manual, explaining in detail their exact requirements. Amongst others, this would typically include instrumental tolerances
130
Total colour management in textiles
and settings, sample presentation and measurement techniques and sample submission procedures. In the absence of this, and in the absence of a clear understanding of the messages by the supply chain, chaos will reign!
7.7
Colour approval – where is it done?
One potential best-case scenario answer to this question would be: ‘As close as possible to production so that corrective action can be taken as quickly as possible as and when things go wrong.’ However, retailers vary considerably in their sourcing operations and structures and, as a direct result of this, we see very different approaches to colour approval in the retail market. In section 7.4 we studied in detail the major disadvantages associated with decentralised or departmental colour approval on a purely visual assessment basis. In this section, we will examine the structural, as well as the process issues, associated with best practice alternatives. Scenario A: An instrumental QC system utilised by many departments at the same Head Office location. Summary A much more accurate and repeatable process than pure visual assessment but there is potential for inconsistent decisions due to the large number of users, plus significant logistical and training issues. Colour approval is still not a high priority task plus suppliers are normally still sending physical samples long distances with associated time lags. Scenario B: A centralised colour department with an instrumental QC system Summary A more accurate, repeatable and consistent process due to focused and production/colour experienced users. Borderline issues are discussed and resolved more easily and colour approval is the primary role of the department. However, potentially suppliers are still sending physical samples long distances with associated time lags. Scenario C: Colour approval is executed at offshore offices in close proximity to the supply chain Summary Though this requires additional capital investment and training, all of the above benefits prevail, plus the ability to significantly improve
Effective colour communication from mind to market
131
lead times due to proximity to supply chain and ability to respond quickly when issues occur. Many retailers continually strive to make their suppliers more ‘colour accountable’ and this is a noble cause. The most important colour decision is the one that is done ‘at source’ and the retailers have a right to expect that submissions that do not meet predefined criteria should at least be minimised and, at best, eradicated completely. This does require time and is very much a two-way process with a number of caveats; however, it is more than do-able given sufficient focus and emphasis, and the rewards speak for themselves.
7.8
Colour approval – how is it done?
A growing number of the world’s larger retailers have recognised the many deficiencies associated with the purely visual method of colour evaluation and have taken steps to implement instrumental colour measurement within their process. The extent of implementation varies considerably and this is a constantly evolving process with equal consideration given to the logistical issues (i.e. where the decision is taken as described above) as well as the technological consideration of how the decision is taken. This section will examine the steps to progression and consider the existing technology currently utilised by the world’s leading retailers and brands.
7.8.1 Instrumental evaluation of physical submits The first step is the implementation of an instrumental QC system to measure and assess physical submits from the supply chain. The original standards are stored and retrieved electronically and the lab-dip or production sample is measured on to the system and assessed for compliance against the established numeric tolerance. Visual evaluation of the submits in a lighting cabinet is also performed and a pass/fail decision results, which is subsequently communicated to the supplier. On occasions where the submit fails, guidelines for corrective action are provided in specific objective production-based terminology. Assuming good levels of operator training, engineered standards and a comprehensive colour manual, this makes for a very accurate and consistent process; albeit one which is limited by the requirement for the samples to be supplied and returned on a physical basis, i.e. by courier post with the associated time and cost implications.
132
Total colour management in textiles
7.8.2 Instrumental evaluation of electronic submits in QC programmes The worldwide application and acceptance of email and Internet connectivity has opened up significant opportunities in the field of electronic communication at a time when the distance between the retailer and their supply base is becoming wider and wider. None more so than in the field of colour where the need to physically handle and assess colour submissions is negated by the ability of the dyemill to make a measurement and send it to any designated location worldwide in a matter of seconds. This is admittedly an oversimplified version of events and is subject to a number of caveats such as: • ensuring that the sample has been measured in the correct manner and under the correct conditions – usually effected by way of supplier audits or accreditations; • ensuring correct levels of inter-instrument agreement between supplier and retailer such that the report the supplier sees in Sri Lanka corresponds with the report that the retailer see in Europe. • making sure that the email attachment that has been sent can be opened and utilised by the recipient system in the same file format. • and finally, getting people to trust the remote measurement that has been made in place of a physical sample that they have historically insisted upon and sought comfort with. Though these are legitimate issues demanding due consideration, they are easily addressed, given an open mind and organisational focus. Many of the early adopting apparel brands have been enjoying the massive lead time, cost and quality benefits that are available through such an electronic programme for some time. Moreover, given the ever-shortening development timelines that retailers are working to nowadays, such programmes are an absolute ‘must-have’ if they are to maintain their future competitiveness.
7.8.3 Instrumental evaluation of electronic submits using ‘true colour on screen’ solutions A further enhancement, and an increasingly popular addition to existing electronic communication packages, has been the introduction of ‘precise colour on calibrated computer monitors’ systems such as Datacolor’s ‘Envision’ product. This high-end application allows the user to view colour on screen in precise colour, thereby facilitating an on-screen pass/fail decision of a remote measurement, together with the ability to assess the effect of colour in context through texture mapping. Such a process can virtually
Effective colour communication from mind to market
133
eliminate the need for physical submissions and provides considerable added value over and above the purely numeric approach available through traditional QC packages.
7.8.4 Self approval by the supply base Selected retailers who are already utilising the latest digital tools have taken their application to the next stage by allowing selected ‘accredited’ suppliers to self-approve their own lab-dip and production samples when they fall within documented tolerances – not so much ‘a leap of faith’ but more a realisation that, given the correct systems and procedures, unnecessary sampling can be avoided and the resultant savings and benefits are significant and easily achievable.
7.9
Electronic colour communication programmes – associated considerations and options
7.9.1 Supplier accreditation An accreditation or supplier audit is widely regarded as a prerequisite and an insurance policy for those retailers who are managing and communicating colour electronically. The phrase ‘garbage in garbage out’ is very relevant here – colour communication packages work extremely well but they are rendered useless if the information being communicated is unreliable. Standard accreditations/audits typically test and verify the instrument, making the measurement together with the operator’s ability to perform colour evaluation, in line with the retailer’s specific procedures. Additionally, the environment in which the system and operator are located are also audited for compliance. Once complete the retailer can be assured that remote measurements from that supplier can be trusted and assessed using electronic procedures.
7.9.2 Spectrophotometer monitoring and correlation A more recent introduction to colour management programmes has been software to allow the frequent performance monitoring of spectrophoto meters over and above that received during the annual service visit. It is widely acknowledged that spectrophotometers drift with time and this phenomenon increases as the instrument ages. There are software packages available that regularly check the performance of the instrument in order to prevent this from occurring and to ensure the integrity of measured and communicated data throughout electronic programmes.
134
Total colour management in textiles
Such software can also be utilised to correlate or profile spectrophoto meters back to the original master instrument or to another selected master reference in order to supposedly offer improved inter-instrument agreement. However, the improvement available via the adoption of such an approach on textile samples is considered to be minimal and is often insufficient to justify the added complication of managing correlated and uncorrelated data in a retailer’s supply chain. (Such packages include Datacolor’s Maestro Product and GMB’s Netprofiler.)
7.10 Electronic tracking and reporting packages As the use of electronic colour programmes continues to increase, so does the amount of data that is produced by such systems. Operators are challenged with the ongoing maintenance, archiving, prioritisation and arrangement of this data into a format that allows them to concentrate on their respective departmental responsibilities. This is at the same time as having clear visibility of the colour process, together with an insight into vendor performance before, during and after the current season’s submissions. In direct response to this, software products exist (including Datacolor’s Track product, GMB’s Netpalette and Ewarna’s X-Match) that concentrate specifically on the colour critical path, enabling the user to prioritise, track and report on individual colours and to filter colour selections based upon the respective commercial detail such as department, season, supplier, etc. Additionally. the user can obtain a ‘snapshot’ approach of any selected colour submission status without necessarily getting immersed in technical colour detail and thereafter run management reports to track vendor performance over selected periods.
7.11 Future trends and conclusion With some exceptions, the general uptake and implementation of colour management technology in the US retail sector is some way ahead of the European sector, though there are encouraging signs that the major players in this market are finally waking up to the inevitable. The reasons behind this are not easy to diagnose accurately, though it is suspected that the size and scope of the American retailers alone was sufficient to justify significant and specific focus and investment in this area at an earlier stage. However, whilst most definitely on the increase, there is considerable scope for improvement worldwide in the implementation of electronic rather than physical programmes. There are really no justifiable technical barriers to implementation any longer – the technology is mature and proven and continues to evolve in line with market demand, the supply
Effective colour communication from mind to market
135
chain has largely ‘bought in’ and endorsed this approach and the benefits could not be more evident. It is foreseen that, in the future, systems are likely to become more automated, user friendly, design integrated and working in one universal language, but this should be viewed as a bonus not as a reason to wait or to procrastinate. The barriers to implementation or initial adoption are mainly human considerations: • maybe a lack of vision; • maybe a lack of understanding of colour theory in retail; • maybe a misunderstanding or misinterpretation of these pressing issues; • maybe a reluctance to realign structures and attitudes in line with market forces; • maybe an unrealistic approach to or expectation regarding capital investment. The phrase ‘the quick and the dead’ is as relevant now as it ever has been – the pioneers are already there, hopefully it is not too late for the rest – maybe!
8 Controlling colourant formulation J H X I N, The Hong Kong Polytechnic University, Hong Kong
8.1
Introduction
Formulating colourant recipes to match target colours is not an easy task. Manual colour prediction often uses a trial and error method, for which the experience of the colourist is essential. The majority of colour matches also require colours to be matched not only under daylight, but also under other artificial light sources, typically cool white fluorescent (CWF) and incandescent sources. A previous recipe archive is very useful for manual colour matching. A colourist would firstly search the previous recipe archive to find out the closest colour matching the target (or standard) and then make some adjustment to the recipe if the recipe colour is not the exact match to the target. However, this trial and error process can be lengthy and arduous even for a professional colourist. Computer colourant formulation is an alternative method. The commercial application of computer colour recipe formulation in textiles was first disclosed by Alderson et al. in 1961.1 The most well-known algorithms for colour recipe formulation are the two proposed by Allen, one for the singleconstant and the other for two-constant formulation.2,3 In recent years, computer colourant formulation has been widely applied, especially when supplying the coloured articles to companies with global sourcing practice, thanks to the formidable advance of the digital computer, especially the personal computer. This greatly improves the lead-time for the colour matching, especially when experienced colourists are not available. It has become a necessity for a modern dyehouse to install a computer colourant formulation system. The flowchart of the coloured goods production process employing a colourant formulation system is illustrated in Fig. 8.1. The spectral reflectance of a target colour is first measured using a spectrophotometer, and the colourant concentrations are computed using the colourant formulation system. These concentrations are used in the colouration process to produce the matching coloured goods. If the colour difference between the target 136
Controlling colourant formulation Measure Target colour
Spectral reflectance
Colourant formulation system
Feed back
Colour difference
Dye
Colourant concentrations
137
Measure Produced colour
Spectral reflectance
8.1 Flow chart of colour production process using colour recipe prediction.
and the produced colour is too large to be accepted, the recipe should be corrected. As the dyeing and colour quality control processes are usually costly and time consuming, it is important and still very challenging for any colourant formulation system to be ‘right first time’ in colour matching with a generally acceptable colour difference.
8.2
Colourant recipe formulation
Colour matching to a target depends not only on the formulation system but also on the accuracy of the recipe preparation, the repeatability of the dyeing process and the colour measurement process. There is a need for quality control at each step in the colouration and measuring processes. The core of the formulation system is based on the theory developed by Kubelka and Munk (K–M theory).4,5 As the theory involves quite a few assumptions that are necessary for the prediction of the colourant concentrations, the real dyes and pigments do not ideally conform to these assumptions. Therefore, the K–M theory is only an approximation to the real pigmented systems. Nevertheless, methods to improve the prediction accuracy are available in certain commercial colourant formulation systems. In addition, methods based on artificial intelligence can also be employed in recipe formulation. The scope of this chapter is delimited to the colourant formulation of the textile materials. The colourant formulation for plastics and paints can be found elsewhere.9
8.2.1 Kubelka–Munk Theory When light passes through a pigmented layer, two things will happen: part of the light is absorbed by the pigments and the medium in which pigments
138
Total colour management in textiles
are dispersed; part of the light is scattered by the pigments. K–M theory is based on earlier research by Schuster,6 who studied the weakening of light from the stars by scattering and absorption before reaching the observer. However, pigmented systems are more complicated because of the interaction between pigments, especially when the pigment loading is high. Therefore, K–M theory has to simplify the real situation in order to derive any useful mathematical equations. In practical colourant formulation, we are restricted to the following assumptions: 1. diffuse illumination and diffuse viewing without polarisation of the light, 2. a plane parallel surface of the object without light losses at the edges, 3. the unit layers of the material are homogeneous and isotropic, 4. the theory does not account for the presence of large particles, agglomeration or orientation of the particle in the layer, and 5. optical contact with the next layer. Figure 8.2 shows a simplified case of light passing through a very thin layer of thickness dx. We consider the downwards and upwards components of the incident light separately, and assume that the absorption coefficient is denoted by K and the scattering coefficient by S. Then, the downward flux (intensity I) is given as dI = -KI dx - SI dx + SJ dx
[8.1]
and the upward flux (intensity J) is given as dJ = -KJ dx - SJ dx + SI dx
[8.2]
We note that light loss through the edges is neglected. The internal reflection that exists when leaving an optically denser medium, e.g. light leaving from textile fibres, which is optically denser than air, is also neglected by the theory. Other assumptions such as uniform distribution of the pigments, etc., may also differ from the real situation.
I
I
K
S
I = Downward flux
X dx
K
S J = Upward flux
J
J
8.2 Schematic diagram of the simplified version of light passing through a finite colourant layer used in K–M theory.
Controlling colourant formulation
139
A series of the solutions can be obtained by solving the differential equations (eqn 8.1) and (eqn 8.2). These solutions and the examples of how they can be used are given by Judd and Wyszecki.8 In textile formulation, the solution can be even more simplified. Firstly, textile fabrics can be considered opaque because the measurements are made at sufficient thickness of the fabric swatches. Secondly, the scattering coefficients of the textile dyes are negligible when compared with the substrate in which they are dissolved, so that only the scattering coefficient of the substrate needs to be considered. Thirdly, it can be assumed that the total absorption and scattering is the summation of those from each individual colourant, i.e. the absorption and scattering are additive. Hence, we obtain the following equation: R• = 1 + (K/S) - [(K/S)2 + 2(K/S)]1/2
[8.3]
and its inverse
K S=
(1 − R∞ )
2
[8.4]
2 R∞
where R• is the reflectance of a colour sample of optically infinite thickness. In K–M theory, the absorption and scattering coefficients of a colour sample can be further represented using absorption and scattering coefficients of the individual pigments or dyes: (K/S)mixture = (K/S)1 + . . . + (K/S)n = K1/Ssub + . . . + Kn/Ssub + Ksub/Ssub
[8.5]
where K1 to Kn are the absorption coefficients of the dyes, Ksub and Ssub are the absorption and scattering coefficients of the substrate, respectively. From (eqn 8.3)–(eqn 8.5), it can be seen that, for a dye system, the reflectance of the coloured sample can be predicted from only the ratio of the absorption coefficient of each dye in the mixture and the scattering coefficient of the substrate, i.e. Kn/Ssub. The various K/Ssub ratios may be considered as a single constant, which are commonly called ‘absorption coefficients’ in textile dye formulation, and the theory is therefore known as singleconstant K–M theory.2 The absorption coefficient for a colourant is related to its concentration and, in many cases, the relationship is a non-linear one with the exhaustion decreasing with an increase in dye concentration. In practice, a range of concentrations will be used to obtain a so-called calibration database by dyeing, and an absorption coefficient is correlated to the corresponding dye concentration in the dyebath. Using this calibration database, the absorption coefficient at a given concentration can easily be found.
140
Total colour management in textiles
8.2.2 Calibration database As described above, a calibration database is necessary for a formulation system to establish the relation between the K/S values and the concentrations for each dye. Because formulations are based on the database, its preparation should be carried out using exactly the same substrate and dyeing conditions as those of the formulated recipes to be subsequently used. For example, if in the calibration database, plain woven cotton fabric is used as the substrate, the formulated recipe should be used only for the dyeing of the plain woven cotton fabric. Any change in substrate, whether it is the reflectance or the absorbance property, would be expected to give inaccurate results. The same is true for other conditions such as dyebath pH, temperature, liquor ratio, etc. One difficulty in that aspect is that laboratory dyeings that are used for preparing the database do not always represent the bulk production dyeings and there is often no clear relation between laboratory and bulk dyeings. Since colour matching relies on the calibration database, accurate preparation of the database is essential. Any error in the database can subsequently affect the formulation. Good computer formulation systems should be able to detect abnormal points in the database preparation only if these points are scarce in comparison with the rest of the normal points. The database consists of the calibration dyeings of each dye used in the system. Usually, six or more calibration dyeings are required for each dye in order to cover the concentration level of, say, from 0.05% to 2.0% of the weight of the fabric. The highest concentration level should comply with the recommendations of the dyestuff manufacturer. Additional errors may be introduced if a formulated recipe uses a dye at much below or much above the concentration range of its calibration dyeing. An example of the acid dye on wool is given below. Figure 8.3 is the reflectance data of calibration samples at different concentrations plotted against wavelength. The K/S value of each calibration sample can be calculated according to (eqn 8.4), and its distribution against wavelength is plotted in Fig. 8.4. The K/S values against concentration at a maximum absorption wavelength of 520 nm are shown in Fig. 8.5. A non-linear relationship exists between the K/S value and the concentration of the colour sample. In commercial computer colourant formulation systems, the K/S values and their corresponding concentrations for all calibration dyeings are stored. The recipe formulation function uses these values for linear interpolation.8,9 For example, at a concentration ctarget, the K/S value of the target is given by: (K/S)target = (K/S)low + B[(K/S)high - (K/S)low]
[8.6]
Controlling colourant formulation 100
R% vs.
141
Wavelength
79 59 R% 39 19 0
400
500
nm
600
700
8.3 Reflectance of dye Ramazol Red 3BS at different concentrations plotted against wavelength.
K/S vs.
20
Wavelength
16 12 K/S 8 4 0
400
500
nm
600
700
8.4 K/S values of dye Ramazol Red 3BS at different concentrations plotted against wavelength.
where B = (ctarget - clow)/(chigh - clow)
[8.7]
and c is the dye concentration and the subscripts low and high denote the two neighbouring points in the calibration data. In some cases, the use of second-order or third-order polynomial equations and curve-fitting techniques to find the correlation between the K/S
142
Total colour management in textiles 20
K/S vs.
Concentration
wavelength = 520
16 12 K/S 8 4 0
0.0
8.0
16.0
Conc.
24.0
32.0
40
8.5 K/S values of the six Ramazol Red 3BS calibration data plotted against the dye concentration (g/l) at 520 nm.
values and concentration can smooth out certain fluctuation of the calibration dyeings to overcome dyeing errors, without the need for re-dyeing. However, if a calibration dyeing sample is far from the trend of the rest of the calibration samples, it should be re-dyed. Equation (8.8) below is an example of a third-order polynomial equation: (K/S)target = a1c + a2c2 + a3c3 + Ksub/Ssub
[8.8]
where a1, a2 and a3 are the coefficients. The coefficients are optimised from the calibration database and are stored by the system for use in recipe formulation. The K/S value for a given concentration can be found using (eqn 8.8). Some early commercial systems may have taken the physical–chemical approach, which models the relationship between K/S and dye-in-fibre using the Langmuir isotherm for dye absorption, which results in a relationship with two coefficients.10
8.2.3 Recipe formulation In this section, we discuss the colour recipe formulation of nonfluorescent textile samples, as the K–M theory cannot be applied to fluorescent materials. It is known that recipe formulation can be carried out using colorimetric and spectrophotometric matching algorithms.9 Spectrophotometric matching algorithms minimise the reflectance difference between the target and prediction:
Controlling colourant formulation
∑ [R
2
λ ,target
λ
− Rλ ,prediction ] → min
143 [8.9]
where l represents wavelengths from 400 to 700 nm with a 20 nm interval. According to (eqn 8.4) and (eqn 8.9), we have the equation at wavelength l: (K/S)l,target = (K/S)sub,l + (K/S)l,l + . . . + (K/S)n,l
[8.10]
where n is the number of colourants in a mixture, which is usually equal to 3. In the case of a 20 nm interval, there are 16 simultaneous equations. This over-determined system can be solved using the least-squares method. Though the spectrophotometric algorithm is very straightforward, it is restricted to a non-metameric match. In other words, the optical properties of the substrate and the dyes used by both the target and the match should be spectrally very similar, otherwise the algorithm will often give poor matching results. Another drawback is that human eyes are more sensitive to lights at certain wavelengths than at others in the visual spectrum. Thus, the reflectance differences at the wavelengths that are more sensitive to human eyes are more important in colour matching than others. The spectrophotometric curve matching does not take this into consideration. Attempts can be made to give carefully selected weights at different wavelengths11:
∑w λ
2 λ
2
[ Rλ ,target − Rλ ,prediction ] → min
[8.11]
where weights wl reflect the importance of different wavelengths for visual perception. It is reported that the spectrophotometric strategy is not as successful in diminishing colour difference for a particular illuminant as the colorimetric strategy.11 Nevertheless, the spectrophotometric strategy does produce more ‘balanced’ colour differences between different illuminants, and can therefore be used to reduce metamerism. In most commercial systems, the colorimetric matching algorithm to minimise DX, DY, and DZ has now been universally adopted. This algorithm is based on the strategy (DX, DY, DZ) Æ (0, 0, 0)
[8.12]
Colorimetric matching is very effective because it minimises the colour difference directly. Smaller differences in DX, DY and DZ result in closer colour match. The drawback of colorimetric matching is that a match achieved under a particular illuminant (e.g. D65) may not be a match under another illuminant (e.g. A), especially when different types of dyes are involved. This type of match is the so-called metameric match, i.e. the reflectance curve of the predicted sample is very likely different from that
144
Total colour management in textiles
of the standard and, when plotted, these two reflectance curves have at least three cross-over points. We show below how a three-dye recipe is formulated using colorimetric matching according to the following steps. 1. Select the starting concentrations c1, c2 and c3. 2. Use (eqn 8.6) or (eqn 8.8) (depending on how the system establishes the relation between K/S values and the concentrations) to obtain the K/S of each colourant in the mixture. 3. Use (eqn 8.5) to obtain the K/S value of the mixture assuming these colourants are additive. 4. Use (eqn 8.3) to calculate the reflectance of the predicted sample. The tristimulus values X, Y and Z can, thus, be obtained using the illuminant required for the matching. 5. Calculate the colour difference between the standard and the match and if the colour difference is smaller than the tolerance, the recipe is found. 6. If the colour difference is larger than the tolerance, the following matrix is devised: ∂X ∂c1 ∆X ∆Y = ∂Y ∂c ∆Z 1 ∂Z ∂c 1
∂X ∂c2 ∂Y ∂c2 ∂Z ∂c2
∂X ∂c3 ∆c1 ∂Y ∆c2 ∂c3 ∆c3 ∂Z ∂c3
[8.13]
where ∂X ∂Rλ = ∑ Eλ xλ , ∂ci ∂ci ∂Y ∂Rλ = ∑ Eλ yλ , ∂ci ∂ci
[8.14]
∂Z ∂Rλ = ∑ Eλ zλ ∂ci ∂ci and in (eqn 8.14): ∂ [ ( K S )λ ] ∂Rλ dRλ = ∂ci d [ ( K S )λ ] ∂ci
[8.15]
Controlling colourant formulation
145
where El is the illuminant spectral power distribution, x¯l, y¯l, and z¯l are the CIE standard observer functions, and ∂[(K/S)l]/∂ci can be obtained from (eqn 8.8) or by linear interpolation using (eqn 8.6), and
d [( K S )λ ] Rλ2 − 1 = dRλ 2 Rλ2
[8.16]
dRλ 2 R2 = 2 λ d [( K S )λ ] Rλ − 1
[8.17]
and thus
Hence, in (eqn 8.13) all the items are known and the differences for each colourant can be calculated by matrix inversion: ∂c1 ∂X ∆c1 ∂c2 ∆c2 = ∂X ∆c3 ∂c3 ∂X
∂c1 ∂Y ∂c2 ∂Y ∂c3 ∂Y
∂c1 ∂Z ∆X ∂c2 ∆Y ∂Z ∆Z ∂c3 ∂Z
[8.18]
The new corrected recipe is then c1,new = c1,original + ∆c1 c2,new = c2,original + ∆c2
[8.19]
c3,new = c3,original + ∆c3 Steps 1 to 6 are repeated until the colour difference between the standard and the prediction is within the tolerance limit. The number of iterations required to obtain the desired recipe depends on the effectiveness of the starting concentration provided in Step 1. If the reflectance of the target is known, the starting recipe can be determined by Allen’s2 method, where the matrix expression of the starting recipe is given as C = (TEDA)-1 TED(F - S)
[8.20]
where C is the concentration matrix, T is the matrix composed of CIE standard observer functions, E is composed of the spectral power dis tribution of the matching illuminant, D is composed of the element dRl/d[(K/S)l], A is composed of ∂[(K/S)l]/∂ci for each colourant, F is com-
146
Total colour management in textiles
posed of the K/S values of the target and S is composed of the K/S values of the substrate. The starting concentration determined by this method is often very close to the target. Usually, only several iterations are required to bring the colour difference within the tolerance limit. Commercial recipe formulation systems have the option to sort the predicted recipes according to the cost and the metamerism under a secondary illuminant.
8.2.4 Recipe correction After recipe formulation, a new colour sample can be produced according to the colourant concentrations suggested. Because of the influence of various variables in the dyeing process, the produced sample may not be acceptable and a recipe correction process may be needed. Laboratory correction Laboratory correction gives a fresh recipe according to what was obtained by previous dyeing. The new concentration calculation is as follows: Cnew = Cpredicted ¥ Cused/Cbatch
[8.21]
Cnew = Cpredicted + Cuse - Cbatch
[8.22]
or where Cnew is the corrected recipe, Cpredicted is the predicted recipe for the standard, Cused is the recipe used in dyeing, which may be equal to Cpredicted, and Cbatch is the recipe back-predicted for the batch dyeing result. Correction method (eqn 8.21) is called weighted (or ratio) correction and method (eqn 8.22) is called additive correction. We note that the use of (eqn 8.21) and (eqn 8.22) above would give relatively similar results for correcting small colour differences. However, for a large colour difference, (eqn 8.22) may give wrong results and the use of (eqn 8.21) is recommended. If a large colour difference exists between standard and batch, correction accuracy is limited. On the other hand, for a small colour difference, the correction accuracy is limited by the repeatability of the dyeing process. Production correction Production correction predicts the additional amount of dyes to be added to the dyeing bath: Cadd = Cnew - Cused [8.23] where Cnew is calculated according to the laboratory correction via either weighted or additive methods. There is no ‘bleed-off’ included in the calculation. It may be added if bleeding is a problem.
Controlling colourant formulation
147
If a batch is already too dark compared with a standard, this correction for exhaustion dyeing will fail, meaning that the absorbed dyes should be stripped before correction. For continuous dyeing, dilutents need to be added to dilute the dye liquor. However, for a slight dark colour, the production correction for exhaustion dyeing should correct hue difference and try to reduce the overall colour difference.
8.3
Improvement of the formulation accuracy
In most current colour formulation systems, K–M theory is applied for recipe prediction. However, there are many situations in which the K–M recipe prediction cannot be used successfully. For example, the prediction result of fluorescent colourants is rather poor. This is mainly due to a breakdown of the K–M assumptions and the failure of the model to describe the optical behaviour of the colourants accurately. Moreover, it is necessary to prepare an accurate calibration database, which would greatly affect the prediction performance. However, sometimes an accurate database cannot be achieved because the samples prepared in the laboratory may not correlate well with the actual dyehouse production samples. Even if the accurate samples can be prepared, a new calibration database will need to be produced if there are changes in dyes and substrates, but for many dyehouses this is prohibitively costly and time consuming. Consequently, the K–M model has many limitations, which make the prediction unreliable in some cases. Drawbacks of the K–M model have motivated the colourists to develop other methods for recipe formulation. As we know, it is possible for a professional colourist to predict the colourant concentrations with high accuracy even for fairly complex situations without being aware of the K–M theory. Colourists accumulate experience of the behaviour of colourants and have the ability to predict the recipe of a new colour shade from previous ones. Therefore, it is possible to use artificial intelligence techniques such as neural networks to mimic the behaviour of professional colourists.
8.3.1 Artificial neural network12 The human brain is a neural network itself. The basic unit of the brain is the neuron, which is a special nerve cell and, although simple creatures may only possess a few thousand neurons, the human brain contains approximately 1012 neuron cells. What makes neurons different from other cells in the body is the way in which they are connected to each other. A neuron receives information from other neurons using its dendrites and sends information to other neurons using its axons. The synapse is the point where the axon of one cell meets the dendrite of another. One of the most impor-
148
Total colour management in textiles
tant properties of the brain that we would like to mimic in computer systems is its ability to learn. The neural network can be considered as a black box that is connected to the world by a series of inputs, which interacts with the world via a series of outputs. Inside the black box, a network system performs a mapping function between its input and output. The units of the neural network are connected by weights that can be modified and they perform a similar function to the connection structure in the brain. Figure 8.6 shows a schematic diagram of the function of a single unit. The unit receives input Ii from n other units. The total input to the unit is the weighted sum of n inputs, that is the sum of each of the n inputs multiplied by the respective value of the weighted connections wi. The output O of the unit is then given by some transfer function f(·) of its weighted input. Thus, mathematically it can be written as n
O = f ∑ I i wi i =1
[8.24]
A typical non-linear transfer function is the sigmoid function f ( x ) =
1 1+e− x
[8.25]
where x is the input of the unit. Non-linear transfer functions are usually due to the function’s ability to approximate complex mapping between input and output vectors, while linear transfer functions are sometimes used for units in the output layer. There are many types of neural networks. The complexity of the neural network is determined by the problem to be solved. One of the simplest and most successful networks is the multi-layer perception (MLP). The MLP consists of simple processing units in layers. Each neuron receives an input and modifies it in a simple way to produce an output. The neural network will contain one or more hidden layers between the input and output layers, and the number of units in each layer also depends on the complexity of the problem. A simple structure of a neural network is shown in Fig. 8.7, which consists of an input layer, a hidden layer, and an output layer. Before a neural network can be used to solve a given task, it must first be trained using known pairs of input and output vectors. The training process consists of adjusting the weights in the network so that, when a certain input is presented in the input layer, the output layer will produce
Controlling colourant formulation
149
I1 w1 w2
I2 . . .
ÂIi w
f (◊)
O
i
wn
In
8.6 The generalised unit of a neural network.
Input layer
Hidden layer
Output layer
8.7 A simplified structure of a neural network.
the desired output. Samples of input–output pairs are presented to the input and output layers of the network, respectively. The input vector is used to generate an output for each unit in the network, layer by layer, until an output is produced in the output layer. The error between the target output and the actual output is then calculated, and the network adjusts the weights to reduce this error. This is repeated until the network can accurately predict the correct output vectors for all the training samples. The process of presenting all training pairs to the network and assessing its accuracy of prediction is called an epoch. To reach a desired prediction accuracy, the training process may need several thousand or more epochs. There are several kinds of learning strategies in a neural network for error minimisation, among which back-propagation of the generalised delta rule may be the most well known. The trained network can then be used to calculate the output for given input vectors.
150
Total colour management in textiles
After the training process, a set of testing samples needs to be presented to the neural network in order to test its prediction accuracy. Input vectors of the test samples are presented to the neural network and the output vectors are then calculated. If the neural network is appropriately designed, the testing error should be comparable to the training error; otherwise, the neural network should be restructured and trained again. A common problem with MLPs is that they may be over-fitting the training data, i.e. producing a very small training error but a very large testing error. This problem arises when there are too many hidden units such that the network actually memorises the training samples rather than learns rules from them. In practice, the training and testing samples are drawn from the same population, and the set of training samples is larger than that of the testing samples. Previous publications on the use of artificial neural networks in colour recipe formulation of textile fabrics can be obtained from references 16 and 17. As claimed by the authors,16,17 the use of neural networks offers several advantages over conventional recipe prediction using K–M theory. 1. It is not necessary to prepare a special calibration database in order to use the neural network method. The network can be trained with real known production samples. 2. The neural network can continue to learn after the initial training period, since future data for the production sample can be fed back into the system and this knowledge incorporated into the network. This gives the network the potential to adapt to changes in important factors such as water supply, substrate properties or colourant strengths. 3. The network may be able to learn the behaviour of colourant systems for which the mathematical descriptions are complex. For example, fluorescent dyes are currently difficult to treat using standard K–M theory.
8.3.2 Fluorescent colourant formulation using a neural network A sample is said to be fluorescent if it contains electrons that can be excited by radiation at a low wavelength and emit radiation at a high wavelength. As the florescent colours can considerably enhance the whiteness and extend the colour gamut, they are widely utilised in fashion items. However, the K–M theory will break down in the colourant formulation of fluorescent dyes due to their anomalous optical behaviour. This section introduces fluorescent colourant formulation by the use of a neural network. For a fluorescent sample, some of the light incident on it is re-emitted with a change of wavelength. At each wavelength, the total re-emitted is
Controlling colourant formulation
151
the sum of the reflection and the fluorescence emission components. In discussing the effects of fluorescence, three spectral radiance factor (SRF) terms need to be defined.9,13 The first one is the reflected SRF bS(l), which is a ratio of the radiance produced by reflection by a sample to that produced by the perfect reflecting diffuser identically irradiated. The second one is the luminescent SRF bL(l), which is also a ratio of the radiance produced by luminescence by a sample to that produced by reflection by the perfect reflecting diffuser identically irradiated. The third is the total SRF bT(l), which is the sum of the reflected SRF bS(l) and the luminescent SRF bL(l): bT(l) = bs(l) + bL(l)
[8.26]
In the above definition the term radiance factor is used instead of the reflectance factor, because the latter applies only to reflected light and not to fluorescent light. According to the ASTM standard,14 the colour of fluorescent samples should be measured as they would be perceived when illuminated by daylight. The recommended measurement geometry is the 45/0 (or equivalent 0/45) illuminating and viewing geometry. In practice, the fluorescent colour is always measured using a spectrophotometer which employs polychromatic illumination of the sample and monochromatic detection of the radiant energy. Since fluorescent emission is related to the spectral energy distribution of the illuminant, usually a light source similar to D65 such as a xenon lamp is used in the spectrophotometer. The coating on the integrating sphere has an effect on the measurement as the fluorescent sample itself generates light at a long wavelength. Therefore, in measurement practice, the sample apertures on the integrating sphere should be as small as possible. In the following, an experimental example will be presented for better understanding of the fluorescent colour recipe formulation based on a neural network. In the experiment, 86 samples, which are polyesters dyed with disperse dyestuffs, are used. Three dyes used were Palanil Brilliant Yellow GN, Palanil Red FD-BFY 200 and Dispersol Navy C-VS 300. The samples were dyed using the three-dye mixture with different concentrations. The dyeing process was done in an Ahiba Nuance laboratory dyeing machine. The dyeing parameters are given in Fig. 8.8. The samples were measured using a Datacolor SF-600 spectrophotometer under the condition of illuminant D65 and 1964 10° observers according to the ASTM standard.14 The spectral range of the measurement is from 400 to 700 nm with a 10 nm interval. The recipe formulation is to predict the colourant concentrations based on the colour measurement results. Therefore, the input layer of the neural network was presented with the measured SRF bT(l), and the output layer
152
Total colour management in textiles
Temperature, deg
140 120 100 80 60 40 0
20
40
60
80
100
Time, min
8.8 Dyeing parameters adopted in the experiment.
was fed with the concentrations of the three dyestuffs. In the investigation, 60 samples were used for training, and the other 26 samples were used for testing. Neural networks with different hidden layers and hence different hidden units were constructed and their performances were compared. It was found that the network with a single hidden layer and 31 hidden units performed well. The relative prediction error of the network was calculated using the following equation:
predicted conc. − target conc. × 1000% target conc.
[8.27]
For the 86 samples used in the experiment, the training error was 2.3%, and the testing error was 4.2%. In addition to predicting colourant concentrations from SRF, the neural network can also be used to predict SRF from colourant concentrations. Bezerra and Hawkyard15 reported such research. They found that the predicted SRF curve was very close to the target SRF curve, with a mean CIELAB colour difference of 7.38.
8.4
A case study for matching a target using a commercial colour recipe formulation system
Figure 8.9 shows the main and additional functions often provided by modern commercial colour formulation systems. In these systems, colour can be measured using various types of spectrophotometers and stored in the system. The measured colours can be reviewed or edited using a house-
Controlling colourant formulation Main functions
153
Additional functions
Colour measurement
Colour specification
Data management
Monitor calibration
Calibration database
Recipe optimisation
Recipe formulation
Shade sorting
Recipe archiving Quality control
8.9 Functions integrated in commercial colour formulation systems.
keeping function, e.g. the data management functions in Fig. 8.9. The colour specifications including reflectance, transmittance, CIEXYZ, and CIELAB can be evaluated. Colour quality evaluations including colour difference, whiteness, grey-scale rating, yellowness, etc. are performed in a quality control function. In some systems, the colour difference evaluation or visualisation can be performed on a calibrated monitor screen in ‘what you see is what you get’ (WYSIWYG) manner, and the physical samples may not be needed. The method of monitor calibration can be found in Chapter 6. Some commercial systems are equipped with recipe optimisation functions using historical dyeing data. Shade sorting,18 a process to sort different batches with small colour differences into groups so that all batches within a group are an acceptable match to each other, may also be provided by a commercial colour formulation system. In this section, we show the workflow of predicting the dye concentrations to match a target colour (or standard) using the SCOPE® system19 developed by Gain Associates Inc., Taiwan. The first step is to measure the target colour using a spectrophotometer, as shown in Fig. 8.10. Then, dye combinations are selected from a list of dyes for which the calibration database has been established as shown in Fig. 8.11. The parameters such as illumination/observer, colour formula and matching criteria can be defined as shown in Fig. 8.12. The colour formulation results given in Fig. 8.13 show the predicted dyestuff concentrations, as well as the predicted colour differences under various illuminations. The predicted concentrations are then used to produce an actual batch sample, which is then measured into the system before using the colour quality control function to determine if the batch is a satisfactory match to the target. Figure 8.14 shows
154
Total colour management in textiles
8.10 Colour measurement.
8.11 Selection of dyestuff combination.
Controlling colourant formulation
8.12 Parameter setting for colour formulation.
8.13 Colour formulation results.
155
156
Total colour management in textiles
the process of selecting a standard sample and a batch sample. Figure 8.15 shows the colour difference results between the standard and the batch sample under various illuminations. The results indicate that the batch colour matches the standard colour well under the predefined colour tolerance (e.g. 0.7 DECMC(2:1) unit). If the colour quality of the batch is unacceptable, i.e. if the colour difference is larger than 0.7 DECMC(2:1) unit, further recipe corrections will be needed.
8.5
Sources of further information and future trends
The colourant recipe formulation using a computer system is widely adopted in the textile industry and in other colour-related industries. In this chapter, we introduced the widely used K–M theory and the alternative method of using an artificial neural network for colourant formulation. In the literature, various optimisation strategies have been presented for the K–M theory-based approach. For example, Sluban proposed a modified colorimetric algorithm to minimise the colour differences under several different illuminants.11,20 He also investigated colour sensitivity and correctability of colour-matching recipes.21,22 In the artificial neural network
8.14 Selection of standard and batch colours for quality control.
Controlling colourant formulation
157
8.15 Colour quality control.
approach, Mizutani et al. proposed a comprehensive evolutionary computing system integrating a neural network, fuzzy classification and a genetic algorithm.23 It is noted that both the K–M theory based and the neural network-based approaches only consider the optimisation problem from a mathematical viewpoint. In practice, however, the accuracy of recipe formulation is not only affected by optimisation strategies, but also by the whole colouration process. Chen et al.24,25 studied the effects of simultaneous and separate changes of several dyeing parameters on the colours. In the colour industry, one of the major objectives is to reproduce colour samples with high right-first-time rate. To improve the prediction accuracy considerably, the complex interaction between dyestuffs and materials needs to be studied further, with the various parameters under close control. In some commercial systems, historic recipes are stored in a database and can then be used to improve the accuracy of new predictions. Recipe optimisation in these cases may use numerical methods. However, to improve the prediction accuracy further, more intelligent algorithms need to be investigated.
8.6
References
1. Alderson, J.V., Atherton, E. and Derbyshire, A.N. (1961). Modern physical techniques in colour formulation. Journal of the Society of Dyers and Colourists, 77, 657–669. 2. Allen, E. (1966). Basic equations used in computer color matching, Journal of the Optical Society of America, 56, 1256–1259. 3. Allen, E. (1974). Basic equations used in computer color matching, II. Tristimulus match, two-constant theory. Journal of the Optical Society of America, 64, 991–993.
158
Total colour management in textiles
4. Kubelka, P. and Munk, F. (1931). Ein beitrag zur optik der farbanstriche. Zeitschrift für Technishen Physik, 12, 593–601. 5. Kubelka, P. (1948). New contribution to the optics of intensely light-scattering materials. Part I. Journal of the Optical Society of America, 38, 448–457. 6. Schuster, A. (1905). Radiation through a foggy atmosphere. Astrophysical Journal, 21, 1–22. 7. Judd, D.B. and Wyszecki, G. (1975). Color in Business, Science and Industry, 3rd edn. New York: John Wiley. 8. Kuehni, R.G. (1975). Computer Colourant Formulation. Lexington, MA: DC Heath. 9. McDonald, R. (ed.) (1997). Colour Physics for Industry, 2nd edn. UK: Society of Dyers and Colourists. 10. McDonald, R., McKay, D. and Weedall, P.J. (1976). Role of instrumental color control in optimization of dyehouse performance. Journal of the Society of Dyers and Colourists, 92, 39–47. 11. Sluban, B. (1993). Comparison of colorimetric and spectrophotometric algorithms for computer match prediction, Color Research and Application, 18, 74–79. 12. Eberhart, R.C. and Dobbins, R.W. (ed.). (1990). Neural Network PC Tools: A Practical Guide. New York: Academic Press Inc. 13. Hunt, R.W.G. (1991). Measuring Colour, 2nd edn. New York: Ellis Horwood. 14. ASTM E 991–98. (1998). Standard practice for color measurement of fluorescent specimens. 15. Bezerra, C.D.M. and Hawkyard, C.J. (2000). Computer match prediction for fluorescent dyes by neural networks. Journal of the Society of Dyers and Colourists, 116, 163–169. 16. Bishop, J.M., Bushnell, M.J. and Westland, S. (1991). Application of neural network to computer recipe prediction. Color Research and Application, 16, 3–9. 17. Westland, S., Bishop, J.M., Bushnell, M.J. and Usher, A.L. (1991). An intelligent approach to colour recipe prediction. Journal of the Society of Dyers and Colourists, 107, 235–237. 18. Aspland, J.R., Jarvis, C.W. and Jarvis, J.P. (1990). A review and assessment of numerical shade sorting methods. Journal of the Society of Dyers and Colourists, 106, 315–320. 19. Xin, J., Huang, H., Luo, R., Lee, S.S. and Lee, D. (1999). A new generation colour physics system: SCOPE. Journal of the Society of Dyers and Colourists, 115, 290–293. 20. Sluban, B. and Sauperl, O. (2003). Least metameric recipe formulation. Croatica Chemica Acta, 76, 161–166. 21. Sluban, B. and Sauperl, O. (2001). A sensitivity model and repeatability of recipe colour. Croatica Chemica Acta, 74, 315–325. 22. Sluban, B. and Nobbs, J.H. (1997). Colour correctability of a colour-matching recipe. Color Research and Application, 22, 88–95. 23. Mizutani, E., Takagi, H., Auslander, D.M. and Jang, J.S.R. (2000). Evolving color recipes. IEEE Transactions on Systems, Man, and Cybernetics – Part C: Applications and Reviews, 30, 537–550. 24. Chen, P.B., Yuen, C.W.M., Yeung, K.M. and Yeung, C.K. (1999). Determination
Controlling colourant formulation
159
of colour sensitivity of dye mixtures for disperse dyes. Journal of the Society of Dyers and Colourists, 115, 378–382. 25. Oulton, D.P. and Chen, P. (1995). Colour change sensitivity of dye recipes. Journal of the Society of Dyers and Colourists, 111, 237–244.
9 Controlling digital colour printing on textiles J R C A M P B E L L, Iowa State University, USA
9.1
Introduction
The variables in digitally printing imagery to cloth using ink jet technology are so great that, if the same print design were digitally printed on a variety of different printers using the same type of ink sets and fabric, the color results would vary widely. Factors such as environmental conditions, ink properties and print head construction can cause results to vary from day to day using exactly the same printer and inks. This can be very troublesome and frustrating for print providers. The approach to getting consistent and appropriate colour requires an operator who has knowledge of fibre and fabric structure, preparatory and finishing processes and the appropriate application of ink/dyestuffs, in addition to a solid understanding of the printing technology, of colour management systems and Raster image processors (RIP)s. This chapter focuses on the multifaceted factors that users of digital textile printing technology must address in order to control the application of colour to digitally printed fabrics. The choices of printers, software and solutions available on the market can be confusing. Users of the technology must evaluate a broad range of software and ink/dye solutions designed to support particular devices. How can the user best evaluate which components interact with each other to produce optimum results? A solid understanding of the printing technology, as well as the software and systems that run it, must be balanced with a clear picture of the role of the end output. Table 9.1 shows the printers that are currently on the market for use in digital textile printing, along with a listing of their key features. Descriptions of ink jet printing technologies are shown below.
9.1.1 Thermal drop on demand (DOD) inkjet Thermal ink jet print heads use heat as the mechanism for forcing the ink through very small openings. The ink is heated in many small chambers by passing a current impulse through a digitally controlled heating element. 160
Ink jet technology
Printing width
Print speed (maximum)
Inks/dyes available
360, 540, 720 dpi
Roll-to-roll web handling (requires special pretreated fabric)
Acid, reactive, disperse, pigment
Stork Piezo 1.6 m 8–10 m2/hour Reactive or Amethyst acid
DuPont Piezo 15–52 m2/hour 2020 (Seiko)
8 heads, process
16 heads, 8 process
Production
Production, sampling
8 heads, process
DuPont Piezo 3.2 m 30 m2/hour Pigment or 3210 (Spectra) acid Artistri (manuf. by Vutek) Roll-to-roll 360 dpi web handling
Sampling, one-of-a- kind artwork, short run production
Intended output/use
Encad Thermal 60 in 62 sf/hour (in Pigment, Roll feed 600 dpi 8 heads, 850/880 (microburst) (1.55 m) 1 ¥ 8 photo acid, and dual CMYK, or configuration) reactive or take-up 8 process disperse
Print head configuration Sampling, short run production
Roll-to-roll 600 dpi web handling
Fabric Resolution handling 12 heads, CMYK process and spot
Displaymaker Thermal 1.65 m 27.9 m2/hour Reactive or Fabrijet (Lexmark) acid XII (MacDermaid Colorspan)
Printer
Table 9.1 Examples of digital textile printers currently on the market
Built-in thermal dryer, 880 allows platen height adjustment
Special Info
Controlling digital colour printing on textiles 161
Piezo 2.2 m 10–15 m2/hour (multi- deflection technology) Reactive, disperse or pigment
Print head configuration
Roll-to-roll web handling
125 mesh resolution
Sampling, preproduction
Intended output/use
8 heads, spot
Production
16 heads, 2 ¥ 8 Sampling, staggered short run arrangement, 8 production colours
Roll feed 720 dpi 7 heads, 6 and take- process up
Fabric Resolution handling
Mimaki Piezo 1.6 m 4.6 m2/hour TextileJet TX1600s
Reactive, acid, disperse, pigment
Roll-to-roll 360 ¥ 360 dpi, 7 heads, 7 web 720 ¥ 720 dpi colour process handling
Proofing, short run production
Colour Piezo 2.23 m 22 m2/hour Pigment, Roll-to-roll 360 ¥ 360 to 16 heads, 2 ¥ Production Booster in 8 colour acid, 2880 ¥ 4, 1 ¥ 8, and (Hollander) mode at reactive, 2880 dpi any disperse, combination of 720 ¥ 360 dpi disperse concentrated transfer and semi- diluted inks
Chromotex SPM 788-F (Zimmer)
Reactive, acid (cartridges)
Inks/dyes available
17.6 m2/hour Reactive, Roll feed 360 ¥ 360 dpi, (high speed acid and take- 720 ¥ 720 dpi (cartridges) up mode), 4.7 m2/hour (high quality mode)
Print speed (maximum)
Stork/Lectra Piezo 1.65 m Sapphire
Printing width
1.8 m2/hour (in highest quality mode)
Ink jet technology
Stork Piezo 1.6 m Amber
Printer
Table 9.1 (Cont.)
Open ink system with antisedimentation system, integrated TFT touch screen
Head height adjustment up to 7 mm, ieee 1394 port
Special Info
162 Total colour management in textiles
24, 8 colour process
Production
Piezo 2.2 m 6.9–27.9 m2/hour Pigment, Roll-to-roll 360 to 8 process Sampling, variable disperse, web 2880 dpi short run dot reactive, handling production (reconstruction acid of Mutoh’s Falcon II)
Endless 360, 720 dpi belt drive
Textile Falcon by Digital Textle
Cenesta® reactive, acid
Piezo 1.6 m 26–78 m2/hour (original design by Epson)
Built-in drying system, head height adjustment up to 10-mm
Continuous integrated washing system to keep rubber feed blanket clean, online dryer
Reggiani Piezo 1.6 m 150 m2/hour CIBACRON® Rubber 600 dpi 7 heads, 6 Sampling, DReAM (Scitex reactive blanket process colours short run, + 1 medium Aprion) (acid, pigment run and production disperse in development)
MonnaLisa by Robustelli
Allows head height adjustment, ieee1394 port, antimeandering loading device helps keep fabric aligned horizontally, optional dryer
Mimaki Piezo 1.6 m 28.4 m2/hour Reactive, Roll-to-roll 360 ¥ 360 dpi, 16 heads, 2 Sample, TextileJet acid, friction 720 ¥ 720 dpi staggered lines short run TX2-1600 disperse roller of 8 colours production (CMYK + four colours specified)
Controlling digital colour printing on textiles 163
164
Total colour management in textiles
The heat produces a bubble of water vapour. The resulting pressure increase ejects an ink droplet on to the surface of the fabric. Temperatures of around 300 °C can be achieved, but a goal for developers using textile dyes would be to minimise the heating while maximising the droplet formation, to reduce the potential of the dye to reacting prior to its application to the cloth. The major manufacturers of thermal DOD ink jet print heads are Hewlett Packard, Canon, Lexmark and Xerox. In the total ink jet market, 85% of all inkjet heads produced are thermal (Ujiie, 2003). This is due primarily to the fact that the technology is inexpensive. Most thermal ink jet printers (sometimes also referred to as ‘bubble jet’) use water-based inks and are good for low-volume printing. High resolution is attained by using a small drop size. The performance of thermal print heads over their lifespan is an exponentially decreasing curve. Their consistency is also somewhat variable because, as the inks are heated up to nearly vapour level for creating the ink droplet, some of the colourant or binder is often deposited on the print head resistor (Tincher, 2003). Figure 9.1 illustrates the structure of a thermal ink jet head.
Ink reservoir
Orifice plate Heating mechanism
Vapour bubble
Ink droplet Ink chamber
9.1 Illustration of a thermal ink jet head through the droplet cycle.
Controlling digital colour printing on textiles
165
The use of DOD thermal heads in digital textile printing can have both advantages and disadvantages. Encad, a Kodak company, originally introduced the Encad 1500TX wide format printer for textiles in the USA in 1998. This printer, as well as Colorspan’s Fabrijet printer, were two of the earliest printers released in the USA, both employing thermal ink jet technology. The relatively inexpensive cost of the thermal heads allowed Encad to create separate ink cartridges for each colour printed and to have them fed by bulk ink tanks that could be refilled while the printer was printing. This meant that, if one colour failed or clogged the cartridge, then that specific cartridge could be removed and replaced without impacting any of the other print heads. A disadvantage of thermal ink jet technology is that it requires the ink to be heated to nearly boiling temperatures to cause the ink droplets to be released from the print head to the fabric surface. Because many fabric dyes use heat and moisture as the primary catalyst for reacting with the fibre, if the user wanted to print dyes that could bond with the fibre for permanent washable colour, thermal ink jet printing raised issues. When the dye was heated in the thermal ink jet head, it was losing some or all of its binding power prior to touching the cloth. As a result, ink developers who were producing dye-based inks for fabric had to modify the dyes to be more resistant to lower level heat so that they could be successfully run through the thermal print heads and still bond to the fabric.
9.1.2 Piezoelectric DOD inkjet Piezo print heads work by the following principle: droplets of ink are mechanically ejected from a nozzle in the ink chamber, when an electrical current impulse activates a piezo crystal. This crystal has the special property of changing its form when an electric voltage is applied, which causes the chamber to be compressed and the ink to be squeezed through the small openings in a manner similar to an oil can. Since the process is a mechanical one, it can be very precisely controlled. Figure 9.2 illustrates a piezo print head configuration. The major manufacturers of piezo DOD heads are Epson (90% of all Piezo heads), Brother, Spectra, Aprion, Trident, Konica and Xaar. Advantages to piezo technology are that the heads are more reliable and have a longer lifespan than thermal heads. They are good for high-volume printing and are able to produce the smallest droplet sizes in the printer world (Epson, June 18, 2004: http://www.epson.de/eng/about/piezo/). Piezo heads are slightly more capable of using a wider variety of inks and pigments, because the head can be created entirely out of inert ceramic materials. Also, the inks do not have to be modified to withstand the temperatures that thermal heads require. The performance of DOD piezo
166
Total colour management in textiles
Ink reservoir
Piezoelectric crystal
Orifice plate (nozzle)
Diaphragm (flexes to produce droplet)
Bubble formation
Ink channel Ink droplet
9.2 Piezoelectric print head.
heads over time is mostly level with a rapid drop-off. The main disadvantage of piezo technology is its cost. Another disadvantage is that piezo print heads are susceptible to entrapped air, which can lead to misfiring nozzles requiring multiple head cleanings (Baydo & Groscup, 2001). The print heads are permanently mounted into printers and are not easy to replace if they do fail. Piezo DOD printers are typically higher priced than the thermal DOD printers.
9.1.3 Multi or binary continuous inkjet head (CIJ) In the continuous process, a constant stream of droplets is generated by a pump and a piezo crystal. The droplets pass through an electric field and become charged. A second electric field transmits a digitally controlled signal to the droplet stream with the result that individual charged droplets are deflected and strike the fabric surface at specific locations. Around 100 000 droplets per second leave the ink chamber, although with very high
Controlling digital colour printing on textiles
167
performance printers up to 625 000 drops/second can be ejected (BASF, 2004). CIJ is the most complex print head technology. Of the three main print head types, CIJ is the least adapted to process colour printing. It is expensive to manufacture and maintain, but clogging is minimised. With CIJ, users must also deal with recycling or disposing of the unused ink. CIJ is well suited for high-speed applications. Major manufactures are Scitex, Stork, Domino and Toxot. The Stork Amethyst is the most popular CIJ printer that is currently being marketed for textile applications.
9.1.4 Initial questions of colour management software Once the basic structures and limitations of the hardware are understood, the user of digital textile printing technology must consider the role and impact of design and printing software. Colour management software for these systems must answer three key questions. Firstly, what is the colour gamut (range of printable colours) of the system, taking into account the printer, inks and fabric to be used? Secondly, are the desired colours inside the printer’s colour gamut? Thirdly, how can the system produce all of the colours that are within the colour gamut for the desired fibre type and fabric construction (Gordon, 2001)? The digital textile printing (DTP) user must initially do some market research to determine which colours are attainable within the limits of specific printers and ink sets. This information can be gathered by contacting a representative from the ink and printer suppliers. If a printer is incapable of producing a desired colour due to its structural set-up (for example, CMYK only vs. CMYK + MedM, MedC, LightM, LightC), no amount of colour management can make it possible. It is also very important to understand that there are some colours that can be displayed on an RGB monitor that are not printable using a CMYK device, and vice versa (CMYK representing the cyan, magenta, yellow and black inks used in process colour printing). A brief discussion of the different colour systems will be addressed later in this chapter. Many of the early printers that were introduced to the digital textile printing market were simply modified versions of graphics printing systems using a standard CMYK process colour set-up. These systems were initially not very well received by large-scale textile mills because the colour gamut obtainable in CMYK is considerably smaller than the gamut of spot colour inks used in conventional rotary screen printing of textiles. If sampling and strike-off printing was to truly take hold, these new printers would have to become able to more closely match the spot colour prints that were standard for textiles.
168
Total colour management in textiles
9.2
Characteristics and variables of digital ink jet printing (DIJP)
After the major decisions have been made regarding hardware and software, the user of DTP technology must be able to conceptualise the printing issues that distinguish this technology from traditional textile printing processes. This section will expand on the physical and user-orientated issues of the technology. DIJP does not require rollers or screens. It is a non-contact printing process; the print heads never touch the fabric, they merely drop the ink droplets on to the surface. As such, it is inherently different from almost every other type of large format textile printing. By its very nature, it allows for the possibility to design without a repeat. DIJP enables artists/designers to be more creative and provides for greater experimentation with colour (up to 24-bit process colour possibility) and composition with much less cost. A designer can realise an idea on cloth in a very short period of time and, if the design is not successful, the design process can be returned to immediately, modifications made and a reprint made for the results to be witnessed. From a textile printing production point of view, DIJP decreases machine downtime in-between designs. This has an important and profitable impact for companies producing strike-offs and samples. DIJP has the potential to lower the costs of short production runs and to produce them with a faster cycle time, thus allowing manufacturers to be more responsive to the marketplace. This technology also virtually eliminates dye waste and, as such, is more environmentally friendly. Another key feature of DIJP is that it eliminates major colour registration problems that are always an issue with roller or rotary-based screen printing. It also enhances the opportunities for personalisation and customisation, allowing artists and designers to interface directly with the end consumer to address their personal preferences and interests. Due to many of these capabilities, DIJP holds the potential to tip the balance of textile print manufacturing back to more industrialised, higher labour cost countries, while reducing the inventory costs that are currently incurred by housing the multitudes of screens and inks. Given that DIJP holds all of this conceptual potential, let us look more closely at the factors that must be accounted for in the day-to-day production of consistent and effective fabric prints. As we have already addressed the print head technology, the physical and use-orientated variables to be discussed below will be printing width/ fabric handling, print speed and features, print resolution, drop size, ink types/characteristics, fibre type, fabric structure, pretreatment, dye/ink penetration, environment, post-treatment and continued care. Each of these
Controlling digital colour printing on textiles
169
variables will impact on both colour matching and colour permanence. The integration of raster image processing (RIP) software, colour profiling and management issues will be discussed in section 9.5.
9.2.1 Printing width/fabric handling The maximum printing width of wide format ink jet printers for textiles ranges from 36 inches (91 cm) to as much as 126 inches (320 cm). The user should purchase a printer based on maximum print width needs. The greatest variable in fabric width lies with how the different printers handle the fabric. Since many of the early printers that were introduced to the digital textile printing (DTP) market were modified graphics printers that were used for printing paper, these printers required the fabric to be paper backed for stability and then fed through the printer in the same way that a roll of wide photo paper would be handled. This added cost to digitally printed fabrics. All of the Encad printers that can be used for textile printing require paper backing, but this also allows for printing of many other media types on the printer as well. Other issues with paper-backing will be discussed further in the section on pre-treatment. The first digital textile printer in the USA to handle the fabric without paper backing was MacDermaid Colorspan’s FabriJet XII. This printer came equipped with a component that the company called an ‘edge tracker’ that would automatically track the edge of the fabric and clip the print image so that ink waste and mess was avoided. As part of the fabric handling system, the FabriJet employed web handling with a web tension range from 500 gm to 5000 gm. If the width of the fabric on the supply roll got smaller, the take-up roll would get larger to balance the tension. Since the FabriJet didn’t require paper backing, it had to have an open platen so that excess ink would fall through sheer fabrics and be handled by an ink blotter. The FabriJet also had a component called the ‘selvage edge handler’ that would allow fabric with fringed selvage to be printed without causing the print head to be dragged across the fringe yarns. These features proved to be far more efficient for textile-specific printing needs than those of the more standardised printers, and they are the key features that continue to be implemented and improved with the more recent printers released to the market for fabric printing.
9.2.2 Print speed and features There are five key features relating to print speed that have an impact on the quality and quantity of ink application to cloth.
170
Total colour management in textiles
Printing speed Printing speed is controlled primarily by the mechanism by which the print heads are mounted on to a carriage and the system for traversing the carriage back and forth across the width of the fabric. Since digital data and ink are constantly being sent to the print heads there must be a flexible chain-like mechanism that can travel part of the distance across the platen. The heads will most probably be positioned on the carriage so that they are staggered along the length of the fabric. This allows for multiple colours to be applied during the same print pass. Print head movement (uni- or bi-direction printing) Printing speed will be greatly affected by whether the printer is capable of bi-directional printing, meaning that it prints ink both as it travels across the fabric and as it returns. These settings can often be adjusted on the printer to improve or reduce the quality or ink lay down, based on the user’s needs. Printing passage (print pass) To allow for the quantity of ink necessary to create the proper hue and saturation on the fabric surface, many printers are required to have multiple print passes of the same ink drop arrangement over the same area of fabric. Most graphic photo paper applications require a maximum of four-pass settings, while it is quite common to see textile settings of six-pass or more in order to lay down enough ink to get the proper colours on these much more absorptive media. Companies that have developed software packages for digital textile printing have often designed ‘double-strike’ settings that cause the printer to lay down twice as much ink as it would have done in a normal six-pass mode. Ink placement The mechanism for creating the precise droplet formation and placement on the fabric surface is a function of both the print head construction and the speed at which the carriage is moving. Software interpolation governs the pattern that the ink droplets will be arranged in, based on the structure and capability of the print heads. Head height To allow for the printing of thicker or textured fabrics such as pile weaves, some printers are able to adjust the height of the print heads or platen. For
Controlling digital colour printing on textiles
171
an extremely thick fabric, it is not recommended to use the fastest print speed settings.
9.2.3 Resolution How the raster image processing (RIP) software handles resolution is very important, but it is the printer technology that governs the geometric resolution (300 dots per inch (dpi), 600 dpi) vs. the perceptual resolution. A good RIP package will allow full control of dot gain adjustment and an ability to control half-toning. All of these adjustments will be heavily influenced or limited by the structure of the fabric that is being printed. There is some debate about the necessity of resolutions above 600 dpi in the use of digital textile printing. Higher resolution does allow for smaller drop size, and thus enables more accurate tonal control for images with large ranging gradient fills, but it is not clear that the finer detail in a photographic image will be attained, due mostly to the structure of fabrics (discussed in 9.2.7).
9.2.4 Drop size and formation A smaller drop size (perhaps 10 picolitres vs. 40 picolitres) should allow the print to capture fine details, reduce graininess and integrate finer tonal curves. The cleaner the droplet of ink, the less likely the occurrence of satellite drops, which can reduce the clarity of the image. The drop pattern is controlled by the firing of the print head and is affected by the viscosity of the ink.
9.2.5 Ink types and characteristics Ink factors that will affect usability include (i) the molecular size of the colourant, (ii) the colour gamut attainable with the dye class, (iii) the stability of the ink to resist precipitation so that it can have a longer shelf-life, (iv) viscosity, (v) the ink’s ability to be de-gassed so bubbles won’t form in the print head, and (vi) the ink’s colourfastness (washfastness, lightfastness, crocking). Inks to be used for DTP will vary based on fibre type, pretreatment, finishing and end-use. For sampling or printing fabrics that don’t require washfastness, pigment inks can be used. Pigmented inks require a binder to hold the colourant to the fabric. In ink jet printing, the binder must be applied either in the ink, by separate nozzle, or by pre-treating the fabric with a receptive binder. For more production orientated fabrics, or one-of-a-kind pieces, washfastness is likely to be more desirable, and so it would be appropriate to use the dye class that is most suited to the fibre type. To determine which ink type will have the best performance on the
172
Total colour management in textiles
major fibre types see Table 9. 2. Reactive dyes are used for cellulosic fibre fabrics, such as cotton. As their name suggests, the dye reacts with the cellulose to form covalent chemical bonds. To complete the chemical reaction, alkali, moisture and heat are required. For ink jet printing, the alkali must be applied to the fabric in a pre-treatment process because the alkalis will interfere with the dye’s effective drop formation and with the nozzle components on the print head (Ervine et al., 1999). Acid dyes are used to print the protein fibres (wool, silk) and polyamide. These dyes typically need a pre-treatment on the fabric to prevent the dye from wicking through capillary action across the surface of the fabric (Ervine et al., 1999). This is especially the case in the use of silk fibre fabrics. Disperse dyes can also be used to print synthetic fibre fabrics (predominantly polyester). The disperse dyes can be printed on to paper or directly on to the surface of the fabric. These dye sublimate into the surface of the fabric, so high levels of heat are required to make the dye permanent in the fabric.
9.2.6 Fibre type There are additional factors related to fibre type that will affect the way the ink is bound to the fabric. For instance, the absorbency, wicking properties, surface structure and length of the fibre will affect how the colourant is applied. Silk, cotton and wool are very absorptive fibres, which means that any water-based ink or dye will soak quite readily into the surface of the fabric. Most synthetic fibres are not nearly as absorptive, so they will require lesser amounts of ink to be applied to the surface of the fabric. Nylon, polyester and silk have good wicking properties, and so are more likely to require a pre-treatment that will reduce the wicking of the ink as it is dropped to the surface of the fabric. Wool has a very scaly and rough fibre surface compared to silk and the synthetic fibres, which adds to its bulkiness and absorption. This means that wool will probably require far more ink to attain the same colour than nylon or silk might. The length of the fibre contributes to the quality of the yarn and thus to the smoothness of the fabric. In DTP, the smoother the fabric, the more accurate and detailed the image can be printed.
9.2.7 Fabric structure The fabric structure will also impact the clarity and colour richness of a digitally printed image. Wicking can also be enhanced or subdued, based on the fabric structure. Weaves with long floats like satin tend towards higher levels of wicking, plain weave fabrics or knit minimise it. If the fabric structure is very bulky or uses a great deal of yarn, as is the case with knitted fabric, they will be more absorptive and require greater amounts of ink/dye
Controlling digital colour printing on textiles
173
Table 9.2 Best performance of ink on fibre type Inks/ Cotton Linen Nylon Polyester Silk fabrics
Viscose rayon
Wool
Acid
✗
✗
✗
✗
Dye- based/UV
✗
✗
✗
✗
✗
✗
✗
Disperse
✗
Pigment
✗
✗
✗
Reactive
✗
✗
✗
✗
✗ ✗
✗
to be printed. The denser the weave or surface structure, the more likely that there will be an increase in surface tension, which might be undesirable if the user is working with a low absorption fibre, because the dye will take longer to soak into the surface of the fabric. Some fabric structures will also tend to increase capillary attraction. The fibres and yarns adjacent to where the ink drop has fallen will be more likely to wick some of the colourant in their direction.
9.2.8 Pre-treatment The pre-treatment needed for the fabric will depend on the ink to be used (dye-based, reactive, acid, disperse, pigment). Many who are interested in entering the field ask, ‘is pre-treatment truly a requirement?’ The answer is most likely yes if we want to achieve the right colour, improve the performance of the ink, and maximise the fibre’s chemical reaction with the ink (Locastro, 2001). One type of pre-treatment that is unique to digitally printed fabrics is paper-backing. Is paper-backing needed? This depends on the printer and fabric. Because of their weight and stability, some fabrics can be run through printers that don’t support roll-to-roll web handling, but generally for these printers the paper-backing provides more peace of mind. An advantage to paper-backing is that, if the fabric being printed is of an open weave or is very light weight, some of the printed dye would actually soak through the back of the fabric. If there were no paper backing to absorb that excess ink, the ink would be deposited on the platen or stage of the printer, creating a potential for discolouration across the back of the fabric. A working dis
174
Total colour management in textiles
advantage to paper-backing is that it may be quite difficult to keep some fabrics on grain during the paper lamination process. A sheer fabric that is printed as sample curtain fabric cannot afford to be off grain when it is presented to a buyer.
9.2.9 Environment To maximise the efficiency of the printer, fabric and dyes, the environment in which the printing occurs should be controlled as much as possible. Reactive dyes require a much more humid environment to reduce clogging and maintain high levels of bonding to the fabric. This is not the case at all with pigments. Typically, the printer components run more efficiently in drier atmospheric conditions. Pay close attention to the suggested operating conditions for the inks, fabric and printer used. Maintaining more consistent environmental conditions will help to ensure more accurate and uniform printing.
9.2.10 Substrate post-treatment So when is post-processing a requirement? This depends on the pretreatment, the ink being used and the purpose of the output. Reactive and acid dyes require steaming for the dyes to bond to the fibre; disperse dyes require heat for sublimation. The two strongest reasons for post-treatment are to increase the colour gamut (many of the dyes display a wider gamut after steaming) and to improve colourfastness to light, crocking or washing. Figure 9.3 (see also colour section) shows experiments conducted at Kimberly–Clark with a 12-colour reactive ink set showing enhanced colour gamut space provided by steaming and washing. Where is post-processing done? Steaming and some fixation processes may be done on premises in either a low or high-pressure steamer. Highpressure steamers undoubtedly provide superior control and consistency from print to print, but are often quite expensive for digital textile artists or small production firms. To attain consistent colour in large batches, washing should most likely be done in finishing mills with conventional machines. Table 9.3 shows the most likely combinations of pre- and posttreatment needs for digital textile printing. Managing colour matching and consistency of digitally printed fabrics is greatly complicated by the post-processing requirements. Because acid, reactive and disperse dyes change colour and expand their gamut after steaming, colour profiling can most likely not occur until after this phase. In addition, some of the colours that have ‘popped’, or become enhanced, through steaming may be lost again during the rinsing process. If a fabric is going to be used in an end product, then rinsing is also a necessity to wash
Controlling digital colour printing on textiles b*
Reactive yellow
Reactive green
175
12-colour reactive ink Reactive set colour golden yellow gamut steamed and Reactive orange washed vs. not steamed Reactive scarlet Reactive red
Reactive black
a* Reactive grey Reactive medium turquoise
Reactive turquoise
Reactive medium red
Steamed Steamed and washed Not steamed
Reactive blue
9.3 Experiments conducted at Kimberly-Clark with a 12-colour reactive ink set showing enhanced colour gamut space provided by steaming and washing.
Table 9.3 Pre- and post-treatment options for digital textile printing inks Inks/ process
Pre- Steaming treatment
Thermal Washing fixation
Colourfastness
Acid
Yes
Yes
No
Yes
Yes
Disperse
Yes
Yes*
Yes
Yes*
Yes
Dye-based UV
Yes
No
No
No
No
Pigment
Yes
No
Yes
No
Yes
Reactive
Yes
Yes
No
Yes
Yes
Note: Some companies, such as Jacquard® Ink Jet have developed a modified post-treatment process for disperse dyes, in which they can be steamed in the same manner that reactive and acid dye printed fabrics would be. This allows users with the equipment to process natural fiber fabrics to also include some synthetics in their repertoire without re-tooling or purchasing more equipment.
176
Total colour management in textiles
off any excess dye that has not been fixed during steaming. In this case, colour matching and profiling will have to occur after the fabric has been both steamed and rinsed.
9.2.11 Continued care of digitally printed cloth As the digital textile printing industry becomes more developed in the production arena, there will be a need for more focused attention on the continued care issues with digitally printed fabrics. Although the overall performances of the acid, reactive and disperse dyes are thought to be similar to what they would be with traditionally printed fabrics, there are differences that could impact on their longevity. The nature of ink jet printing technology is such that very little ink is used per square metre of the fabric. Ink drops will predominantly soak only into the outer surface of the fabric after it has been printed, especially in tightly woven or pile fabrics. The dyes often do not penetrate the fabric completely. This means that digitally printed fabrics are more susceptible to colour loss from the wear of rubbing over time than they might be from washing. Laundering and storage issues of digitally printed fabrics need to be further explored to inform the design and development stages of printing. Washing conditions and agents are related to consumer preference; colourfastness problems may be minimised by consumers’ appropriate selections of washing conditions and agents. Therefore, in order to provide consumers with appropriate care information, it is necessary to examine how different washing conditions or agents affect colourfastness in the laundering of digitally printed fabrics.
9.3
Design potential and limitations of digital textile printing
This section will elaborate on the design issues and potentials/limitations of the technology for digitally printing to fabric. These issues are ultimately tied to the successful and consistent application of colour due to the designer’s role in conceptualising the end product. The following are issues that have been determined to have an impact on the design approach: (i) the use of repeat designs versus non-repeating image creation, (ii) the use of photo-realistic imagery, (iii) the potential for greater variations in size and scale, (iv) the possibility for creating more producible engineered digital three-dimensional forms, (v) limits of file size with contemporary software and hardware and transportability of files to the digital printer (Parsons & Campell, 2004; Ujiie, 2003).
Controlling digital colour printing on textiles
177
9.3.1 Use of repeat designs vs. non-repeating image creation Using traditional textile printing processes, the physical size of a print design is limited to the size of the screen or roller. With direct digital textile printing, it is possible to develop large-scale print designs in which the elements of the print design are never repeated.
9.3.2 Use of photo-realistic imagery In creating digitally printable imagery for textiles, the designer can incorporate the use of high-resolution images to push the limits of photo-realistic printing. The use of high-resolution images is only technically limited by the ease of use with currently available hardware and software and the current (and yet constantly changing) storage and transport media for dealing with large file sizes (Parsons & Campbell, 2004). Photo-realism opens the possibility for the use of multi-layered imagery, ghosting effects, an unlimited use of color, extreme tonal images, surface/ texture simulation, digitally created effects, and a number of possibilities that are not cost effective or even possible to produce through traditional printing methods.
9.3.3 Variations in size and scale Digital printing allows the designer to quickly print out the same image at different scales to test the visual impact, or to add variation to a textile collection without having to design multiple images. This is a function that would require a great deal of time and money, if it were done through traditional printing techniques. Designers can also use digital capture technology to obtain high resolution images of the microscopic world and transform them easily into huge scale prints, with little loss of visual quality.
9.3.4 Engineered digital three-dimensional forms Apparel, furniture and/or sculptural pieces can be designed for digital textile printing in which the imagery is continuous around the form. Through the integrated use of textile and apparel design software, printable designs can be tailored directly to pattern pieces for a garment. By engineering the textile print designs into each garment pattern shape, print designs can become more personalised and body specific. The image-filled pattern pieces are all that are printed to the fabric, leaving all areas of the fabric that are not used in the garment unprinted, thus saving ink. These pieces
178
Total colour management in textiles
can simply be cut out and sewn together to create a finished garment (Parsons & Campbell, 2004).
9.3.5 File size and transportability limitations Usability of the software and hardware involved in DTP is determined by how well the systems can handle the large computer files that are needed to be able to print large-scale images. Many of the RIP and spooling software packages that have been developed are specific to an operating system, so files must be converted and saved in formats that are acceptable across platforms (Mac, Windows, UNIX, etc.).
9.4
Role of end output: artist and industry approaches
This section will include an analysis of the ways in which the textile and apparel industry, as well as artists, are using the technology. The approach will differ greatly depending on the end use. Many artists have adopted the technology to create one-of-a-kind or limited edition artworks. Small-scale companies such as Gild the Lily, based out of Providence, Rhode Island in the USA, print scarves and garments on demand as they are ordered from their pre-designed choices on the Internet. Digital textile printing use is now almost completely integrated into the apparel and interiors manufacturing industry as it is used for sampling and prototyping. More recently, companies like Direct Digital Printing (DPP) in Sydney, Australia have begun to focus on collaborative production quantity digital printing for the interiors and apparel markets.
9.4.1 Recognising the role of the end output To get the right output, digital textile print providers must choose the right software, integrator, printer, ink, fabric, RIP and the right studio. Because many of the printer manufacturers traditionally are coming from a graphics perspective, research and development investments are not allocated for textiles. Users must understand the special features of the components to be used with the printers, i.e. inks, media and RIPs. For every given end product, there will be different requirements for speed, quality, colour matching, colour fastness, etc.
9.4.2 Sampling The first really cost-effective approach to users of DTP was to apply the technology to print pre-line or sales samples, design proofs or strike-
Controlling digital colour printing on textiles
179
offs. Much research and development has gone into making the prints that are coming off the ink jet printers look as close to screen printed fabrics as possible. This somewhat odd use of the technology makes great sense when the cost of a digitally printed strike-off is compared with one created traditionally at a mill. The technology can also be used to make a sample product to use as a prototype or production floor example. Perhaps the digitally printed sample could be used for a photo shoot to create catalogues for distribution to the public prior to beginning the production process for the product in an overseas factory. Other uses for sampling could include indoor banners/soft signage or tradeshow backdrops. For this type of use, pre-treatment of the fabric allows inks to be printed to the sample fabric without wicking or bleeding, promoting the highest quality image possible. The pre-treatment also allows more ink to be applied, resulting in more saturated colours. In most cases, these samples do not need to be water resistant, so no post-processing may be required.
9.4.3 Production Production style printing can be approached from a number of different angles, but the key similarity is that the end product needs to conform to the ultimate consumer’s expectation for style, innovation, usability, purchase-ability, and serviceability. Fabrics printed for production will most often need to be post-processed for colourfastness. Possible output for short-run production could include products used for marketing, sales or promotional activity (e.g. the cosmetic company Clinique could give away digitally printed bags to their customers to promote the company). One-ofa-kind products like garments sold in speciality boutiques, costumes for cinema or theatre productions (like the Austin Powers movie, for example), or artwork to be displayed in galleries would also qualify as production style printed media. In the more recent months, a few companies across the world have started providing production printed yardage to the apparel and interiors industries by using printers like the DuPont Artistri® 3210 or the Reggiani Dream® Machine. Pre-treatment with post-processing allows inks printed for these cir cumstances to bind to fabric. Most of the inks used for production printing penetrate the fabric, instead of simply resting on the surface. Some ultraviolet curable inks allow colours to be UV protected. The inks used for production will typically be water resistant after post-processing.
9.5
Ensuring accuracy and uniformity
A brief discussion on using and creating profiles for the different media types will begin this section. This will lead into a survey of the different RIP
180
Total colour management in textiles
software packages that are currently supporting textile printing. Profiling and testing is increasingly complex if the intention is to create wash-fast fabrics; and so ensuring accuracy will involve not only calibrating printers and monitors for the media, but also using consistent methods for steaming, rinsing and finishing the fabrics.
9.5.1 Process colour systems Looking at the different process colour printing systems will help in understanding the realistic colour capabilities and expectations one should have of these systems. CMYK is a four-colour printing process using three subtractive colour primaries with black, cyan, magenta and yellow. The colour limitations of CMYK lie in the difficulty of reproducing bright reds, greens and blues, as well as many of the colours required by the textile industry. The CMYK process is improved by including extra colours that cannot be reproduced by dithering or mixing cyan, magenta and yellow. The strongest complaints about digitally printed fabric from the textile industry are the visible dither of colours and limited colour range (or total gamut – hue, saturation and value) compared with traditional textile screen printing (Gordon, 2001). The dithering results from the printers’ lack of ability to create a sufficient number of steps in value of any given colour. So, when a very light colour is reproduced on the fabric, it is printed as very few dots spread out across the surface of the fabric. Because the dots are in isolation, they become more perceptually visible. The introduction of 7, 8, and even 12-colour ink jet textile printers brings the gap closer to achieving the results desired by the industry. As a general rule, the greater the number of colours (not print heads) that are in a printer, the larger the gamut of colours that can be reproduced. For example, a 12-colour printer with ten individual colours and two light shades will provide a much larger colour gamut than a 12-colour printer using CMYK with light shades. It is important, however, to have a balance of colourants to light shades to eliminate visible dither. When using textile inks such as reactive, acid or disperse, the full potential of these colour spaces are not realised until the colours have reacted with the fabric, which occurs during any postprocessing such as steaming and washing (see Fig. 9.3). The hardware and ink options available to the textile industry are a reflection of a growing market that has yet to develop any standards. As an example, Mimaki and Mutoh printers are available in versions that support both CMYK and Hexachrome® colour systems. The Mimaki can be con figured with any six or seven colours as well as CMYK with light shades or in Hexachrome®. DGS offers the Luxor 7, which is a Mimaki printer that supports three different inksets, using CMYK with special colours such as
Controlling digital colour printing on textiles
181
CMYK + (blue + green + gold), CMYK + (blue + grey + gold), CMYK + (C light + M light + K light). In addition, the ColorSpan 12 colour printers can be configured as either CMYK with light shades of cyan and magenta, or to use an 8- or 12-colour textile inkset. The inks determine the colour space, but the RIP drives and manages those colours. Digital textile printers are developed, tested and marketed with the use of specific inksets in co-operation with ink vendors offering inks specially formulated for the textile market. While established users such as fine artists and graphic designers have been known to stray from these established formulae in hopes of finding their unique niche in the market, playing with ink chemistries and established ink/hardware formulas is not for the faint at heart. Nor is it advisable for companies under tight timelines (Gordon, 2001).
9.5.2 Defining and profiling colour Components of RIP packages and colour management systems that should be considered are described below. The colour management system (CMS) and RIP software must be able to work together to create consistent colour from monitor to fabric. The International Color Consortium (ICC) CMS is the most common and it employs device-independent references of the International Commission on Lighting (CIE) L*a*b* colour space. ICC’s CMS characterises input colour devices, colour displays and output colour devices by using software and a spectrophotometer to create a small data file in the form of a conversion chart, which is called a profile. The profile creates a conversion chart by cross-referencing information from the input, display, or outputs’ colour space and determining the equal value of that information in the CIE L*a*b* colour space. An example of a crossreference chart is shown in Fig. 9.4. CIE L*a*b* colour space is one of the colour standards used by the textile industry. The CIE realised that every colour the human eye perceives could be defined using three numbers: L* indicates luminosity, lightness from white to black. The a* and b* are the chromaticity coordinates that indicate colour directions: +a* is the red direction, -a* is the green direction, +b* is the yellow direction, and -b* is the blue direction. The centre is achromatic – hues of grey. As the a* and b* values increase and the point moves out from the centre, the chroma or purity of the colour increases. The pythagorean distance between two colour points plotted in the colour space relates to the visual colour difference between those two points. In this way, colour variation between points and a standard may be expressed using numbers (Gordon, 2001). Colour management and RIP software manage colour by creating profiles or characterisations specific to the printer, ink, fabric and any postprocessing, such as steaming and washing. All of these variables have an
182
Total colour management in textiles Based on colour range (gamut) that can be printed on a device Uses device dependent to/from device independent transformations
R 253 251 248 243 230 217 204
G 4 4 4 1 9 9 8
B 3 3 3 11 9 9 8
L 59 56 55 54 52 50 47
a 73 71 70 69 67 65 63
b 75 75 73 70 68 65 64
L 54 52 50 47
a 69 67 65 69
b 70 68 65 64
Source profile
C 0 0 7 13
M 89 90 91 91
Y 100 100 100 100
K 0 0 1 2
Destination profile
9.4 An example of a cross-reference chart.
impact on colour and each variation must be profiled to ensure accurate colour match. When a design is printed, a profile is selected based on the printer/ink/media combination to ensure that the colours in the original design or target colours match the digitally printed output. The process of creating a profile or characterisation of a digital printer begins by printing out a linearisation file of the inks in the printer, typically from 0–255 saturation. The linearisation chart is measured with a spectrophotometer to calculate the distance between mathematical tonal ink values and actual printed tonal ink values. Next, a number of colour targets are printed and measured to map the colour space of printable colours. From all of these data points an algorithm is used to calculate the colour space and the profile is created. Various software packages offer different levels of profiling capabilities, from supporting standard ICC profiles created in third-party profiling software, to vendor supplied profiles, to end user capability to create custom profiles using proprietary colour systems. The International Color Consortium (ICC) colour profile is a standard profile format that characterises the colour-reproduction capabilities or colour gamut of devices such as scanners, digital cameras, monitors and digital printers. The price points of various software packages are often determined by the level of profiling and colour management capabilities. For instance, a RIP without profiling capability may be less expensive than one that uses proprietary systems to enable the end user to generate profiles. These
Controlling digital colour printing on textiles
183
options offer colour management of digital printing systems to customers who may not want to delve into profiling themselves.
9.5.2 Raster image processors (RIP) The RIP software is essentially a sophisticated printer driver that allows for greater user control in rasterising the image (converting the data) for the printer. The RIP software must be written for the specific printers in order to take full advantage of the hardware capabilities. The RIP ultimately controls the printer and the colourant. In addition to colour management capabilities, textile specific software is needed to handle print design images such as flat and continuous tone designs, separation files and files prepared for repeat printing. Important software features for DTP include its ability to accept textile industry file formats from CAD design and screen separation programs (such as CST, MST, PUB, GRT, SEP, SCN, XPF, etc.). The software must also be able to accept common graphic file formats like TIFF, Indexed 8-bit TIFF, PSD, EPS, AI, BMP, TGA, etc. Print layout functions such as step and repeat, design colourways, colour chips, multi-image placement, scaling, rotating, spooling or batching are also desirable features. A good RIP package for textiles should also be able to manage expanded ink sets beyond CMYK. In addition, it should allow for extra ink control functions to manage the higher ink densities required for colour saturation of printed fabrics. It should be able to create colour catalogues, colour palettes, and/or use the Pantone Textile Color System®. Profiling may be supplied by the vendor, but it is rare that the vendor would have appropriate profiles for all of the different fabric types and ink sets available, so custom profiling can be an important capability of RIP systems. A special feature of some RIP packages is the ability to create a colour gamut visualisation and comparison to see if the target colour is attainable. The screen print simulation features of software like DuPont’s Color Control and Management System (CCMS) are highly desirable if the digital output needs to match to screen printed production yardage. For the large sector of users who are solely trying to create strikeoffs for screen printed fabrics, it is helpful to have screen simulation features that can be viewed on screen to bridge the gap between digital and screen printed fabrics. Several software vendors have incorporated features useful in simulating and matching to screen printed production fabric, such as simulating screen resolution, raster simulation or screen mesh size; colour mixing, colour overprinting and colour trapping; incorporating gradation curves for tonal separations; and even profiling the textile printing mill’s colour space to link colour data to a textile mill’s colour kitchen. Table 9.4 displays the companies offering colour management and RIP solutions for digital textile printing. The colour management and RIP soft-
Printers supported
File types supported
Colour management
Special features
Artistri CCMS DuPont® Dupont®3210 Stork®Public, Custom printer profiles Screen print simulation Stork®Separated, specific to ink and features Tiff, RGB, TiffLab fabric, mill CCMS links the colour space characterisation and of both the colour profile digital printer and the textile Gamut mapping to screen printing mill in order compare ink jet to create digitally printed printer gamut vs. fabric that matches the textile printing mill textile printing mill. This is gamut accomplished by creating a Supports black, cyan, Mill Characterisation and lt. cyan, magenta, lt. colour profile of a mill’s magenta, yellow, unique color set, and a orange, and green printer profile to map inkset combinations of fabrics and ink. Evolution Digifab HP printers, All standard Set of ICC profiles Step and Repeat, scale, rotate, textile Systems Roland, graphic file included with RIP flip, multiply, mirror, cut, RIP and Mutoh, formats Supports ICC profiles measure, and layout RIP Plus Epson created by third capabilities. RIP Plus adds a 9000, party programs colouring system, Mimaki colourways, colour gradient, TX1600s, advanced colour Encad management, colour 1500TX & database and colour chips 800 series
RIP package Vendor
Table 9.4 Raster image processing software packages, with key features indicated
184 Total colour management in textiles
ProofMaster DPInnovations EPSON Stylus Indexed 8 bit Colouring tools: Print layout function includes Studio, Mill, Inc. Pro 3000 TIFF, CST, colourways, colour step and repeat, page and Pro to 9500, MST, PUB, editing, colour positioning, scaling, rotating, ENCAD GRT, SEP, catalogues, Colour mirroring, cutting out, colour TX1500 TIFF, BMP, gamut mapping chips, spooling, batching, and PROe AI, TGA, Adjustment of colours etc.
Vision NedGraphics Encad, Mimaki, Stork, PS2, Colouring tools: Preview of design on simulating Colorspan, all industry colourways, colour multiple substrates, printed HP, Konica standard editing, colour using multiple printing fabrics and Iris formats catalogues, custom and dyeing techniques (SPF) catalogues, Pantone Step-by-step UNDO Textile Color Library function included Simulation features: print Colour gamut mapping order, overlapping colours, Printer calibration: rasterisation, screen mesh, custom profile dye type, trapping, pad or creation by vendor, resist effects, dye opacity/ Colour data link to fabric absorption simulation colour kitchen PrinterServer Lectra Stork Amber, Stork, Colour gamut mapping Print functions include step and Amethyst, Lectra, Printer calibration: and repeat, job queuing, Stork Zircon, Tiff and custom calibration spooling, multiple image and TCP other profiles, vendor layout 4000 CAD supplied profiles Screen simulation features files Colour data link to Network to multiple CAD colour kitchen stations
MatchPrint II DGS Mimaki Tx Standard Custom profiling, Can be combined with Dua & Tx2, graphics built in colour atlas, Ramsete III software to do Graphic Encad 850, formats direct linearisation screen print simulation and Systems FabriJet XII generate colour recipes
Controlling digital colour printing on textiles 185
TexPrint Ergosoft EPSON Stylus All standard Colour and brightness Scaling, Rotation by 90°, 180°, Pro 3000– graphics correction or 270°, Tiling, Cropping 10 000, formats Specifying colour Textile repetition with any Mimaki (all), values for imported number of repetitions HP, ENCAD, separations: colour horizontally and vertically ROLAND separations (DCS or including displacement and HifiJet monochrome TIFF mirroring FJ 40/50/500, files) for output OPI support: PostScript files MUTOH (all) Separation for screen with Tiff files which are ORION, Fargo printing: Using the linked using OPI links Pictura 310S, colour management Font converter for MAC PS SUMMA to create separations fonts DuraChrome, Spot colour support Setting up various print PC 1500, via Photoshop spot environments SELEX colour channels or Optional features: Print and SG-950, DCS files with more cut module
series, HP PSD, etc. on multiple Simulation features: print DesignJet substrates, i.e. order, overprint and series, non-white fabric reservation, trapping, tonal Mimaki Ink control functions: screens such as Penta, Nova, TX1600S maximum ink level Galvano, Gravure, tonal and JV2, control, user gradation curve Ichinose, controllable ink Monitor ink level in printer to ColorSpan mixing insure sufficient ink for print DMXII and Printer calibration: job FabriJet XII profile wizard to printers guide end user step by step, vendor provides set of profiles, colour data link to colour kitchen
Table 9.4 (Cont.)
186 Total colour management in textiles
Wasatch Wasatch EPSON Stylus All common Colour catalogue, Print functions include step SoftRIP TX Inc. 3000 to graphic colourways and repeat, colour editing, and SP EPSON formats Printer calibration, layout features 10000, ENCAD supports ICC profiles Supports both indexed colour printers, and photographic imagery MacDermid within same print file ColorSpan DMXII, HP DesignJet printers, Mimaki, Mutoh, Roland
Textiler Image FabriJet XII, Tiff, CDI, Printer calibration; Print functions include step Technologies Mimaki Lectra supports ICC profiles and repeat, colorways, TX1600s, NedGraphics, created by third party colour editing, image Encad series HighTex, software programs, positioning, scaling, rotating, printers, SpeedStep, Linearisation mirror, and spooling. Mutoh, Epson Sophis BMP, capability On screen visualisation of best 7000/7500 EPS PCX, etc. colour match depending /9000/9500 upon output device
GRAPHTEC than four separations ‘Hot folder’ to output printable JX2150, data SignJet Print queue additional licences Pro 24/54, ICC profile generation for CALCOMP calibrating monitors, Techjet scanners, printers/media 5524/5536 with colourimeters and spectrophotometers provided by X-Rite, Gretag, SpectroCam, Colortron, etc. Specifying mixed colours
Controlling digital colour printing on textiles 187
188
Total colour management in textiles
ware options available to the textile industry reflect a growing market and growing acceptance of digitally printed fabric for proofing, sampling and short-run production.
9.5.3 Workflow issues To further ensure accuracy and uniformity in the process of DTP, design and printing studios must employ people who understand computer integrated design tools, colour, fabric issues and the components to use with the printers. Employers must have dedicated people to work with equipment, and must be organised with all of the variables, digital and non-digital. One of the difficult workflow issues, even after profiles have been created, is in trying to attain true blacks in production-orientated digitally printed fabrics. This is an issue that pervades the entire textile industry, but is aggravated by the fact that it is predominantly acid and reactive dye-based inks that are available for the DTP market. These dye classes traditionally have a difficult time creating a true black. In addition, fabrics that are very absorptive or bulky will pull in and distribute the relatively small amount of black ink that is printed to the surface, so that the overall result is more of a dark grey. If the user tries to compensate by printing even larger quantities of black ink, then bleeding commonly occurs and destroys the quality of the image. Another vexing issue is the continued need for calibration of printers and monitors, which requires a great deal of labour hours and downtime for printing. Without consistently scheduled calibration procedures, maintaining colour accuracy can be quite difficult. In addition, because of the relatively small lots of fabric that are being pre-treated for digital printing, when printing to natural fibre fabrics, it is very common for one lot to vary greatly from the next in terms of quality, absorbency and pre-treatment. This means that the user is likely to have to build or modify profiles of even the same fabrics from time to time.
9.6
Future trends
This section will have a discussion of upcoming improvements in the technology, as well as suggestions for future approaches to colour printing on textiles. It is no longer enough to be able to simply print fabric digitally; the industry is requiring colour matching and management throughout the design workflow, from scanning, to calibrated monitors, spectrophotometers, to the printing. Future hardware and software developments for the textile industry may include increased combinations of spot and process
Controlling digital colour printing on textiles
189
printing systems, and customisation of a wider range of ink colours that can be selected depending upon the colour space requirements of the design to be printed. Print providers like First2Print®, in New York, are leading the way in testing various inkset combinations on the fly in order to get the results their customers desire. With all of these future developments, the colour management and RIP software will be the engine that swiftly drives these constantly evolving systems.
9.6.1 Business and marketing model As a result of the continued growth and acceptance of the technology, quick response and short-run production will become more pervasive and help to shift some of the design focus away from large-scale production fabrics. As consumers become more aware of what is truly possible with DTP, there will be an increased move towards customisation and personalisation. Developments in the larger-scale production printers like the Reggiani Dream® machine will assist mills to become more effective with agile manufacturing. Digital textile printing may soon become combined as one component of the various production steps of print, cut and sew that are becoming mechanised for on-demand mass customisation practices of marketing. In addition, DTP will continue to provide a simplification of the path from apparel specification through to the manufacturing of finished product, and at the same time will help to meet the trend of vast customisation (the ability to customise a range of products in a variety of substrates: textile, paper, metal, glass, etc.).
9.7
Sources of further information and advice
[TC]2, based in North Carolina in the USA, continues to be a leading nonprofit research and support company for the textile and apparel industries. They host a textile and apparel technology information forum on www. techexhange.com that is an endless source of useful information. The Center of Excellence in Digital Ink Jet Printing at Philadelphia University, which is currently led by Hitoshi Ujiie, has been engaged in a great deal of research on the technology and often holds workshops for people who are interested in learning more about the technology. The Information Management Institute often holds conferences on ink jet technologies, including focused events on digital textile printing. As a newly formed institute, the Creative Institute for Design and Technology (KrIDT) in Denmark is likely to become another world leader in DTP research and teaching. These groups and institutions represent a growing number of people who are actively involved in continued research, development, design and production in digital textile printing. The issue of managing the
190
Total colour management in textiles
application of colour to digitally printed fabrics will continue to be refined and evolve as these groups bridge the gap between consumers and producers of digitally printed goods.
9.8
References
BASF, Retrieved on June 2, 2004 from: http://www.basf.de/en/produkte/farbmittel/ farben/textil/inkjet/technologie/continuous/?id=2vYR859hRbsf459. Baydo, R. and Groscup, A. (2001). Getting to the heart of ink jet: printheads. Recharger Magazine, July 10, 10,12,14. Epson, Information retrieved on June 18, 2004 from http://www.epson.de/eng/ about/piezo/. Ervine, S., Siegel, B. and Siemensmeyer, K. (1999). A simple universal approach to the ink jet printing of textile fabrics. AATCC Book of Papers, 1999 International Conference and Exhibition. Charlotte, NC, October 12–15, pp. 500–504. Gordon, Susu, Kimberly-Clark Corporation (2001). Color Management and RIP Software for Digital Textile Printing Managing Color for Optimal Results, July. Published by [TC]2 at www.techexchange.com, retrieved June 1, 2004 from http:// www.techexchange.com/thelibrary/DTPColorMgmt_RIPS.html. Locastro, D. (2001). Digital textile printing: forging the frontier for many clients. Presentation given at the CAD Expo, New York City, NY, August 22. Parsons, J.L. and Campbell, J.R. (2004). Digital apparel design process: placing a new technology into a framework for the creative design process. Clothing and Textiles Research Journal, 22 (1/2), 88–98. Tincher, W.C. (2003). Overview of digital printing and print head technologies. AATCC Review, July, 4–7. Ujiie, H. (2003). Digital inkjet fabric printing. Presentation given at the Surface Design Association Conference, Kansas City, Missouri, USA, June 7.
10 Colour management across the supply chain R L AW N, Consultant, UK
10.1 Introduction This chapter explains the structure of a typical colour critical supply chain in the apparel industry. This industry is chosen as an example because colour is so important there and garment supply chains are unusually complex, but the general points apply to plastic components supply chains for the automotive or electronic industries, for example, and to many other supply chains. Changes to the apparel supply change in recent years (essentially more complexity, faster) are covered, leading on to the colour processes involved and their information requirements. Three methods of dealing with colour information are then discussed: physical/visual, digital standalone and digital centralised, with a subdiscussion on the option of digital centralised being provided on an application service provider (ASP) model or not. The balance of pros and cons is clearly in favour of the digital centralised model, although elements of the other colour information methods will always be required alongside this. Finally, in future trends the eventual subsuming of colour software into more normal supply chain management solutions is proposed as the logical outcome of the application of web service technology approaches. Future trends also covers increased use of multi-spectral images.
10.2 Colour supply chains 10.2.1 Supply chain explanation The supply chain referred to here is a group of commercial operations involved in producing and selling a coloured product. In this chapter the comments focus on the apparel colour supply chain as an example, but the same trends highlighted below (more fashion, wider ranges, longer supply chains but shorter lead time targets) can also be seen in almost every industry where product colour is important (Fig. 10.1). 191
192
Total colour management in textiles
HQ sends seasonal palettes to Asia office Merchandisers sends standard to vendor
Vendor sends standard to mill
Merchandisers asks vendor for re-dip, with comments
Vendors adds info. sends lab-dips to merchandiser
Mill sends labdips to vendor
Mill produces lab-dips
Merchandiser passes lab-dips to colourists
Colourist checks lab dips vs. std
Rec. fail Rec. fail Rec. pass
Rec. pass
HQ colourist checks labdip vs std
Colourists pass lab-dip to HQ Rec. pass
Merchandiser requests base test from vendor
Vendor requests base test from mill
Mill produces base test
Mill sends base test to vendor
Re-try
Technologist Fail discusses with vendor/mill
Agree to allow or re-try
Allow Pass Merchandiser requests first bulk from vendor
Vendor requests first bulk from mill
Merchandisers asks vendor for re-dip, with comments
Vendor adds info. sends first bulk to merchandiser
Mill sends first bulk to vendor
Mill produces first bulk
Mechandiser passes base test to lab
Vendor adds info. sends base test to merchandiser
Mechandiser passes first bulk to colourists Colourist checks first bulk vs. app lab-dip
Rec. fail Rec. fail
Rec. pass
HQ colourist checks first bulk vs. appr lab-dip
Rec. pass
Lab checks base test vs. stds
Colourists pass first bulk to HQ Rec. pass
Merchandiser orders bulk from vendor
Vendor orders bulk from mill
Mill produces bulk
Mill sends bulk samples to vendor
Merchandisers rejects failed rolls
Vendor adds info. sends bulk samples to merchandiser
Rec. fail
Merchandiser accepts passed rolls
Rec. pass
Mechandiser passes bulk samples to colourists Colourist assesses bulk samples against appr lab-dip
End
10.1 The workflow for a typical apparel colour supply chain (for a large UK retailer).
The colour supply chain of the apparel industry would, as a minimum, include the mill colouring the fabric, the factory cutting and sewing that fabric into a garment, and the retailer buying the garment for resale to a consumer. In practice, supply chains are usually much more complicated.
Colour management across the supply chain
193
The retailer may buy from a brand owner, who may subcontract to a licensee, who may handle regional sourcing via a local office or sub-contract to a buying office, who may buy from a major vendor who subcontracts to several factories who buy fabric from a fabric agent who deals with the mill. Then there are all the companies supplying secondary components of the garment – thread, zips, buttons, straps, other trim, and so on. In all, up to 100 or so locations may be involved in getting the coloured parts of a garment produced, assembled and shipped to the buyer. Shipping the finished garment to the retailer’s warehouse is not the end of the colour supply chain, however. Colour management can extend to ensuring that garments of different sizes that hang on the same rack in the same store are the same colour, or that Internet-ordered matching top and bottom are indeed identical in colour, even if one was made several months before the other.
10.2.2 Recent changes to colour supply chains Most areas of the supply chain for apparel have changed enormously in the past decade or two. Twenty years ago, it was not unusual for a branded apparel supplier to be selling four seasons per year of styles planned long in advance, with a fairly small colour range. Production would almost all be in the same country as the brand owner, from factories and mills that were either part of the same company or at least closely linked commercially and geographically. Retailers were mainly department stores, which typically stocked branded goods sold by the brand owners. The key changes in this situation include: • 70%–80% or more of apparel sourcing has now switched to overseas, leading to the multi-player, multi-country supply chains described above. • Typically, the colour specifier knows little about the supply chain, and may not know even the name or location of the final mill. When supply was from one local vertical manufacturer, most brand owners trusted that manufacturer with colour development, feasibility checks (can we make this colour on this fabric?) and production colour quality (batchto-batch or roll-to-roll variation) control. When the mill is unknown, such trust tends to break down, and final buyers add processes such as the lab-dip approval cycle just to check that the mill can supply the required colour/fabric combination. • ‘Colour specifier’ is itself a new term, as many garments are now designed and sourced by major retailers directly (‘own label’) rather than from brand owners. Companies like the biggest of these, Walmart, have built
194
Total colour management in textiles
their fortunes on supply chain management and want to control much of it themselves. • Partly because of Walmart and others, the real prices of apparel have been dropping for most of those two decades. The search for ever cheaper production locations means ever longer supply chains. • Most of the personnel in these supply chains are involved in colour decisions in so far as they decide whether to pass on, or not, colour submissions from below them in the chain. But only at the mill end and at the main colour specifier are there likely to be trained colourists, or probably any colour measurement equipment. The other personnel have principally clerical roles, but they are part of the commercial relationships and can not be left out. • In parallel with the increasing complexity of such supply chains, the colour demands placed on them have also multiplied. Not only are there more colours in each season, there are more seasons, with the majority of companies aiming for ‘fast fashion’ – the ability to bring out a new line in time to catch a trend at its start. Many companies now aim for monthly product line adjustments, and some such as Zara restock their lines every 2 weeks.
10.3 Supply chain colour process requirements 10.3.1 Colour processes involved in the supply chain Overall, then, the colour supply chain is being asked to handle more colours, faster, through a more complex web of companies. The supply chain retains the same basic colour processes as before the changes, but the added complexity has meant the pressure to modernise those processes has increased. These processes (1–8) are discussed below. 1
Choosing colours
Usually done by designers, and the range of their inspiration is already too well covered to go into again here. It is a fact that, although superb supply chain management can keep an apparel brand in profit, only improved designs can double sales overnight, so getting precisely the right colour is very important. Designers tend to need to choose from real physical examples; the challenge is to create a useful example of the required colour – a physical sample that is close enough to the desired colour to be approved by the designer. At the same time it must also be feasible to produce with minimal metamerism on all the materials it will be required on in ways that conform to other requirements for the garment (washfastness, ecologically sound colourants, and similar).
2
Colour management across the supply chain
195
Checking that suppliers can produce the desired colours
As mentioned, supply chains are hugely complex nowadays and the fabric dyeing or printing facility may be on a different continent and separated by many commercial ‘layers’. It is therefore usual for a colour specifier to require proof, before ordering from a particular colouring unit, that the unit can create a colour close to that required. This proof is usually a ‘lab-dip’ – a small piece of the required fabric coloured to the right shade. Actually, usually several ‘rounds’ of lab-dips are required, as the colour specifier and colour producer ‘negotiate’ with the initial lab-dip being rejected as too far from requirement, and successive ones getting closer until the specifier is happy or until time pressures force an agreed ‘concession colour’ on the specifier. During this process, the specifier and supplier also discover any issues of impracticality with the colour if not spotted before (e.g. achieving exactly the colour target with particular fastness requirements may mean use of a high cost dye). 3
Checking that bulk production matches approved lab-dip
Fabric dyeing or printing machines of production size rarely produce exactly the same colour if fed with the same recipe and fabric as the laboratory machine that produced the lab-dip. Specifiers often therefore request to see the first actual production colour as well. 4
Secondary component matching
Most garments have at least five or six components, even if that only means main fabric, second fabric (for example for a collar), lining, thread, zip and buttons. Complex lingerie items may have 20 or more components, produced from a wide range of materials, using a similarly wide range of colourants. In every case the secondary components need to be matched to the standard, but often they are also checked or re-matched later to the accepted lab-dip or even to the first production colour for the main fabric. In some cases this matching process is a colour search of a provided shade range, in others, colours are produced to match and therefore a lab-dip loop or loops are usually involved. These processes are usually managed by the main vendor or the specifier’s office local to the main vendor – the degree of specifier involvement usually increases with increasing garment complexity. 5
Production checking
Long production runs invariably produce colours that vary through the run, from roll-to-roll on continuous machines or from batch-to-batch on batch
196
Total colour management in textiles
machines. The requirement is to ensure that the variations are within a tight enough tolerance for all the batches or rolls to still be close to the required colour. Some specifiers leave this to the suppliers whilst some have the suppliers send examples of selected rolls or batches. 6
Colour sorting
Once the rolls arrive at a garment maker for cutting and assembly, they will need to be sorted into groups or sequences that can be cut and then assembled. The acceptable colour tolerances between two sewn pieces (for example, between the main fabric and waistband and patch pockets on khaki pants) are generally tighter than the overall tolerances that have to be accepted across any total production run. So it is necessary to group the rolls of fabric into clusters of close colours, and to adjust the laying and cutting order accordingly so that all the components cut from one fabric for any given garment are from rolls with very closely matching colours. Usually this process is left to the garment maker. 7
Distribution choices
The processes just mentioned for roll grouping mean that, whilst particular garments will be highly colour consistent, the colour variation between garments will be perceptible if they are hung together or worn together (if a suit jacket made from one fabric group is worn with trousers from another group, for example). So, in a high colour quality process it is also necessary to ensure that the grouping choices made as described create a whole group – a complete size group, for example, or even a complete group of all sizes for a particular retail channel or retail region. To do this in all cases is an ideal (or perhaps the ideal is zero roll-to-roll variation, but that is rarely achieved), so most apparel supply chains do not achieve this currently, and just have to cope with returns due to non-matches, or in-store problems with obvious variations on hanging garment racks. A further example of the need for the colour grouping process is the increasing volume of apparel purchased on the Internet. A typical purchaser can tolerate a certain variation between the colour displayed on a web page and the colour that arrives by mail, but not any variation between two purchased items that are supposed to form a set. 8
After purchase
Finally, apparel buyers carry out several colour processes themselves, finding accessories or other apparel that match their garment as closely as possible. This is currently left almost entirely to the apparel purchaser.
Colour management across the supply chain
197
10.3.2 Requirements of these colour processes The colour information requirements of these processes are fairly simple. • All parties involved in the colour supply chain should be trying to match exactly the same initial standard. • They should be aware of the tolerances allowed from that standard. • They should be able to easily access an objective measure of how far any lab-dip or production fabric deviates from the required standard. • They should have an objective framework of description for differences between colours. • Colour information should be able to quickly move up and down the supply chain (e.g. lab-dip submits returning to the specifier, or roll-toroll data going to the cutter). • Any party involved should be able to access all the data necessary for their colour role, for example a garment maker should be able to search for trim to colour match an approved lab-dip. • Changes to colour information should be easily shared with all parties (for example, if the approved first production becomes a secondary standard, then everyone needs easy access to that new standard). • Colour information must not degrade over time, during travel from one party to another, or be substantially affected by the environment in which it is accessed. Finally, the non-colour information requirements must also be covered. Colour processes only work if all the people involved also have access to a great deal of additional information, on fabric types, design style references, even things as simple as lab-dip number, and this information must be reliably connected to the relevant pieces of colour information. Non-colour information also includes records of colour transactions – for example, a record of whether a particular party decided two colours were acceptably matched. Colour decisions can be very valuable or very expensive depending on the context, so keeping accurate records of them is also a requirement of all the colour processes listed above.
10.3.3 Method choices available for colour communication The available methods for storing and communicating colour information are physical and digital, and for communication and management purposes digital colour information can be further split into standalone or centralised. To explain these in more detail. Physical/visual The physical/visual method is still the one used for perhaps 99% of all colour transactions. This means that colours are stored and communicated
198
Total colour management in textiles
as physical examples – fabric swatches, for example. So, a specifier will send out physical examples of their standard colour, and mills in the supply chain will send back physical samples of their lab-dips or production. Non-colour information is handled in the physical method by typing it on to pieces of paper attached to the physical swatches. Digital standalone ‘Digital standalone’ means that colour information is stored digitally but at many standalone locations, typically on PCs. From an IT perspective, the most important feature of digital colour measurements is that they are meaningless on their own – any digital colour transaction requires a ‘colour engine’ to handle the data (for example, to calculate whether the objects described by two measurements are close enough in colour terms in a particular light). This aspect of digital colour data is covered in more detail in Future Trends below, but the relevance here is that, although in a standalone digital model, the data can be transferred from one party to another, usually by point-to-point email, any party that works with it must also have access to a colour engine, normally standalone PC based colour software. Note that physical colour information can be evaluated visually or by using a spectrophotometer, which converts the physical information into digital, so the two methods above can mix (lab-dips are sent physically, but they are checked against the standard visually and instrumentally, for example). Digital centralised With the digital centralised method, colour information is again stored as digital colour measurements, but in this case on one central database which can be accessed, in different ways by different parties, over the Internet (usually via a simple web browser). The server hosting the centralised database also hosts a full feature colour engine, which means parties in the supply chain can access and work with the data without any extra software. Table 10.1 shows how these three options perform against the requirements listed above. The comparison applies to their use in processes a) to f) and mainly to processes a) to e). Only the digital centralised method can really be applied to processes g) and h), and then only after the developments described in Future Trends.
10.3.4 ASP or not The combination of digital colour measurement and centralised data storage is now seen as the way forward by most technology leaders in the apparel
Physical
Digital standalone
Digital centralised
(d) They should have an Colour variations are very Numerical differences are Numerical differences are objective framework of hard to describe in text- consistent and have same consistent and have same description for differences only terms in a consistent meaning for everyone. meaning for everyone. between colours. and meaningful manner. All parties must have access Calculations can be provided to colour software. on a web browser.
(c) They should be able Visual assessment is Colour variations are Colour variations are to easily access an inherently subjective, no numerically described. numerically described and objective measure of standardised output. All parties must have access easily understood by all. how far any lab-dips to colour software. Calculations can be provided production deviates from on a web browser. the required standard.
(b) They should be aware Very hard to describe Tolerances can be simply Tolerances can be simply of the tolerances allowed tolerances allowed given described in numerical terms. described in numerical terms. from that standard. only physical standards Updating them or using Updating them globally is and text. different tolerances for instant, and a choice of different situations can be tolerances can be easily complex via email with offered. many parties involved. Calculations can be provided All parties must have access on a web browser. to colour software.
(a) All parties involved in Engineered physical Properly calibrated, Properly calibrated, profiled the colour supply chain standards are available profiled and maintained and maintained should be trying to with guaranteed variation spectrophotometers can spectrophotometers can match exactly the same < ±0.5 DE (CMC 2 : 1). have inter-instrument have inter-instrument initial standard. Many other physical variation < ±0.3 DE variation < ±0.3 DE (CMC 2 : 1). standards vary by much (CMC 2 : 1). Checking this Centralised data storage of all more than this. across many instruments measurements allows using email can be variations to be easily tracked. complex.
Requirement
Table 10.1 Colour methods choices vs. supply chain requirements
Colour management across the supply chain 199
Physical
Digital standalone
Digital centralised
Non-colour information Only method is attaching Non-colour information can be Non-colour information can be must be accurately paper details to physical stored with digital stored with master record, recorded and reliably swatch, can become measurement in same file or always linked and available. linked to relevant colour disconnected easily. record, though multiple Editing possible but centrally information. storage locations and frequent controlled. passing on and editing can allow mistakes to enter.
(h) Colour information Physical standards fade, Digital measurements are Master of each digital must not degrade over can be damaged or spoilt, constant, though email measurement is stored on time, during travel from and are affected by communication can be server, cannot be corrupted or one party to another, or temperature and corrupted. edited. be substantially affected humidity. by environment in which it is accessed.
(g) Changes to colour Issuing any new data New colour measurements New colour measurements can information should be means sending new can be easily circulated, be made instantly available easily shared to all physical examples to though ensuring everyone and any replacement of parties. everyone involved and has changed over can be standards is centrally ensuring changeover complex. controlled. happens everywhere.
(f) Any party involved Anyone requiring to carry Digital measurements can all Anyone can access all the should be able to access out a colour transaction be shared amongst everybody measurements they need, for all the data necessary for must collect all the in theory. In practice many example for online search of their colour role. physical data (such as measurements and many a live trim colour database lab-dip plus shade cards parties to be constantly without any local software. for trim),impractical for updated means this is very all players to do this. complex, and again all parties must have access to colour software to carry out trim search or similar.
(e) Colour information Physical swatches must Digital measurements can be Digital measurements once should be able to quickly be sent by mail or sent by email. saved can be accessed by move up and down the courier. anyone with permission. supply chain.
Requirement
Table 10.1 (Cont.)
200 Total colour management in textiles
Colour management across the supply chain
201
sector, and the growth of this method is rapid considering that only a few years ago the concept of broadband Internet connections to a dye mill in China would have seemed a long way off. One follow-on choice to be made once a digital centralised model is chosen is whether the centralised database and colour engine should be hosted by the specifier itself (internal) or by the software solution provider (the application service provider or ASP model). The pros and cons depend mostly on the size and complexity of the supply chain involved and the IT support skills of the specifier, but the ASP model has many advantages for the apparel industry. This is because an apparel supply chain may have more than 1000 parties involved, in many countries and time zones, and the actual parties may ‘churn’ (i.e. some are dropped, some added) continually. If the specifier IT team is responsible for managing this churn and supporting each party, the work load can be huge. Also, such support generally includes colour science education as well as software help, and is typically in a language different from that of the specifier’s home country. There are also serious IT security concerns if all parties are to be brought inside a specifier network. A useful analogy is with the generally existing physical colour communication systems, which are always handled by external parties – courier companies such as UPS, Fedex, and DHL. An apparel supply chain should have its digital colour communication supported on a third-party platform for the same reasons as apparel specifiers use Fedex rather than starting their own courier company (Fig. 10.2). In other industries this reasoning may be less valid: FMCG companies, for example, tend to run sophisticated global IT networks, and their supply chains have fewer players and fewer layers, so internal hosting may be appropriate.
10.3.5 Best practices Most apparel supply chains will use some combination of all three methods with the centralised digital method gradually becoming the default – almost all designers and many mills will not drop their preference for physical examples of standards, and there are some benefits to offline software (for example, for recipe prediction where Internet connections are difficult or expensive). Generally accepted best practice in this area includes: • use of a single digital measurement as the master standard for each colour, with this measurement being shared with all parties from a centralised database and colour engine, via the Internet; • use of engineered physical standards as visual representations of the digital standard, with online ordering and recording of their use; • optionally, the ability to switch from the original standard to a ‘production standard’ after first bulk production has been approved on the main
202
Total colour management in textiles
Other Specifiers? Specifier Standards supplier Vendor
Vendor
Mill
Subcontractor
Trim supplier A Trim supplier B Trim supplier C Trim supplier D
Test house 1
Subsidiary Subsidiary
Mill Dye supplier Z Dye supplier Y Dye supplier X
Test house 2 Local agent Remote agent
Vendor Subcontractor Fabric agent
Vendor Mill Trim supplier 1 Trim supplier 2 Trim supplier 3 Trim supplier 4
Mill
10.2 An example of the many parties involved in typical colour supply chain for the apparel industry.
fabric. This can be useful where the main fabric colour must be adjusted to take account of significant production issues (such as colourant costs and ecological effects) and where many other components must then be matched to the main fabric, especially if the garment will be viewed under many lighting conditions. If the original standard is maintained in this situation, problems of metamerism between main fabric and secondary components may arise; • use of measurement variation tracking modules within the centralised database. As it remains normal for the specifier to still ask for a final physical example of what they have approved digitally, even if only for archiving purposes, a centralised system allows a digital re-measurement of, say, a lab-dip, to be compared automatically to the supplying mill’s original measurement of it. The centralised nature of the database means any authorised user can then see live reports on the differences between the readings that refer to real textile measurements, and to act accordingly; • continuous checking of the reasons for any measurement variations seen and addressing of them, which will typically includes training and certification in sample preparation and measurement procedures;
Colour management across the supply chain
203
• additionally, the use of instrument profiling modules that use software to reduce measurement differences between spectrophotometers in the network. If a centralised digital solution is being used overall, the use of a centralised (for example Internet-based) profiling module is appropriate; • constant monitoring and mining of the data stored on the database to spot bottlenecks and track improvements in colour quality, process lead times, workloads and similar. Overall, perhaps the most important best practice in improving colour supply chains is to see colour management as yet another business process which can be improved, by measurement, by shared information and requirements, and by continuous effort. Although digital colour measurement and calculations are not perfect, for the reasons given elsewhere in this book, they are certainly good enough to give immediate improvements to most colour processes (Fig. 10.3).
10.3.6 Demonstrated benefits The most commonly demonstrated benefit of the best practices just listed is a reduction of several weeks in colour approval lead times compared to the physical process. This typically happens because the physical process may have included between three and ten ‘loops’ of lab-dip submission where a specifier rejects a lab-dip, another is produced and sent, and so on (Colour Process 2 above), whereas in a centralised digital submission the supplying mill will ensure that the submit is within tolerance digitally before sending a physical sample – the specifier approves remotely and digitally and uses the physical sample only for confirmation. After a few months of using a digital colour centralised architecture, specifiers typically are approving more than 90% of the first physical submit they see, for example. Other benefits seen in most cases include: • lead time reductions where roll measurements are shared instantly with garment assemblers to allow pre-planning of cutting and pre-selection of rolls; • faster trim selection via online colour search, which also saves trim suppliers the cost of maintaining updated physical shade cards; • reduced mill production costs from lower re-dips; • lower courier and postage charges arising from all the other benefits; • lower return costs in the retail stores; • minimised need for spectrophotometers – standards, submits, bulk samples and rolls, etc, need only be measured once at point of production, with everyone else simply using that data – nothing should ever be re-measured except to check measurement variations.
204
Total colour management in textiles Specifier design team
Standards producer
Measure physical standards into core folder to create master digital standards
Designers move digital standards into palette folders based on review of physical examples
Central colour store and engine on an Internet server
Specifier merchandiser team
Merchandisers access palettes to create colour requests (jobs) in web-based merchandising system
Supply chain layers (buying offices, garment assemblers, Mill adds etc.) pass colour jobs up measurements of their and down chain, submits to jobs, reviews against standards, always able to access standards and review decides whether or not submits on web to pass up the chain, all browser on web browser
Supply chain layer 1
Supply chain layer 2
Colouring unitmill or dyehouse
Job flow in colour approval system
10.3 The workflow of a typical colour supply chain on a digital centralised model.
Over time, the continued use of shared systems and agreed data means the specifier may pass authority for colour approval further down the chain, eventually perhaps using the roll-to-roll variation data just as a continuous monitor of the ongoing colour quality performance of the dye mill. If this method is reached, the supply chain will have returned close to the ‘old’ route of trusting a local supplier, but with the cost benefits of long-distance sourcing.
Colour management across the supply chain
205
10.4 Future trends 10.4.1 Links to other software As mentioned above, non-colour information is a major part of what needs to be communicated down a colour supply chain. Yet almost all of that information is already available inside other software solutions used in the supply chain, and typically gets re-typed into the digital colour software solutions. So the more advanced digital colour solutions are already using the latest software communication protocols (such as XML file transfer or application protocol interfaces) to take this data seamlessly and invisibly from the CAD or product lifecycle management solutions and to return status data (on where in the colour cycle a particular set of jobs has reached) back to those solutions.
10.4.2 Standardised data Links will be made easier if all parties can agree on what the various data field headings mean in this area – there are almost as many definitions of what the difference is between ‘material’ and ‘fabric construction’ as there are specifiers involved. The American Association of Textile Colorists and Chemists, AATCC, has been working on standardising these definitions whilst their UK counterpart, the Society of Dyers and Colourists, SDC, has been addressing the standardisation of instrument measurement settings and their descriptions. This work will definitely lead to easier communication through the colour supply chain.
10.4.3 Increased use of multi-spectral imaging The discussion above centres around the use of spectrophotometers to create digital colour information. Such instruments can measure only one colour at a time, yet many garments include textile prints with many colours on them. The mismatch is not as bad as that statement implies – almost all multicoloured textiles have their colours added one colour at a time (one yarn, one ink, etc.) and these can be individually monitored (and there are usually not many basic colours involved). The problem arises more from the enormous complexity of calculations needed to model the human response to variations in one colour in a multi-colour object. Whilst the maths needed to model human response to variations in one colour is now good enough to be used for objective colour management processes (in other words, better than any individual human checker), that is not true for the maths for multi-coloured objects. World-class research is going on in this area, however, and will definitely form part of future colour supply chain technology.
206
Total colour management in textiles
That in turn will lead to a greater need for multi-spectral imaging devices – whose output is a picture in which the spectrum of each pixel is measured or derived. These are not so far removed from current spectrophotometers – which can be described as a long thin CCD camera with a diffraction grating instead of a lens, but commercial units are only just becoming available. It is possible that such devices will replace spectrophotometers in the longer term, however.
10.4.4 The end of colour software? To look further ahead, though, it is necessary to ask why there should be digital colour software at all. After all, there are digital measuring instruments for length, weight, yarn strength, fault levels and many other types of data that are communicated around the apparel supply chain, and around many other supply chains. So why does no one sell ‘digital length software’, or indeed write book chapters on ‘weight in the supply chain’? There are many differences between colour and weight as product attributes, but these differences relate principally to human perception. There are some similarities too, because weight depends on the environmental gravity just as colour depends on environmental illumination, but that is beyond the scope of this chapter. When we deal with what we call colour measurements (actually measurements of the reflectance properties of a coloured object: colour is a sensation in the brain of the observer and almost impossible to measure, as covered elsewhere in this book) we are referring typically to a set of 31 numbers output from a spectrophotometer, digitally not very different from the numbers output from a digital weighing machine. So the question remains – why do we need specialist software for dealing with colour measurements but not with weight measurements? Because not only are there globally accepted systems for weight measurements (kilograms, for example) but the maths of dealing with them is available everywhere. An object weighing 1.21 kg meets the standard 1.2 kg ± 0.02 kg. In a database sorting, the 2 kg object will always show between the 1 kg and the 3 kg one. The same database can be searched for the record of the object closest in weight to my target, and if I have 10 kg of a dyestuff I can always manage a recipe requiring 3.35 kg of that dye. In other words, standard mathematical functions, of the type built in to every spreadsheet, database, CAD solution, ERP program or any other software solution, can work with weight data. There is no need to build special systems to check whether or not the supplier’s fabric weight is within 2% of the specifiers standard – any purchasing system can confirm this if fed the weight data. None of this is true for colour measurements. There are no spreadsheets sold that, given two sets of data from a spectrophotometer, can calculate how close they are in colour terms, and even the most powerful databases
Colour management across the supply chain
207
from Oracle or IBM or Microsoft cannot search a set of colour measurements and output the closest one to my required target (the equivalent of searching a trim shade card for a match to a lab-dip swatch). This is because, to carry out these operations, the software needs to include a colour engine that not only has all the integral maths functionality needed for working with reflectance data, but also stores all the other data (such as illuminant profiles) that is required alongside reflectance readings. And, until recently, colour engines were only available in standalone PC-based colour software. Although, in theory, Oracle could include them in a database solution, or Microsoft in a spreadsheet program, they have not chosen to do so. Colour engines can now also be Internet based, interfacing directly with spectrophotometers (and with standalone colour software) and delivering colour measurements plus the required colour functionalities (colour compare, colour search, formulate, etc.) to ‘clients’ such as web browsers, with no additional software needed by the user. That is a major step forward for colour in the supply chain, but logically it will not stop there. Most of the ‘big’ systems that run the world’s supply chains (enterprise resource planning – ERP systems, inventory systems, supply chain management – SCM systems, and similar) can also act as clients for miniapplications delivered over the Internet, provided the interfaces are configured correctly. The term for the correctly configured services is Web Service Components. Almost all these big systems are built on a centralised database architecture (no one would run a global purchasing system by email), so it is simple to tag a record relating to a garment order in a big purchasing or planning system (which will have a colour name in it but no colour measurement) to a record on the centralised colour system that stores the colour measurement for that named colour. Whenever the big system wants to carry out a colour calculation on two or more items, it ‘asks’ the colour system for the answer, as shown in Fig. 10.4, and then carries on. For example, a supplier mill might still enter their lab-dip colour measurement into a web page to see if they met specification, but that web page might be the web-based part of a specifier’s purchasing system. The colour server can also provide the data needed to show the colours on screen correctly within the purchasing system if required. Such an approach will also address colour processes 7 and 8 above – distribution and retail logistics are already almost always controlled by big centralised database systems, so an inventory management system could ‘decide’ which sets of apparel to move to which shop by asking the colour server for a colour grouping – Process 7. And, the retail inventory system can ask the colour server to search all the colour measurements relating to all the items in stock to find a match for a customers target colour – Process 8.
208
Total colour management in textiles Colour server
Inventory item/SKU records in mainstream IT systems
Record ID in mainstream database linked to colour measurement record in colour server
Requirement for colour sorting of inventory
Mainstream system provides colour server with listing of records to be sorted, and parameters for sort
Colour measurement records
Colour engine, with algorithms for colour search, sort, compare, formulate, etc, plus illuminant, observer profiles
Colour server colour sorts the requested records, returns the sort results in terms of record IDs for the mainstream records
Mainstream solution then works with sorted inventory records in same way as with results of other sorts (e.g. sort by productions date)
10.4 The workflow for a combined big system and colour server colour transaction.
The leading colour software vendors are already working on achieving this vision. When they do, users will no longer use ‘colour software’, they will just work with colour inside there normal supply chain solutions in the way they work with weight now, and many colour decisions will be made automatically by those big systems. But there will be many more digital colour measurements and digital colour transactions taking place than there are now and the requirements of colour in the supply chain will finally be being met in full, or close to it.
Colour management across the supply chain
209
10.5 Conclusions Apparel is used here as an example but all colour supply chains have, in recent years, seen greatly increasing complexity, coupled with requirements for much faster throughput. These challenges can only be met by the fuller use of digital colour measurement and communication, and the current best practice is for a combination of centralised digital colour solution (with database and colour engine available via the Internet) and engineered physical standards. But the full needs of colour in the supply chain will only be close to being met when colour software as a specialist solution has faded from sight and colour functions are built into generic supply chain management and distribution software via the technology of Web Service Components. At that point, in addition, whole new business opportunities will open up for colour processes.
10.6 Further reading Description of the apparel supply chain, see Birnbaum’s Global Guide To Winning the Great Garment War by David Birnbaum. Methods and best practices to be used in digital colour communication. The AATCC website at http://www.aatcc.org has a wide library on recommended practices, including Color Management Principles. This site also covers the work on standardised descriptions for non-colour data. Explanation of spectrophotometer profiling technology. A market leader in this area explains their approach at http://www.gretagmacbeth.com/index/products/ products_color-communication/products_color-compliance/products_netprofiler. htm Explanation of other supply chain management solution areas. There are numerous books and papers on this area. TC2 has a wide library of those relating to apparel and textiles at http://www.techexchange.com/apparel-supply-chain.html Explanations of ASP and web service technology models. Out of the Box: Strategies for Achieving Profits Today and Growth Tomorrow Through Web Services by John Hagel III. John Hagel has written widely on this field, and many more papers can be found at www.johnhagel.com.
11 Quality assurance management for coloured goods M S B A L L, Consultant, UK
11.1 Reproduction of colour 11.1.1 Traditional identification of colour attributes and recipe/reproduction forecasting The reproduction of a specific shade or colour is a fundamental part of textile manufacture. Commercially successful products and designs require the concept of colour to be recreated in the closest manner possible so that the holistic effect of the cut, shape, pattern and texture are carried into the retail product. Because of the way in which inspirational colours are found or accumulated by designers, from articles as varied as feathers, minerals, ceramics or other surfaces, the task of interpreting these into a dyed textile has fallen to the dyer working in collaboration with the designer. The dyer uses expert knowledge of practical dyestuff combinations and effects to predict or create the closest match to the vision of the designer. Sometimes a close approximation is not possible due to the technical limitations of process or materials – but usually a compromise can be found if the communication is good between the parties. Some apparel producers try to reduce the difficulties of recreating a designer’s colour choice by insisting that only textile materials are put forward for development – this means that there is a greater likelihood of being able to match the colour closely and that instrumental techniques may also be employed. As with any evaluation or assessment of colour, it is vitally important that the viewing conditions, i.e. the illumination and environment, are defined and standardised. Unfortunately, this is still not widely understood by noncolourists but failure to establish these basic rules can needlessly invalidate much effort and work. Ideally, one or more internationally recognised illuminants (such as D65, TL84, A (Tungsten), etc.) should be specified for viewing. 210
Quality assurance management for coloured goods
211
Samples should always be viewed in a neutral environment, such as in one of the commercially available colour assessment cabinets1 – remembering to shield the cabinet from exterior light pollution and ensuring that the clothing of the observer is not reflecting colour back into the cabinet! The physical condition of the samples also affects the perceived colour so the samples should preferably be at a constant temperature and humidity if at all possible. For textiles this is recommended to be 65% ± 2% relative humidity and 20° ± 2°C – the normal standard atmosphere found in physical testing laboratories. Where these conditions cannot be met in the working area, environmental cabinets,2 that maintain the correct temperature and humidity conditions and expose the samples to a consistent level of light prior to measurement can be used to condition the samples immediately prior to assessment. Once it is certain that the colour is being observed correctly, other factors essential to the design or product need to be clarified prior to production. • • • •
What is the material (or substrate) to be coloured? Will it need to be after-treated? How will it be cleaned? Will it be exposed to high levels of light (e.g. summer wear) or only be worn in the dark, or in nightclubs? • Will the light sources change during normal wear (indoors/outdoors)? • Are other coloured materials to be used in combination? The answers to these questions help the producer to select the most appropriate materials and processes to meet the specification. Historical or experience-based forecasting The most common way of developing a recipe for a new colour has been to review the results of previous recipes – the historical shade library. All dyers retain records and samples of previous dyeings and these have the advantage of being ‘known’ properties. They are ‘known’ to have produced a given shade under the local conditions found in the dyehouse. If they are deemed to have been ‘successful’ recipes, they will be reliable, reproducible and efficient means of obtaining given colours with the available dyestuffs, chemicals, water, equipment and substrates. In absolute terms, they may not be the cheapest or best – but they will work predictably. By finding colours similar to the required one and adjusting the relative ratios of elements in the recipes, it is usually possible to find a reliable recipe fairly quickly – and the behaviour of the recipe can be predicted reasonably well so that unforeseen production problems do not occur. With an experienced dyer, this method can be a reliable and quite a speedy way of recipe generation; unless a new colour happens to be com-
212
Total colour management in textiles
pletely outside the gamut (colour range) of previous experience or if the substrate is new with unknown behavioural characteristics. In these cases it is necessary to go back to basics in the laboratory and develop a recipe empirically. Empirical recipe development Because of the number of variables affecting the colouration of materials, it is essential to establish a valid experimentation method that will allow predictions to achieve a reasonable probability of success. In the case of application of dyes to textile materials, the methods need to assess the colouration or yield effects of a particular dyestuff in isolation and in combination with others, on typical or actual substrates (in the state or form that they will be finally processed), using the same water and chemicals as the industrial process. Whilst this preliminary work on the ‘raw materials’ can be accurately and consistently performed in the laboratory, to find the ‘working recipe’, the scaling-up of the recipe to ‘bulk’ often requires changes in order to achieve the same result. These changes are necessitated by: • the different fluid dynamics inside large-scale machines compared to laboratory-scale equipment; • the physical state or positioning of material in the dye liquor and the degree of control existing in the system and plant – slight temperature variations, pump rates and pressures, liquor ratios, etc.; • the small but inherent variations in water quality, chemical compositions and dosing. The difficulties of the move from the ‘lab-dip’ to ‘bulk production’ can often be underestimated by those not involved in production and may lead to some unrealistic expectations. However, most successful dyers understand and can predict the differences in their own circumstances and therefore ensure that the laboratory work takes this into consideration or that the bulk recipes are suitably modified. To undertake empirical recipe development is an important but expensive and time-consuming exercise. Some organisations choose to limit this work – and avoid the errors or mistakes that can happen with inexperienced staff – by ‘subcontracting’ this work to professional laboratories or by using the services of dyestuff companies. Third-party or subcontracted recipe development Most, if not all, dyestuff manufacturers offer customers their services in the form of recipe development or advice using their products. The advantages
Quality assurance management for coloured goods
213
of such a service are that the recipes are usually ‘optimised’ for reliability and performance and the quality of the result with regard to colour fastness and durability is usually indicated. If the service provider uses water from the same source and all the same chemicals, the results will be very reliable. It will just be the time and scaling elements that will need to be considered. Dyestuff suppliers offer development services for major clients with a rapid turnaround (i.e. days rather than weeks). Smaller clients may have to wait significantly longer.
11.2 Instrumental or computer recipe prediction One of the advantages of computer-based systems is the ability to retain a total history of empirical, practical and imported data for purposes of predicting likely colour outcomes of specific colourants, methods of application and substrate combinations. These databases are generally site specific and peculiar to the local factors in play (local water, actual equipment design, stocked dyestuffs, available chemicals, regular substrates). Importantly, they are also normally an average or consensual value of the work of technicians and colourists and therefore more reliable, being less prone to bias or oneoff situations.
11.2.1 Target measurement As with all instrumental methods, the care taken in measuring the target will repay itself by requiring fewer iterations and corrections to achieve the desired end result. In other words, if the target is not measured properly, it is rather pointless trying to use the data for predicting anything. The importance of technique has been well documented elsewhere but it is always worth remembering that more errors are made due to inconsistent or poor measurement technique than probably all other instrumental calibration or correlation factors combined. Firstly – can the target actually be measured accurately? Is the sample large enough to be able to present an opaque, representative area to a spectrophotometer?3 Representative in this case means large enough (folded if necessary) to be able to use an aperture sampling size that is consistent with the material texture and form so that repeated measurements are comparable (typically a difference between readings £ 0.3 DE CMC 2 : 1). Note that although the choice of a colour difference formula to be used by instrumental colour measuring systems is not discussed here, it is essen-
214
Total colour management in textiles
tial that the same formula is applied by all parties sharing or using the same colour data and systems to make comparisons or judgements. With most commercial spectrophotometers it is possible to change aperture size to suit the available samples. However, it must be remembered that readings from different aperture sizes are not generally comparable due to the differing amounts of light falling on, and reflected from, the samples during measurement. Sometimes the structure or shape of a desired target may make meaningful measurement impossible – for example, a feather cannot be easily measured with a spectrophotometer. However, recent developments in the use of digital cameras, standardised viewing conditions and software to create reflectance data, mean that in future many more items – previously thought to be impossible to measure – may be able to be viewed and used to target colour predictions and matches.4 Secondly – has it been conditioned? Much work has been published regarding the change of perceived or measured colour due to the effects of heat and moisture. It is therefore important to ensure that measurements are only performed on materials that are in equilibrium with a standardised atmosphere (as referred to in Section 11.1.1 above). Similarly, light will affect the behaviour and appearance of certain colourants so measurements should only be made on samples that have been exposed to normal or standard levels of light radiation so that they are in a stable condition. Because of this sensitivity to light with some materials, repeated measurements with a spectrophotometer xenon flash may also change the colour being measured, so care must be taken when deciding how many readings (flashes) to take to obtain a specific reflectance value. Thirdly – does the target have any common physical or material features with the predicted substrate? Widely differing surface characteristics, coating chemistry and pigment types may make any prediction non-viable due to the lack of similarity in such critical factors as appearance, colour component combinations and surface measurement geometry with the proposed manufactured product.
11.2.2 Predictive systems Predictive systems have been available for many years from various providers of colour applications.5 Essentially, they all provide the same function
Quality assurance management for coloured goods
215
and allow a recipe to be created and then modified, based on measurement of the actual results. Some systems have been enhanced to ‘learn’ or become ‘expert’ in the results achieved and to modify the first ‘pass’ or prediction thus increasing the likelihood of a first-time success and reducing the need for further iterations. Any predictive system depends on accurate base data for the generation of suitable recipes. As with empirical recipe development, the use of representative data is fundamental to the accuracy and usefulness of the system. However, with a digital system using only one data set for all calculations, the accuracy of this data set is crucial. Recently, there have been web-based prediction systems established on the Internet6 and these can be used to locate and identify ranges and mixtures of dyes that offer the promise of a close match with identifiable and comparable matching under different illuminants. These tools can be used to help dyestuff manufacturers offer a rapid ‘virtual’ sampling service to existing or would-be clients – prior to them investing in the laboratory time and cost to create a working recipe from the most suitable dyes. The tools can also be used to share accumulated knowledge and methodology across multi-site operations, avoiding the duplication of effort (and eliminating the cost and errors of that effort).
11.2.3 Primary data creation or sourcing Textile dyes exhibit different performance and yield attributes depending on their concentration in the application medium, the quantity actually taken up by the substrate and the combination of different quantities and types of dyes used to achieve a given shade. Advice on these characteristics should be obtained from the dyestuff manufacturers so that needless work, and fundamental errors, can be avoided. For a prediction system to generate reliable results, it is important that the colour yield attributes are accurately known over the full range of commercial application rates, i.e. the quantities of dye known to provide acceptable levels of fastness or stability on the substrate. This primary data is generated by completing a series of monochromatic dyeings using the ‘standard’ substrate, water, chemicals, liquor ratio and dyeing method to be found in the laboratory or production unit, with an evenly spaced range of dyestuff concentrations between the recommended minimum and maximum application rates. The ultimate accuracy of this data is always going to be determined by the number of correct and representative dyeings performed and measured into the system. Smooth and accurate prediction curves are unlikely to be achieved with fewer than 12 gradations – although as few as six can be seen in practice.
216
Total colour management in textiles
The work required to achieve the required level of accuracy should not be underestimated, and high levels of personnel training and care are essential if the database is to be reliable, and therefore usable. Automated dispensing systems7 can certainly help to eliminate some of the work and errors. Although these can be relatively expensive for smallto medium-sized producers, they do save much time and cost. However, they do depend on the same high level of operator care to load and maintain the standard solutions of dyestuffs used in the dispenser. Alternatively, the work can be subcontracted to professional laboratories who have been supplied with all the information and materials (water, dyes, chemicals and substrates) to permit the generation of data specific to site condition. Whilst these data will be consistent and accurate, it is essential that the currency of these data is ensured and maintained by constant updating whenever material changes are made to dyes, chemicals, process conditions, equipment or substrates.
11.3 Colour variation evaluation and monitoring Commercial reproduction of coloured materials is only possible if the quality and repeatability of a colour can be assured. Systems to enable this control are essential elements of any manufacturing process.
11.3.1 Establishment of colour standards Practical colour standards need to be created in order for any colouration process to be controlled effectively. This means creating colour standards that are relevant and consistent with the products and processes being used, essentially by using the same colourants and substrates as the production items. Colour standards are often produced ‘in-house’ as laboratory dyeings or special small-scale production lots, depending on the requirements and use of the standards. For internal use, these can be stored as fibre tufts, yarn windings or fabric swatches, and are usually measured into a colour library database for routine usage in controlling production, the ‘original’ physical standard being filed, purely as a reference. However, colour standards are often created commercially for specifiers or buyers and distributed (or sold) to the various producers in the supply chain. These so-called ‘engineered standards’ are actually small production runs on standard substrates – or special substrates to order.8 These engineered standards are often supplied with the recipe (or at least the dyestuff concentrations) used to create the colour, as well as the standard reflectance value – which is effectively the mean of all the values of
Quality assurance management for coloured goods
217
a standard measured on a reference quality spectrophotometer under prescribed conditions. The coloured materials are checked for consistency prior to sale or shipment and are warranted to be within a certain colour matching tolerance to the master reflectance value reading (typical examples may be £ ± 0.5 DE CMC 2 : 1 from the mean value). Whilst these engineered standards are invaluable for enabling buyers and agents to specify and evaluate the purchased materials, they may constitute merely a guide, albeit a very detailed and accurate guide, to the production unit. Most production units have a different source of raw materials to those used for the engineered standard and will therefore create a local production standard, based on the engineered standard, but using local substrates and dyestuffs normally sourced by that unit.
11.3.2 Ensuring continuity and constancy Once a colour standard has been established, every production lot or batch must be compared to this and be within an agreed commercial tolerance before it is approved for release. If the colour standard is a local one, matched to a client’s engineered standard, both standards should be used when assessing the acceptability of a particular lot. A good rule of thumb is that the colour of any lot should always lie between the two standards. This eliminates any tendency to drift away from the colour expected by the client. The maintenance of continuity records for every production lot is essential in order to achieve this level of control. Modern computer-based systems simplify this process by providing measurement and reporting features able to display current and historical data simultaneously. With larger, multi-unit groups or buyers requiring production from more than one manufacturer, it is essential that common data is shared and used to ensure colour consistency across the supply chain. Effectively, this means maintaining a single colour standard for any given product – no matter how many production units are involved – and for each and every dye lot to be matched against this standard and every other dye lot produced to the same colour! The result is a single, multi-unit continuity record, where all units are linked in real time over the Internet so that no drift or pass/fail disagreement can occur.9 All measurement data must be typical of the material being produced and steps must be taken to ensure that production batches are coloured consistently and evenly throughout. For continuous dyeing and finishing ranges, on-line systems are available and are being developed using the latest remote and no-contact sensor technology with high-speed sampling.10
218
Total colour management in textiles
As with all colour measurement, it is vitally important to ensure that the substrate condition (i.e. temperature and humidity) is regulated or known during measurement if the results are to be meaningful and comparable.
11.3.3 Practical re-standardisation Sometimes it is necessary to change a local standard for production – even if the master or engineered standard has not been changed. This can be caused by the temporary or permanent unavailability of one or more of the original raw materials, such as the withdrawal or lack of colourants, or perhaps a change in substrate composition, e.g. new cotton crops, synthetic merge changes, etc., or even a forced change in production conditions: water supply problems, plant breakdown or closure, etc. Whatever the cause, the result is a practical and technical inability to reproduce the exact colour conditions previously established. A new local or production standard must be created to control subsequent production but the impact of this change needs to be carefully evaluated and considered. Sometimes the change may be so small as to be within expected normal process variation. This is not usually an issue. At other times, the result may be a significant move from the original standard under one or more illuminants. In these instances, it is essential to advise the client, and probably the rest of the supply chain, so that the impact can be assessed and measures taken before unexpected costs and problems are incurred. Even if the adjustment is accepted in principle by the client, a sudden arrival of off-standard goods, which may need to be re-matched by other processors and trim suppliers, can cause serious problems and delays to product manufacture.
11.3.4 Instrumental systems and methods Colour standard and continuity maintenance are best achieved by the use of instrumental measuring systems and digital record keeping. These are the only means of ensuring consistent application of objective decision criteria and the preservation of colour records in a state free from degradation due to handling or poor storage. Spectrophotometric measurements of materials prepared and presented in standard forms create highly consistent and reliable data for use as standards and in continuity assessment. Any data used for colour assessment need to contain all the important parameters of the measurement conditions so that any differences can be noted and the comparability, or not, of the results taken into consideration. Ideally, these data should show:
• • • • •
Quality assurance management for coloured goods
219
the type and model of the measuring instrument; the calibration data condition; the choice of measuring aperture; the use of any UV or other filtering in the sample illumination; the percentage reflected light at given specified intervals across the ‘visible’ spectrum.
Because of the different designs, ages and condition of spectrophotometers in use throughout the industry, some extra method (other than normal calibration) of assessing comparability of results had long been felt to be desirable. Internet applications are now available to check the performance of instruments against a virtual standard and to correct the output, correcting any drift or degradation in performance.11 Whilst it is important that instruments are correctly maintained and calibrated, it must be remembered that the biggest influence on the accuracy of spectral data is the human operator! Web-based applications that can correlate and compare the actual results of the same samples measured by different users12 should also be considered as another essential part of a quality assurance scheme to ensure the reliability of data. Once accurate digital information is available, these data can be used to generate consistently uniform images on calibrated CRT monitors, allowing identically coloured images to be viewed simultaneously at any location. Such techniques are excellent aids to describing colour relationships and variations. Photorealistic images can also be incorporated into these systems allowing ‘texture’ to be seen and taken into account when colours are modified or changed.13 The use of digital camera technology allows ‘real’ images to be presented rather than facsimile masks but one limitation of all these systems lies in the inability to show a complete gamut or range of colours. High quality monitors can provide stable, consistent and reproducible colours but the actual appearance of a coloured image on a glowing monitor may not look like the original fabric in a light assessment cabinet. For quality assessment and control purposes this may not matter as the comparing images will only be seen ‘on screen’ as relative colours and the ‘real’ information for control is digital. Many other factors affect the accuracy and reliability of spectral data and codes of practice, or other measuring conditions, should also be specified in the data or taken into account. Colour record files are available in many formats from different manufacturers or different equipment and systems and contain varying amounts of information. There has been some rationalisation in the types used, usually driven by retailers wanting commonality in their systems, but the creation of a single file format and colour difference formula for use by everyone has yet to be achieved.
220
Total colour management in textiles
At the time of writing, work was in hand by the Society of Dyers and Colourists and by the American Association of Textile Chemists and Colorists to create an international form of data record that would include all the relevant factors. In addition to spectrophotometric measurement of colour, the use of digital cameras to capture coloured images in standardised conditions, which can then be processed to create real or synthetic, i.e. software derived, reflectance curves for use in comparison with ‘traditional’ spectral data is also available. Data storage and communication is also being developed rapidly away from so-called stand-alone systems, where colour data was created by local measurement only, to open Internet connected systems where colour data can be stored, shared, searched and exchanged in the same way as financial data. It is this paradigm change in communications that now enables instrumental systems to be used creatively in the global manufacturing industry for control and standardisation.
11.4 Colour performance The management of colour must include the delivery or assurance of colour performance in addition to the colour reproduction. Coloured materials are expected to retain their colour properties for the service life of the product. For example, a disposable napkin will have a colour quality requirement that is totally different to an Atlantic trawlerman’s waterproofs but they both have to deliver user satisfaction.
11.4.1 The quality factor – fitness for purpose The quality of any item is judged by fitness for purpose, and this is no different for colour properties. Fitness for purpose in colour terms usually means a retention of appearance in keeping with expectation, and this varies enormously in apparel. For example, indigo-dyed jeans are expected to ‘wash down’, but a black formal shirt or sweater will be expected to stay black after washing. The quality of materials used and the expertise required to apply the colourants is probably no different between the two, but both need to be applied consistently and reliably so that the product will behave as expected by the consumer. How to ensure this performance at a known cost is the question to be addressed when considering fit for purpose quality. Contributing factors Many things contribute to the colour quality of an item but the main areas usually considered in textile apparel centre around the choice of materials,
Quality assurance management for coloured goods
221
the methods of process and the abuse or otherwise of the finished product during its service life. Raw materials The choice of the correct raw materials is fundamental to any manufacturing process. Whilst it is not normally possible to ‘make a silk purse out of a sow’s ear’, neither is it guaranteed to make the best product from the most expensive materials available. Raw materials need to be selected, based on the required final performance and acceptable cost. Processes used in colouration invariably have a negative effect on raw materials due to the chemical, physical and thermal treatments involved. Raw materials also need to be selected so that they will withstand these processes in an acceptable fashion. Fibres, yarns and fabrics should be of an acceptable form, strength and constitution that will withstand the mechanical stresses imposed when wet and/or hot. They should be consistent or homogeneous in composition so that colourants are absorbed and retained regularly and in a predictable manner. Process materials similarly should be of consistent quality; this includes everything from water, which should be adequately filtered and treated, through the chemicals and additives that must be of consistent purity and strength, to the colourants used to achieve the desired shade. Colourant types and chemistry are substrate dependent, but the consistency and quality of these products with regard to their purity and physical form and preparation (powders, grains or liquids) is of paramount importance. High quality materials can be compromised by poor storage and housekeeping, which will reduce the reliability and quality of colour obtained. Apart from cleanliness and cross-contamination concerns, environmental factors are tremendously important to dry dyestuffs as they may be hygroscopic (thus affecting the colour yield per unit of mass). In textile dyehouses, which can be hot and humid, the colourant storage areas should be climate controlled to reduce the chance of moisture absorption, which would affect the dosage and may even react with the materials, reducing their efficacy still further. Liquid dispersions and solutions should be carefully controlled to ensure homogeneity and freshness, so that results remain predictable. Application techniques The accurate application of colourants is a science and technology of extreme breadth and depth. For consistent quality, these techniques must be followed to the letter. The vast number of variables involved in the col-
222
Total colour management in textiles
ouration process demands that every possible attempt is made to standardise and control as many factors as possible. The design and automation of colouration equipment has improved this area immeasurably but care still needs to be exercised. This not only means controlling the mass of substrate, the concentrations of colourants and the time/temperature profiles, but also the standardisation in design of the equipment and the consistency of the material packages, beams or pieces. Ideally, repeat colours should always be dyed in the same machine as this will reduce the risk of other variables coming into play, which may result in a re-process or re-dye, which will add cost and time to the process as well as reducing the quality of the end product! The results of poor or inconsistent application techniques are often evidenced by; uneven and inconsistent dyeing, poor colourfastness and damaged fibres – in addition to extra costs and delays to market. After-treatment and finishing Product colour quality is often ‘made or broken’ by post-dyeing treatments. For some dyeing processes, an after-treatment may be used purely to remove unabsorbed or unreacted colour from the surface of the fibres, providing a ‘clean’ surface that will not contaminate or stain adjacent materials. This process may also ensure that the ‘pure’ colour is observed. Other aftertreatments may add chemicals, which combine or react with the colourants to make them more resistant and long lasting. More frequently, after-treatments are applied to change or modify the fabric properties and handle. These finishes can affect both the appearance and durability of the colour and so need to be carefully considered before use. Fabric finishing may involve the use of heat, moisture and pressure to achieve a particular effect. Care must be taken that the colourants used are suitable for these processes and able to resist induced changes due to these factors. For example, colourants that have a low sublimation fastness should not be used in fabrics that will be treated at high temperatures in curing or pressing. Aftercare and ‘wear and tear’ The selection of colourants must take into consideration the final destination of the product, and how the purchaser will expect to be able to treat the product. Normally, textile materials are washed to clean them, and the quality of colour must be sufficient to allow this process to take place. International
Quality assurance management for coloured goods
223
conventions on the labelling of materials exist so that suitable washing and cleaning processes can be recommended for various types of textile materials and construction.14 Rapid colour loss and staining in normal washing cycles are evidence of poor colour quality and inferior processing. Certain luxury fibre types and colourants are more susceptible to damage by incorrect cleaning but these should be clearly marked by the correct labelling. The application of labels stating that ‘this garment should not be cleaned’ has been evidenced but is really an indication of poor design and inconsideration by the manufacturer, unless the garment is a design statement or work of art only to be used for display. All products will suffer colour degradation due to ‘wear and tear’ but the relative rate of change can be designed into the product by the choice of materials and methods of manufacture. This effect should be consciously built into garments and assessed by wearer trials so that the product can be adequately labelled and marketed. Colour durability Assuming that colour has been applied consistently and at the required quality level, the retention of the colour is subject to several factors, which are outside the control of the manufacturer. Abrasion and wear Textile materials are flexed and abraded in use and this mechanical action gradually ruptures and removes the surface of the fibres and the fibres themselves. The type of fibre and method of colouration will affect the change of appearance and loss of colour due to this process. Certain fibres and yarns may be coloured consistently through the material so that erosion or abrasion does not remove colour; whereas other fibres and yarns may be surface dyed and relatively little abrasion may eliminate the coloured effect.15 Typically, colour loss effects are welcomed in fabrics such as denim but are usually to be totally avoided in dark woollen garments, for example. Washing and cleaning As mentioned previously, washing and cleaning can have a drastic effect on colours that are not well bonded to substrates;16 but colour loss can be attributed to other factors as well. Many commercial detergents include enzymes and fluorescent brightening agents (FBAs) in their composition. Enzymes will erode certain fibres
224
Total colour management in textiles
(particularly protein fibres such as wool and silk) under given conditions and this may cause a change in colour (as well as a structural weakening of the fabric!). FBAs are added to some detergents to brighten whites and colours. However, if these products are applied to dark colours, they effectively ‘fade’ the colours by masking the shade with a lightening effect. These effects are cumulative and proportional to the amount of exposure during wash and wear cycles. Modern detergents are usually labelled as to their suitability and alternative products without enzymes or FBAs are marketed for all fibre types, but consumer education in this matter is still wanting, so colours are still at risk! Light and environment Electromagnetic radiation in the form of visible and ultraviolet light can have a marked effect on colour retention and on the catalytic degradation of fibres. When designing coloured materials for environments with high exposure to light, it is essential that the correct fibre/colourant combinations are employed and suitably tested.17 The environment existing during light exposure is also critical as this may accelerate or modify the effects. For example, the effects of sunlight in the desert will be totally different from the same intensity at the coast.
11.4.2 Testing methodology The evaluation of colour quality or durability is traditionally performed by trained observers using standard test methods. Usually, these tests entail various processing of test fabrics (normally a multi-fibre strip, composed of bands of common textile fibres) in contact with a dyed sample of known dimension or mass. The two components are then assessed for colour loss and/or transfer (staining). However, the subjective nature of this method has long been recognised as a concern and more objective means are desirable. Traditional (subjective) assessment systems For the assessment of colour loss or staining, special scales of grey coloured tiles or chips are used in standard methods. One set (grey to grey) is used for colour change or loss and another set (white to grey) is used for assessing the staining of adjacent fabric. The results are quoted in half steps from 1 (maximum staining or loss of colour) to 5 (little or no change or staining).
Quality assurance management for coloured goods
225
The grey tiles are grouped in pairs of increasing contrast. The skilled observer uses these scales to assess a similar degree of contrast between a sample ‘before and after’ test. The method of illuminating and viewing the samples and the scales is defined by the test methods but the results on a given pair of samples can be slightly different from observer to observer due to differences between the observers! This is especially so in the middle ranges, and this is where the most of the problems lie! There is usually no disagreement about little or no staining or colour loss; nor for severe staining or colour loss. The difficulties arise in the 3 to 4 grade regions where a buyer’s grading may be lower than a seller’s and the commercial implications are serious! A more consistent method would be preferable. Instrumental (objective) assessment systems Digital camera-based systems that are able to assess staining and colour change are now available and have been proved to be far more consistent than human assessment across a range of samples and laboratories. One of the pioneers in this area is the DigiGrade system.18 The DigiGrade equipment consists of a standardised sample mounting and illumination system, which is scanned by an accurate digital colour camera. The images are then processed by sophisticated software and the results shown on a calibrated monitor together with their fastness rating. Being digital, these data and images can be copied or shared via computer systems to other DigiGrade installations, allowing all parties to see the actual results.
11.5 Future trends As production areas become ever more dispersed from the final consumer market and design centres, the importance of fast, reliable communication increases. Even within the manufacturing supply chain there can be multiple levels of fabrication and processing that are separated geographically and culturally and need common technology and data formats to enable processes to be managed. The Internet increasingly will provide the means to bring all the disparate elements into a cohesive structure. This impact of web communication will be seen at all levels of endeavour from the remote monitoring and control of production machinery, garment assembly and identification to the tracking and delivery of goods to the final consumer. The use of linked web services allows simply tasked systems to use queries and searches in extremely complex and powerful ways at a fraction of the resource and cost required for individual stand-alone systems.
226
Total colour management in textiles
Design software will use the same source colour data shared with production, quality assurance and logistics systems; there will be no need to duplicate or copy data, merely to access and share it. Once a colour has been measured and approved, it can be eternally linked to the product containing that coloured material. This means that true and reliable colour becomes a product attribute that can be managed and specified, just as fabric weight, yarn titre and garment sizes are used in product codes. The agreed establishment of standardised colour evaluation systems (colour difference formulae and means of recording reflectance data) will be necessary to provide truly global communication but the technology is already capable of delivering the data; it is just the development of the human interface that will impede or hasten these moves.
11.6 Notes and references The list of contacts given below is representative and not exhaustive. 1. Suppliers of colour assessment cabinets: www.datacolor.com www.gain.com.tw www.gretagmacbeth.com www.verivide.com 2. Suppliers of environmental cabinets: www.vindon.co.uk 3. Suppliers of spectrophotometers: www.datacolor.com www.gretagmacbeth.com www.hunterlab.com www.konicaminolta.com www.xrite.com 4. Suppliers of digital camera-based colour measurement: www.color-aixperts.de www.digieyeplc.com 5. Suppliers of recipe prediction systems: www.datacolor.com www.ewarna.com www.gretagmacbeth.com 6. Suppliers of web-based recipe prediction systems: www.ewarna.com 7. Suppliers of laboratory dispensing systems: www.datacolor.com www.gain.com.tw 8. Suppliers of engineered standards: www.archroma.com www.dystar.com
Quality assurance management for coloured goods
9. Suppliers of web-based colour QC software: www.ewarna.com 10. Suppliers of non contact measuring systems: www.hunterlab.com www.xrite.com 11. Suppliers of spectrophotometer correlation software: www.datacolor.com www.gretagmacbeth.com 12. Suppliers of web-based user correlation application: www.ewarna.com 13. Suppliers of textured imaging software and systems: www.datacolor.com 14. ‘Caring for Your Clothes’ guide. www.asbci.co.uk 15. Test method for abrasion (example): BS EN ISO 12945-2:2000 16. Test methods for fastness to washing and dry cleaning: BS EN ISO C06:1997 BS EN ISO D01:1995 17. Test for fastness to light: BS EN ISO 105 B02:1999 18. Supplier of instrumental colour fastness rating equipment: www.digieye.com
11.7 Sources of further information Society of Dyers and Colourists www.sdc.org.uk American Association of Textile Chemists and Colorists www.aatcc.org
227
Index
AATCC guidelines 84–5 absorption of light 9, 139 accuracy of colourant formulation 147–52 of digital colour printing 179–88 of instrumentation 47 observer accuracy 66, 106 acid dyes 172, 174 additive mixing 14–15, 26–7 after-treatments 174–6, 222 aftercare 222–3 appearance of colour 19–20, 36–8 appearance of samples 87 application of colourants 221–2 approval systems 123, 130–3 artificial neural networks 147–52, 156–7 ASP (application service provider) model 198, 201 ASTM D1729 guidelines 85–6 attainable colours 167, 180–1 automated dispensing systems 216 benchtop instruments 49–50 best practices 201–3 binary continuous inkjets (CIJ) 166–7 blackbody 8 body colour calculation 110–11 business and marketing model 189 calibration database 140–2, 219 cameras 51–2, 55, 219 carpets 90 choosing colours 194 chromacity diagrams 37–8 chromatic adaptation 18–19, 71–3 CIE94 colour difference formulae 61–2 CIE system 17, 19–20, 24–40 ASTM guidelines 86 colour difference formulae (CIELAB) 58–60 colour matching functions 31–2 illuminants 30–1, 32, 34–5
light sources 30 primary colours 15, 27–8, 29–30 reflectance measurements 35, 36 standard observers 31–2, 35 tristimulus values 28, 29, 32–4, 36 and colour appearance 36–8 usefulness and limitations 38–40 viewing conditions 32, 35, 82–3, 211 see also description specification systems CIEDE2000 colour difference formulae 62–5, 73 CIELAB colour difference formulae 58–60 CMC colour difference formulae 60–1 CMYK colour system 167, 180–1 colour cathode-rays (CRTs) 98 colour difference evaluation 58–65, 68, 73, 153 colour display characterisation 98 colour engines 207 colour loss 176, 223, 224–5 colour mapping algorithm 100–2, 113–14 texture images 99–100 colour matching functions 31–2 colour simulation 97–115 body colour calculation 110–11 colour display characterisation 98 colour synthesis 108–13 dichromatic-based modelling 109–10 grey-scale comparison method 105 image synthesis 97 mapping algorithm 99–103, 113–14 texture effects 103–8 colour sorting 153, 196 colour synthesis 108–13 colour vision 13–15 testing 81–2 communication choices 197–8 manual 121–4 technological 124–7 computer formulation 136, 213–16
229
230
Index
concept colours 120–1 cones in eyes 11, 12, 13–14 response equation 13 constancy of colour 69–73, 217–18 chromatic adaptation 18–19, 71–3 and the human vision system 13, 18–19 inconstancy index 70, 71 metameric matches 70 spectral matches 70 contrast 17–18 corduroy 90 cornea 10 correlated colour temperature 8 custom instruments 52 daylight distribution 8 DDP (Direct Digital Printing) 178 description specification systems 22–42 additive mixing 14–15, 26–7 chromacity diagrams 37–8 Munsell system 25, 40–1 naming of colours 24–5 NCS (Natural Colour System) 41 ordering systems 40–1 specifier systems 41–2 subtractive mixing 26–7 see also CIE system; measurement systems design of colours and digital printing 176–8 see also mind to market communication detergents 223–4 dichromatic-based modelling 109–10 difference evaluation 58–65, 68, 73, 153 diffuse reflectance 9, 45 diffuse transmittance 46 DigiGrade system 225 digital capture technology 177 digital colour printing 160–90 accuracy and uniformity 179–88 attainable colours 167, 180–1 business and marketing model 189 colour loss 176 colour management system 181–3 design potentials and limitations 176–8 dither of colour 180 drop size 171 environment 174 fabric handling 169 fabric structure 172–3 fibre type 172 head height 170–1 ink placement 170 ink type and characteristics 171–2 multi (binary) continuous inkjets (CIJ) 166–7 paper-backing 173–4 photo-realistic printing 177 piezoelectric drop on demand (DOD) inkjet 165–6
post-treatment 174–6 pre-treatment 173–4, 179 print head movement 170 print pass 170 print speed 169–71 printing width 169 production printing 179 raster image processors (RIP) 183–8 repeat designs 177 resolution 171 for sample production 178–9 software 167, 178, 181–8 thermal drop on demand (DOD) inkjet 160–5 three-dimensional forms 177–8 washing printed fabrics 176 workflow issues 188 Direct Digital Printing (DDP) 178 disperse dyes 172, 174 display characterisation 98 distribution of daylight 8 dither of colour 180 drop size 171 durability of colour 223–4 Dvorine test 81 dyes 172, 174 electromagnetic radiation 7 electronic sample evaluation 132–3 electronic tracking 134 Encad 165 engineered standards 127–30, 216–17 enzymes 223–4 eyes 7, 10–13 fabric handling 169 fabric structure 172–3 FabriJet 169 fast fashion concept 118–19 fibre type 172 finishing 222 fitness for purpose 220–4 flare 79 flat woven textiles 89–90 fluorescent colourants 147, 150–2 fluorimeters 52–3 formulation of colourants 136–57, 211–13 accuracy 147–52 artificial neural networks 147–52, 156–7 calibration database 140–2 colour difference evaluation 153 computer formulation 136, 213–16 fluorescent colourants 147, 150–2 historical shade libraries 211 Kubelka-Munk theory 137–9, 147, 156–7 pigmented systems 138 recipe correction 146–7 shade sorting 153
spectrophotometric matching algorithms 142–6 third-party development 212–13 trial and error process 136 frequency of purchase 119 gain-offset-gamma (GOG) model 98 ganglion cells 16 geometries of measurement 44, 46 Gild the Lily 178 goniometric instruments 51 grey-scale comparison method 105 hand-held instruments 48–9 head height 170–1 Helmholtz reciprocity 45 historical shade libraries 211 human vision system 7, 10–17 colour vision 13–15 testing 81–2 constancy of colour 13, 18–19 eyes 7, 10–13 spatial vision 15–17 humidity 81 illuminants 30–1, 32, 34–5, 77–9 image synthesis 97 inconstancy index 70, 71 industrial colour tolerance 65–8 information requirements 197 initial colour development 87–8 ink jet technology ink placement 170 ink type and characteristics 171–2 multi (binary) continuous inkjets (CIJ) 166–7 piezoelectric drop on demand (DOD) inkjet 165–6 thermal drop on demand (DOD) 160–5 see also digital colour printing instrumentation 47–53 accuracy 47 and approval systems 131–3 benchtop instruments 49–50 calibration and maintenance 219 cameras 51–2, 55, 219 custom instruments 52 fluorimeters 52–3 goniometric instruments 51 hand-held instruments 48–9 inter-instrument agreement 53–4 light boxes 48 multiangle instruments 49 portable instruments 48–9 precision 47 scanning instruments 50–1 single-scale instruments 47–8 software for 54
Index
231
specifications 47 spectrophotometers 9–10, 97, 133–4, 142–6 sphere-based 49 visual instruments 48 see also measurement systems inter-instrument agreement 53–4 Internet, online shopping 119 iris 10–11 Isihara test 81 JPC79 colour difference formulae 60–1 knitted textiles 89–90 Kubelka-Munk theory 137–9, 147, 156–7 lab-dips 195 laboratory recipe correction 146 lead times 203–4 light 7–10 absorption 9, 139 correlated colour temperature 8 diffuse reflectance 9 distribution of daylight 8 measurement of coloured light 25 scattering 9 surface reflectance 8–9, 109–10 wavelengths 7–8, 29–30 light boxes 48, 79 light degradation 224 light sources 30, 77–9, 82–3 AATCC guidelines 84–5 evaluation and quality 79 secondary sources 78–9 selection 78–9 standard illuminants 78 liquid-crystal displays (LCDs) 98 loose fibres 90 maintenance of instruments 219 management systems 181–3 see also supply chain manual communication 121–4 mapping algorithm 100–2, 113–14 marketing model 189 measurement systems 23–5, 44–55, 218–20 coloured light measurements 25 electronic sample evaluation 132–3 geometries of measurement 44, 46 instrumentation 47–53 inter-instrument agreement 53–4 physical sample evaluation 131–2 radiance factors 46 reflectance measurements 44–5 sample-induced effects 46–7 traceability 53–4 transmission factors 46 transmittance measurements 44, 45–6 see also CIE system
232
Index
metamerism 68–9, 70, 79 ASTM guidelines 86 mind to market communication 117–35 approval systems 123, 130–3 concept colours 120–1 electronic tracking 134 engineered standards 127–30, 216–17 fast fashion concept 118–19 manual communication 121–4 palette development 120–1, 125–7 reporting packages 134 shopping habits 118–19 supplier accreditation 133 target colour distribution 121–2, 129–30 technological communication 124–7 see also supply chain mixing colours 14–15, 26–7 MLP (multi-layer perception) 148, 150 multi (binary) continuous inkjets (CIJ) 166–7 multi-spectral imaging 205–6 multiangle instruments 49 Munsell system 25, 40–1 naming of colours 24–5 nature of colour 7–8 NCS (Natural Colour System) 41 neural networks 147–52, 156–7 observers accuracy 66, 106 metamerism 68–9 repeatability 66 standard observers 31–2, 35, 81 training 86 uncertainty 67–8 viewing environment 32, 35, 82–3, 211 see also visual evaluation online shopping 119 opaque materials 9 opponent processing 14–15 opsin 11 ordering systems 40–1 palette development 120–1, 125–7 paper-backing 173–4 perception of colour 7–20, 77–83 appearance 19–20, 36–8 contrast 17–18 human vision system 7, 10–17 light sources 77–9, 82–3 nature of colour 7–8 physical basis of colour 8–10 texture effects 103–8 three-dimensional nature 17, 25 see also constancy of colour photo-realistic printing 177 photopigments 11–13 physical basis of colour 8–10
physical sample evaluation 131–2 piezoelectric drop on demand (DOD) inkjet 165–6 pigmented formulation systems 138 pigmented inks 171 pile textiles 90 portable instruments 48–9 post-treatments in digital printing 174–6, 222 pre-treatments in digital printing 173–4, 179 precision of instruments 47 predictive systems 214–15 primary colours 15, 27–8, 29–30 primary data 215–16 principle of univariance 13 print head movement 170 print pass 170 print speed 169–71 printing width 169 production printing 179, 195–6 quality control 88–9, 195–6 recipe correction 146–7 pupil 11 purchase frequency 119 quality evaluation 57–73, 210–26 after-treatments 222 aftercare 222–3 application of colourants 221–2 colour difference formulae 58–65, 68, 73 colour loss 223, 224–5 constancy of colour 69–73, 217–18 durability of colour 223–4 finishing 222 fitness for purpose 220–4 industrial colour tolerance 65–8 light degradation 224 metamerism 68–9, 70, 79 observer accuracy 66, 106 observer repeatability 66 observer uncertainty 67–8 predictive systems 214–15 primary data 215–16 production quality control 88–9 raw materials 221 reproduction forecasting 210–13 target measurement 213–14 testing methodology 224–5 variation evaluation 216–20 washing and cleaning effects 223–4 wear and tear 222–3 wrong decision measures 66–7 see also working standards radiance factors 46 raster image processors (RIP) 183–8 raw materials 221 re-standardisation 218 reactive dyes 172, 174
recipe correction 146–7 recipe formulation see formulation of colourants reflectance measurements 35, 36, 44–5 regular reflectance 45 regular transmittance 45 repeat designs 177 reporting packages 134 reproduction forecasting 210–13 resolution in digital colour printing 171 retina 10, 11, 13–14, 15 rhodopsin 11 RIP (raster image processors) 183–8 rods in eyes 11, 12 roll grouping 196 samples appearance 87 conditioning 81 electronic evaluation 132–3 physical characteristics 80 physical condition 211 physical evaluation 131–2 preparation 80–1, 83, 85 production 178–9 sample-induced effects 46–7 testing 89 viewing conditions 32, 35, 82–3, 211 scanning instruments 50–1 scattering 9 SCOPE system 153–6 scotopic vision 13 secondary component matching 195 secondary light sources 78–9 shade libraries 211 shade sorting 153 shopping habits 118–19 simulation see colour simulation single-scale instruments 47–8 software 206–8 colour engines 207 for digital colour printing 167, 178, 181–8 for instruments 54 raster image processors (RIP) 183–8 sorting 153, 196 spatial vision 15–17 specifier systems 41–2 spectral matches 70 spectrophotometers 9–10, 97, 133–4, 142–6 sphere-based instruments 49 standard observers 31–2, 35, 81 standardisation 205 standards see working standards steaming 174 subcontracting 193, 212–13 subtractive mixing 26–7 supplier accreditation 133
Index
233
supply chain 117–18, 191–209 ASP (application service provider) model 198, 201 best practices 201–3 choosing colours 194 colour sorting 196 communication choices 197–8 information requirements 197 lab-dips 195 lead times 203–4 multi-spectral imaging 205–6 production checks 195–6 recent changes to 193–4 roll grouping 196 secondary component matching 195 standardisation 205 subcontracting 193 workflow 192, 204 see also mind to market communication surface reflectance 8–9, 109–10 target colour distribution 121–2, 129–30 target measurement 213–14 technological communication 124–7 temperature 81 correlated colour temperature 8 testing colour vision 81–2 testing methodologies 224–5 texture 99–100, 103–8 thermal drop on demand (DOD) inkjet 160–5 third-party colour formulation 212–13 threads and yarns 90 three-dimensional forms 108–13, 177–8 three-dimensional nature of colour 17, 25 traceability 53–4 translucency 9 transmission factors 46 transmittance measurements 44, 45–6 transparency 9 trial and error colour formulation 136 trichromacy 14–15, 17 tristimulus values 28, 29, 32–4, 36 and colour appearance 36–8 uniformity of colour 23 univariance principle 13 variation evaluation 216–20 velour 90 viewing environment 32, 35, 82–3, 211 visual evaluation 76–92 AATCC evaluation procedure 84–5 appearance of samples 87 ASTM D1729 guidelines 85–6 colour vision testing 81–2 of flat woven textiles 89–90 and humidity 81 industrial guidelines 83–6
234
Index
initial colour development 87–8 of knitted textiles 89–90 light sources 77–9, 82–3 of loose fibres 90 object being observed 80–1, 83 observer accuracy 66, 106 observer environment 83 observer training 86 physical and psychological influences 82 of pile textiles 90 production quality control 88–9 sample preparation 80–1, 83, 85 sample testing 89 standard observers 81 and temperature 81 viewing environment 32, 35, 82–3, 211 of yarns and threads 90
see also human vision system; perception of colour visual instruments 48 washing and cleaning effects 176, 223–4 wavelengths 7–8, 29–30 wear and tear 222–3 wool 140 workflow in digital colour printing 188 in the supply chain 192, 204 working standards 80–1, 123, 216–17 engineered standards 127–30 re-standardisation 218 wrong decision measures 66–7 yarns and threads 90