Aspects on Colour Rendering, Colour Prediction and

Colour Control in Printed Media

Marianne Klaman

Stockholm 2002

Doctoral Dissertation

Royal Institute of Technology

Department of Numerical Analysis and Computer Science

Media Technology and Graphic Arts

Thesis for the degree of Doctor of Technology to be presented with
due permission by the Royal Institute of Technology, for public examination
and criticism in Room D2, at the Royal Institute of Technology,
KTH, Lindstedtsvägen 5, Stockholm, on May 8, 2002, at 10 AM.
The thesis will be defended in English.

Opponent Professor Ulf Lindqvist

Examination committee Professor Mladen Lovrecek

Dr. Per-Åke Johansson

Dr. Jon Y. Hardeberg

Supervisor Professor Nils Enlund

Chairman Professor Roger Wallis

ISBN 91-7283-275-4
TRITA-NA-0207
ISSN 0348-2952
ISRN KTH/NA/R--02/07-SE
© Marianne Klaman, April 2002
Abstract
This thesis work deals with analysis and development of methods to
characterize colour rendering, to predict colour and to control colour in
printed media. There is often a need to make an accurate assessment of the
available colour gamut or the difference between a test print and a
reference, whether it be an original, a proof or a given set of inks on
different paper grades. It may also imply the comparison of different
printing techniques. The process treated is the production of printed media,
including the conventional printing represented by offset, digital proofing
and different digital printing technologies.

The work includes development of reliable methods to ensure that the

degree of fidelity in the colour reproduction chain in graphic arts media can
be satisfactorily assessed. The work is furthermore focused on development
of methods for the characterization of crucial parameters that influence
colour in the printing stage. This in turn implies the study and development
of methods to assess the available colour gamut, methods for colour
comparison, the analysis and understanding of how variations of techniques,
materials, and process parameters in each process step influence colour.
Since customers always judge the printed product at the end visually, it is
crucial that technical measures developed are in accordance with the
perceptional evaluations. Finally, the work deals with the relationship
between measured quality factors and the mechanisms causing the
variations in colour rendering.

For the analysis of colour in a way that is relevant for the printing process a
comprehensive test model for proofing based on densitometric and
colorimetric measures is proposed. The relationship between technical
measures and perceptional assessment also have to include several
parameters in order to, in a more complete way, bear a resemblance to the
comprehensiveness of the human vision. A test model based on crosswise
use of ICC-profiles (an ICC profile in this case characterizes the colour
performance of the printer or press) is developed for the evaluation of
tolerances concerning the substrate.

It is shown that even small deviations in inking level will cause

considerable colour shifts. It is demonstrated that substrate parameters as
paper surface roughness, paper whiteness and absorption can be used to
characterize the influence of the substrate on colour rendering.

Keywords: Colour quality, printed media, colour rendering, colour

differences, colour control, colour gamut, proofing, substrate parameters,
visual assessment.
Errata (so far found!)
Page 5, 1. 3. Lines 8-9, “An explanatory vocabulary...is added after the
reference”. There is no vocabulary. This was planned at an early stage but later
on the intention was changed since explanations were successively included in
the text. Unfortunately, the reference here was not removed.
Page 13 The second paragraph, “very similar”, should be: “like”.
Page 15 First line, “specifications” should be “specification”.
Page 19 The first paragraph, “match another” should be: “match one another”.
Page 24 Eq. 15, the last words preceding the formula, “and is expressed as:”,
should instead be: “and the Y-N modification of the M-D. eq. removes the
light-scattering effect and thus gives the physical dot area, as expressed here
in Eq. 15”.
Page 33 (McDowell, 1998) should be: (McDowell, 1998, p. 220).
Page 39 The (Dolezalek, 1990) reference is missing in the Reference List but
will be found in Paper III. The year, however, should be: 1994!
Page 42 DE94 should be: CIE94.
Page 44 (Baudet and Rousset, 2001) should be (Baudin and Rousset, 2001).
Page 66 Eq. 17 (here, and also in Paper VI, page 5) should be:
L 0 ..............
V = 1/6 ...................
.............
Page 70 The third paragraph from the end, “The colour difference measure”...
should be: “The colour difference measured in skin tone”.
Page 83 10.1.1.7, ...a*, should be: b*.
Page 90 Collet ...21st, should be: 26th.

Paper 1, page 8. The indices 1in the first line, and 4 in the last, in the formula
for the tetrahedron, have disappeared. Further: 10th line, the reference (Bristow,
1996) should be: (Bristow, 1995). 14th line, “each octagon” should be: “each
hexahedra”. Considering terminology, the term dodecahedron used is probably
not adequate, since, although referring to a polyhedra with twelve faces, the
expression dodecahedron generally seems to be used for polyhedra with twelve
pentagonal faces, and not as here with triangular faces. The correct term seems
to be hexagonal dipyramid.
Paper III, page 2. Under Background, last line in the first paragraph, “solid
tones of ± 2.5%”, refers to density.
Paper IV, page 4. Table 1, under notation for profile, second line,
Art1a 2 should be: Art 2 a 2 .
Paper IV, page 5. Delete visual; perception will be enough.
Paper VI, Figure 1. Delete one of the two letters C.
Contents

1 INTRODUCTION 3
1.1 THE IMPORTANCE OF COLOUR 3
1.2 BACKGROUND 4
1.3 THE THESIS OUTLINE 5

2 THE OBJECTIVES OF THE STUDY 7

3 THE METHODS USED 9

4 COLOUR SCIENCE 11
4.1 THE EYE 13
4.2 COLORIMETRY 14
4.2.1 CIE Standard Colorimetric Observer and colour matching
functions 15
4.2.2 CIE Colour Spaces 16
4.2.3 Colour difference formulas 17
4.2.4 Fluorescence 18
4.2.5 Some important phenomena 19

5 COLOUR REPRODUCTION IN MEDIA AND GRAPHIC ARTS

SYSTEMS 21
5.1 ADDITIVE AND SUBTRACTIVE PRINCIPLES OF COLOUR REPRODUCTION
21
5.2 HALFTONING 23
5.3 DOT GAIN 24
5.4 PROOFING 25

6 EVALUATION OF COLOUR 27
6.1 TECHNICAL MEASURES 27
6.2 PERCEPTIONAL ASSESSMENT 29

7 COLOUR CONTROL 31
7.1 LIMITATIONS OF COLOUR GAMUT 31
7.2 COLOUR MANAGEMENT AND ICC 32
7.2.1 Gamut mapping 34
7.3 COLOUR APPEARANCE MODELS 35

8 CONTEMPORARY RESEARCH IN RELATED AREAS 37

8.1 COLOUR MANAGEMENT APPLICATIONS 37
8.2 PROOFING 39
8.3 COLOUR GAMUT 40
8.4 COLOUR AND COLOUR DIFFERENCE MEASURES 42
8.5 FACTORS INFLUENCING COLOUR 44

9 SUMMARY OF ORIGINAL WORK 47

9.1 THE AUTHOR’S CONTRIBUTION TO PAPERS 47
9.2 PAPER I – A TEST MODEL FOR PROOFING SYSTEMS 48
9.3 PAPER II – IMPROVED PRODUCTIVITY AND IMAGE QUALITY BY USING
COLOUR MANAGEMENT SYSTEMS 54
9.4 PAPER III – COLOUR SHIFTS IN FOUR-COLOUR PRINTING 56
9.5 PAPER IV – THE INFLUENCE OF PAPER ON ICC-PROFILES 60
9.6 PAPER V- THE INFLUENCE OF PAPER WHITENESS AND OTHER
PARAMETERS ON THE CREATING OF ICC-PROFILES FOR DIGITAL
COLOUR PRINTERS AND CONVENTIONAL OFFSET PRESSES 63
9.7 PAPER VI – COLOUR RENDERING ASPECTS IN DIGITAL PRINTING 65
9.8 APPENDIX - THE COMPARISON OF COLOUR GAMUTS 72

10 DISCUSSION AND CONCLUSIONS 75

11 ACKNOWLEDGEMENTS 87

12 REFERENCES 89
12.1 LINKS 95

2 (95)

Marianne Klaman Royal Institute of Technology 2002

1 Introduction
Colour reproduction for printed media includes the reproduction of an
original in such a way that colour fidelity of print with original is achieved.
The process includes the conversion of original colour values to values for
proof, hard or soft, and to values for the final print.

1.1 The importance of colour

Colour fidelity and colour matching have increased in importance in media
since use of colour has increased tremendously, not only in printed media
but also in other media due to the large addition of multichannel publishing;
print, TV, film, Internet and others. Multichannel and cross-media
publishing with distribution of information, advertising and marketing also
sets increasing demands on the colour fidelity and true colour rendering of
an original. Colour management is one of the areas pointed out as an
important research field in what Lindqvist calls “the restructured media
field” (Lindqvist, 2000). Imagine purchasing a sofa over the Internet or
from a printed catalogue. You have decided to buy a red sofa, a special red
sofa and naturally you do not want to have the sofa, red to be sure, but more
yellowish red or more bluish red than that red you ordered.

Colour plays a varied and important role throughout human life. We use
colours to decorate our surroundings and also ourselves in a pleasing way.
Sometimes even the choice of colours may “imply the difference between
life and death. The basic heraldic tinctures (heraldry has been given its own
language) are comprised of gold (mostly yellow in modern practice), silver
(or white), and the colours: red, blue, black and green. Tinctures must not,
however, be combined in just any fashion. Metals may not touch each other,
and colour is not allowed to meet colour. Arms composed in compliance
with these criteria will be clear and easy to identify. Why is this important?
The mediaeval knight was concealed inside his suit of armour. On the
battlefield, it was consequently vital for him that the coat-of-arms on his
shield be easily distinguishable at a long and safe distance. Otherwise he
could mistakenly be killed by his companions.
And take a look around in the streets. The colour scheme of our traffic signs
mainly follows the mediaeval heraldic tincture rules. Traffic signs, too,
must be easily distinguishable at a long and safe distance. This is not to say
that roads and streets are battlefields (but now and then they are not far from
it)” (Jarnvall, 2002).

3 (95)

Marianne Klaman Royal Institute of Technology 2002

Colour also helps with the recognition of different objects. Certain objects
are very closely linked to their colour and help us immediately recognize
them although we are in fact not entirely aware of the shape and other
features.

1.2 Background
Looking at colour reproduction it is of importance to develop reliable
methods to ensure that the degree of fidelity of reproduction from one stage
to another in the media process can be satisfactorily assessed. The need to
make an accurate assessment of the available colour gamut or the difference
between a test print and a reference original arises in different situations,
not only when final prints are compared with each other and with an
original, but also in the comparison of the potential of different sets of inks
on a given paper, of a given set of inks on different kinds of paper, the
comparison of different colour copiers and printers and the comparison of
different proofing systems.

In media technology and graphic arts reproduction there is a need to be able

first to characterize colour rendering, then to be able to predict the rendering
in different situations and lastly, manage to control how colour will be
rendered in different reproduction and printing processes, with different out-
put media and on different substrates.

The area of colour control and colour reproduction includes fields of re-
search such as colour appearance modelling, gamut mapping, development
of algorithms in the field of colour management and other related areas.
There is also a need to improve workflows within colour reproduction using
the different tools developed for colour management.

Accordingly, the need of relevant evaluation and measuring methods are ap-
parent. Among other things, the factors when talking about print quality
includes sharpness, contrast, detail rendering and colour fidelity. When high
quality colour printing is addressed, the colour quality expressed as colour
fidelity and colour matching is crucial.

To be able to control the reproduction of colour on printed media it is

essential to have the knowledge of how colour will be rendered in different
processes and on different substrates and how each of the parameters
involved in the reproduction process will influence the colour rendering.
Each step in the process includes several essential parameters that will
certainly influence the colour rendering to a lesser or greater degree. It is
4 (95)

Marianne Klaman Royal Institute of Technology 2002

therefore vital to control every single step and also to possess the
knowledge of how the different materials involved, the different parameters
and the different techniques influence the colour.

1.3 The thesis outline

The thesis starts with a description of the objectives and the methods used.
It is then followed by a general survey of colour science and colour
reproduction including colour evaluation, proofing and colour control. Also
included in this more general section is some contemporary research in
these fields, such as gamut mapping and colour appearance modelling. After
that follows an overview of studies within areas more closely related to the
author’s investigations. This is concluded with the author’s exposition and
discussion of her own research. An explanatory vocabulary of some
technical words and expressions is added after the references. Included at
the end are the six papers and one appendix.

The mentioned papers are:

Paper I - A Test Model for Proofing Systems (1995).

Paper II - Improved Productivity and Image Quality by using Colour
Management Systems (1997).
Paper III - Colour Shifts in Four Colour Printing (1999).
Paper IV - The influence of Paper on ICC-profiles (1999).
Paper V - The Influence of Paper Whiteness and other parameters on
the creating of ICC-profiles for digital colour printers and conventional
offset presses (1999).
Paper VI - Colour Rendering Aspects in Digital Printing (2001).

Also included is an Appendix. Many of its assumptions created a basic

platform for further research work. Thoughts formed in discussions with the
main author, J. A. Bristow, as well as the results gained from empirical
studies conducted in cooperation, influence the present thesis. It is therefore
essential also to include this paper but in the form of an Appendix, since J.
A. Bristow did most of the writing.

The Appendix is named:

The Comparison of Colour Gamuts (1993).

5 (95)

Marianne Klaman Royal Institute of Technology 2002

 6 (95)

Marianne Klaman Royal Institute of Technology 2002

2 The objectives of the study
The work in this study deals with the analysis and development of methods
to characterize colour rendering, to predict colour and to control colour in
printed media. All these factors are crucial for achieving, in the end, a high
print quality with the main focus on colour quality. The need to make an
accurate assessment of the available colour gamut or of the difference
between a test print and a reference original arises, not only when final
prints are compared with each other and with an original. It also arises in
the comparison of the potential of different sets of inks on a given paper, of
a given set of inks on different papers, and further, in the comparison of
different colour copiers and printers and of different proofing systems.

The two main objectives of the thesis work are:

To develop reliable methods to ensure that the degree of fidelity of
colour reproduction from one stage to another in the graphic arts media
process can be satisfactorily assessed.
To develop methods for the characterization of crucial parameters that
influence colour in the printing stage.

This means that the objectives include approaches such as:

The development of reliable methods that can be used for the
comparison of colour deviations.
Developing accurate assessment methods for the available colour
gamut.
The analysis and understanding of how variations in the steps in the
process influence the colour, according to techniques, materials and
process parameters.
Investigation and validation of the technical measures used, and to what
extent they are in accordance with the perceptional evaluation.
Evaluation of the relationship, if possible, between the measured
quality factors and the mechanisms causing the variations, as crucial
parameters in the out-put printing section are assumed to have a strong
impact on the colour rendering.

The process concerned is the production of printed media, including the

conventional printing represented by offset, digital proofing and various
digital printing technologies.

7 (95)

Marianne Klaman Royal Institute of Technology 2002

 8 (95)

Marianne Klaman Royal Institute of Technology 2002

3 The methods used
Disturbing factors in the colour reproduction process will cause colour
deviations experienced by human vision as unintended and undesired colour
shifts. The control of colour in a reproduction process includes control of
the different process steps and use of tools for prediction of the print. To
predict the final print, proofing is used as a check of the reproduction work
performed and in this aspect is essential as a simulation tool. The reliability
of the proofing systems then becomes crucial. Since the resulting print
quality concerning colour fidelity is ultimately based on the acceptance of
human vision, the validation of colour measures against perception
assessment is an important part of the work.

The following methods are used:

Development of test models. One is the development and validation of

a test model, primarily aimed for testing proofing systems, but with the
intention to be possible to use also for other colour comparisons (Paper
I). Another is the development of a test method for ICC-profiles and
their accuracy and tolerance for different substrates (Paper IV).
Theoretical analysis, which focuses on the use of colour management
and on benefits and disadvantages possible to be assumed in a normal
production of printed media. This is discussed in terms of quality and
productivity (Paper II).
Empirical studies. The sensitivity of the eye to colour shifts caused by
density variations due to variations in the inking level is analysed The
correlation between visual assessment and technical measuring is
investigated (Paper III). Empirical studies of the influence of different
parameters in press, mainly paper properties, on the colour rendering
expressed in terms of colour gamut size, colour differences and
influence on ICC-profiles are conducted. An analysis is performed of
the possibility to work with ICC-profiles in a way that enables a
productive handling of the colour workflow in printing companies.
Comparison is made of the performance of digital printing versus
conventional printing (Papers IV-VI).

9 (95)

Marianne Klaman Royal Institute of Technology 2002

 10 (95)

Marianne Klaman Royal Institute of Technology 2002

4 Colour science
The study of colour is a pronounced interdisciplinary activity involving
many arts and sciences. Further the colour reproduction in media is part of
the colour science theories. In this chapter some of the fundamentals in
colour science are explained. Also, some contemporary research in this field
is referred to. The general survey forms a background to colour reproduc-
tion and as such in a high degree a background to the present thesis work.

The fundamentals of colour are, among others, expressed in the work of

(Wyszecki & Stiles, 1982).

To experience colour in objects, light is a basic requirement in viewing.

Human vision of the objects illuminated, together with the interpretation of
the human brain, gives us the resulting colour experience.

These basic facts are stated in many works in the colour science field, such
as (Hunt, 1987, p 17) where he declares that colour matching involves “the
three basic components of colour: sources of light, objects illuminated by
them, and observers”, and in the colour reproduction area (Field, 1988, p.
23) expressed that “to understand the sensation of color, it is necessary to
examine the illuminant, the characteristics of the sample, and the human
factors, physiological and psychological”. It is not enough to examine
colour match in only one type of light and often more than one opinion is
needed to verify the match. (Field, 1988, p23) further expresses colour as
“essentially a complex visual sensation, influenced by physiological and
psychological factors that probably make one person’s perception of color
slightly different from another’s”. This complexity of not being a physical
phenomenon alone but also possessing physiological and psychological
elements makes the investigation of colour complex and makes it necessary
to validate the measures or methods used, by means of visual assessment,
which is the case in several of the papers of this thesis.

Isaac Newton, concluded that, based on experimental facts, white light was
not a simple homogenous entity but was composed of a mixture of all the
colours in the spectrum, see Figure 1. Since then the understanding of the
mechanisms of colour and colour vision has progressed. The development
of theoretical models for colour vision and for the mathematical
representation of colour has increased dramatically, as well as the
development of colour measuring methods.

11 (95)

Marianne Klaman Royal Institute of Technology 2002

Figure 1. To detect the spectrum of white light, a prism that causes the refraction of
the light when it goes from one medium to another is needed (in the case of the
prism, from glass to air). It is the same phenomenon as when light passes through a
raindrop and the spectrum is made visible.

A well-known phenomenon is the influence of surrounding colours. When

we look at a single colour spot, we get one impression of the colour if the
colour spot is isolated as opposed to if other colours surround it.

This is illustrated in figure 2. Two different colours surround the colour spot
in the middle and though the colour spot is identical in both cases, it appears
to be different. There exist a great deal of phenomena such as this that
influence the appearance of colour, like simultaneous contrast (analogous to
different colour frames but with different grey frames instead) and
crispening (defined as the increase in perceived magnitude of colour
differences when the background on which the two stimuli are compared is
similar in colour to the stimuli themselves), all described by (Fairchild,
1998). These phenomena illustrate in a simple way why it is not enough to
measure colour and try to match the measures gained. Also, the
physiological and psychological effects have to be taken into consideration
in colour matching.

12 (95)

Marianne Klaman Royal Institute of Technology 2002

Figure 2. The well-known phenomenon of how the colour in the middle is perceived
depending on the surrounding colour.

4.1 The eye

Literature on human vision and the function of the eye and colour vision
can be studied in many works on colour or colour reproduction, such as
(Wyscecki & Styles, 1982; Hunt, 1987; Fairchild, 1998; Field, 1988) to
name a few. There is also (Guyton, 1986), classical literature for many
medicine students.

The fundamentals of colour vision are based on the fact that “the human eye
acts very similar to a camera. The cornea and lens act together like a camera
lens to focus an image of the visual world on the retina, located at the back
of the eye, which acts like the film or other image sensor of a camera. These
and other structures have a significant impact on our perception of color” as
described by (Fairchild, 1998, p. 3).

The fovea is the area of the retina where we have the best spatial and colour
vision, see Figure 3. Rods and cones are the retinal photoreceptors. There
are three types of cones and they serve colour vision. (Fairchild, 1998)
claims that the most appropriate names are L, M and S cones. The names
refer to the long-wavelength, middle-wavelength and short-wavelength
sensitive cones, respectively. Contrary to the cones there is only one type of
rods, which makes the rod system incapable of colour vision. The visible
spectrum ranges from 380 to 780 nm.

13 (95)

Marianne Klaman Royal Institute of Technology 2002

Figure 3. Schematic diagram of the human eye. Source: Fairchild, 1998 p. 4.

Since the mechanisms of colour vision are very complex, there have
historically been many theories that attempt to explain the function of
colour vision (Fairchild, 1998). Fairchild mentions in his book some of the
more modern of these, such as the “Thrichromatic Theory”, “Hering’s
Opponent-Colors Theory” and the “Modern Opponent – Colors Theory”.

The development of colour appearance models has become important not

least within the media industry. In our fast developing information society,
now that publishing has become an issue of multichannel and cross-media
publishing the importance of relevant colour reproduction has also become
even more essential. Colour Appearance modelling is also a key factor in
the development of modern colour management systems.

4.2 Colorimetry
To be able to communicate colour and what colour looks like, the special
field of colour science, namely colorimetry, was developed. The need to
specify colour in numerical terms forced the development of a definition of
the physically defined stimulus in such a way that (Wyscecki & Styles
,1982, p. 117) claimed:

“(a) when viewed by an observer with normal color vision, under the same
observing conditions, stimuli with the same specifications look alike (i.e.
are in complete color-match).
(b) stimuli that look alike have the same specification, and
14 (95)

Marianne Klaman Royal Institute of Technology 2002

(c) the numbers comprising the specifications are continuous functions of
the physical parameters defining the spectral radiant power distribution of
the stimulus.”

4.2.1 CIE Standard Colorimetric Observer and colour

matching functions
The base of applied colorimetry is the CIE 1931 Standard Colorimetric
Observer (CIE = Commission Internationale de l’Eclairage).

“The colour matching properties of the CIE 1931 standard observer are
defined by the colour matching functions x (λ ) , y (λ ) and z (λ ) . They
represent the amounts of R, G and B needed to match a constant amount of
power per small constant-width wavelength interval throughout the
spectrum” as expressed by (Hunt, 1987, p. 46). This Standard Colorimetric
Observer is assumed to represent the colour-matching characteristics of
those among the human population possessing normal colour vision. A
viewing angle of 2° was defined for the judging experiments from which
the colour-matching functions were derived. The Supplementary Standard
Colorimetric Observer defined by CIE in 1964 and referred to as the 10°
Observer, complemented this 2° Observer.

The “color-matching functions are used in the calculation of CIE tristimulus

values X, Y and Z, which quantify the trichromatic characteristics of colour
stimuli. The X, Y and Z tristimulus values for a given object (characterized
by its spectral reflectance or transmittance) that is illuminated by a light
source (characterized by its spectral power distribution) can be calculated
for the CIE Standard Colorimetric Observer (characterized by the CIE
color-matching functions) by summing the products of these distributions
over” the visible wavelength (λ ) range as (Giorgianni, Madden, 1998, p.
20-21) expresses it in their survey. (Hunt, 1987, p. 48 declares): “The CIE
colour-matching functions”…”are the most important spectral functions in
colorimetry. They are used to obtain X, Y and Z tristimulus values from
spectral power data. If two colour stimuli have the same tristimulus values,
they will look alike, when viewed under the same photopic conditions, by
an observer whose colour vision is not significantly different from that of
the CIE 1931 Standard Colorimetric Observer; conversely, if the tristimulus
values are different, the colours may be expected to look different in these
circumstances”.
The chromaticity coordinates x, y and z [Eq. 1-3] (defined as a type of
relative tristimulus values) are derived from the tristimulus values as:

15 (95)

Marianne Klaman Royal Institute of Technology 2002

 X
x= [Eq. 1]
X +Y + Z

Y
y= [Eq. 2]
X +Y + Z

Z
z= [Eq. 3]
X +Y + Z

The Y tristimulus value corresponds to the measurement of luminance. In

colour-imaging applications, as in media, the measurement of luminance is
of particular importance because luminance is an approximate correlate of
one of the principal visual perceptions namely brightness (Giorgianni,
Madden, 1998).

(Fairchild, 1998, p. 90) points out that “chromaticity coordinates should be

used with great care, since they attempt to represent a three-dimensional
phenomenon with just two variables”. This is important and not to be
forgotten. The three variables characterizing colour are mostly expressed in
terms of lightness, hue and chroma (saturation).

4.2.2 CIE Colour Spaces

The development of the CIE colour spaces, CIELAB and CIELUV made
the CIEXYZ “largely obsolete” (Fairchild, 1998, p. 91). In these two colour
spaces the tristimulus colorimetry is extended “to three-dimensional spaces
with dimensions that approximately correlate with the perceived lightness,
chroma and hue of a stimulus” and the perceived differences of colours are
represented more uniformly. The CIE 1976 (L* a* b*) colour space is
defined by Eq. 4-8.

L* = 116(Y / Yn )
1/ 3
− 16 [Eq. 4]

[
a* = 500 ( X / X n )
1/ 3
− (Y /Y n )
1/ 3
] [Eq. 5]

b* = 200[(Y / Y ) − (Z / Z n ) ]
1/ 3 1/ 3
n [Eq. 6]
16 (95)

Marianne Klaman Royal Institute of Technology 2002

 *
C ab = (a * 2
+b * 2 ) [Eq. 7]

hab = tan −1 (b * / a *) [Eq. 8]

X, Y and Z are the tristimulus values of the stimulus and X N , YN and

Z N are the tristimulus values of an appropriately chosen reference white.
L* represent lightness, a* approximate redness-greenness, b* approximate
*
yellowness-blueness, C ab chroma and hab hue.

4.2.3 Colour difference formulas

The CIELAB space is frequently used in the graphic arts field of media. Of
*
specific importance is the specification of the ∆E ab -value [Eq. 9] which is
the colour difference measured in the CIELAB space as the Euclidian
distance between the coordinates for two stimuli. The goal of the CIELAB
colour space design was that the colour differences should be perceptually
uniform in all areas of the colour space. However, this goal was not fully
achieved and that is also the reason for continued research in the field of
establishing relevant colour representation systems.

∗
∆E ab [(
= ∆L∗ ) + (∆a ) + (∆b ) ]
2 ∗ 2 ∗ 2
1/ 2
[Eq. 9]

Improvements of the colour difference formula were performed by (Clarke,

McDonald and Rigg, 1984) and addressed in the work of (McLaren, 1986),
i.e. the CMC (l:c) formula. This formula was developed to evaluate small
colour differences in the colorant industry. The formula itself varies the
relative weights of differences in lightness, chroma and hue depending on
the position of the colour in the gamut. The possibility of weighing the
importance of lightness and chroma in the formula is assumed to give it a
larger potential for measuring small differences. The CMC formula is used
in one of the papers (III) in order to investigate its usability for media
applications.

17 (95)

Marianne Klaman Royal Institute of Technology 2002

The CIE (CIE, 1995) recommended a new colour difference formula for
industrial use, called the CIE 1994 ∆L∗ ∆C ab
*
( *
∆H ab )
colour difference
*
model with the symbol ∆E 94 and abbreviated as CIE94.

CIE94 colour differences are calculated as seen in Eq. 10-13:

1/ 2
 ∆L *  2  ∆C * 2
  ∆H ab*

2

∆E *
94 =   +  ab
 +    [Eq. 10]
 k L S L   k C S C   kH SH  

SL = 1 [Eq. 11]

*
S C = 1 + 0.045C ab [Eq. 12]

*
S H = 1 + 0.015C ab [Eq. 13]

The factors k L , k C and k H are used to adjust the relative weighting of the
lightness, chroma and hue components, respectively, for various viewing
conditions and applications that depart from CIE94 reference conditions.
The CIE (CIE, 1995) established a set of reference conditions for the use of
the CIE94 colour-difference equations concerning illumination, viewing
mode, sample size and sample colour-difference magnitude, among others.

4.2.4 Fluorescence
A topic of importance when performing colorimetric measuring of materials
is fluorescence. Fluorescent materials are characterized in that they absorb
energy in one region of wavelengths and emit the energy in another region
with longer wavelengths (Fairchild, 1998). Fluorescent materials are
characterized by their total radiance factor. This factor is dependant on the
light source of the measuring device, which is not the case with non-
fluorescent materials. They are insensitive to the light source. Thus,
measuring fluorescent materials accurately can be difficult. Since some inks
and paper are fluorescent, it is therefore of importance to take this fact into
consideration when measuring.
Fluorescence in paper can for instance prove troublesome when trying to
achieve exact colour matches of light pastel colours (Field, 1988).

18 (95)

Marianne Klaman Royal Institute of Technology 2002

4.2.5 Some important phenomena
There are some phenomena to be explained. One is metamerism. Many of
us have certainly faced the problem, for instance, of matching a jacket with
trousers, which can look almost identical in colour when trying them in the
fitting-room but when light source is changed (outdoors!) they look
remarkably different. This is explained by (Hunt, 1987, p. 141): “when two
samples match another, but are different in spectral composition, they are
said to be metameric, and the phenomenon is referred to as metamerism.
The phenomenon of metamerism is of great importance, because the greater
the degree of metamerism, that is, the greater the degree of difference in
spectral composition, the greater will be the likelihood that, if the spectral
composition of the illumination is changed, or if the observer is changed,
the colours will no longer match one another”. Metameric colour stimuli
have identical tristimulus values but different spectral radiant power
distributions (Wyszecki & Styles, 1982). Metamerism is one of the factors
that implies the need in media industry to use standardized illumination in
critical assessment of prints, proofs and other subjects. In perception
assessments in scientific matters the viewing conditions are crucial.

Besides other properties, colours help us to recognize objects. Objects are

illuminated under a very wide range of conditions where both the power
level and the colour of the illumination vary. Colour constancy is the
phenomenon defined by the human visual system and its ability for
adaptation. This means that the system has the capacity to compensate for
changes in both the level and the colour of the illumination and still retain
the same impression of a certain colour. This adaptation capacity of the
human visual system is one reason to develop measures instead of merely
rely on visual judging of colours in production control of print quality. The
eye, although it is an excellent instrument in comparisons, is also due to this
adaptation and thus insensitive to gradually changing colour variations.

19 (95)

Marianne Klaman Royal Institute of Technology 2002

 20 (95)

Marianne Klaman Royal Institute of Technology 2002

5 Colour reproduction in media and graphic
arts systems
Colour reproduction in graphic arts is in many ways based on the same
principles as colour photography but with its own specific characteristics.
(Yule, 1967) described the main principles of colour reproduction and the
basic rules. Some of them are still valid, although of course the
development of new techniques and digitalisation has changed the processes
and the work involved.

“ Color reproduction can be defined as an optical process of producing a

close color representation of some original scene or object. Photography,
electronics, and the physical transfer of a colorant to a substrate may each
play a role in this process, depending on the form of the reproduction. In a
broad sense, the process includes making photographic transparencies and
prints, television images, and printed reproductions”, as declared by (Field,
1988, p.1). But naturally, there also exist other ways to define the process.

Among more recent literature in the colour reproduction field (Schildgen,

1998) can be mentioned as a handbook, as well as the documentation
gathered (Lindsay, McDonald and Luo, editors, 1999). The latter one deals
with colour reproduction in a broad sense, including multimedia colour
reproduction.

5.1 Additive and subtractive principles of colour

reproduction
In printed media the basics of the additive and the subtractive colour mixing
principles are essential. Wavelengths of light from three sources in equal
proportions and projected on a white screen will give the perceived colours
of red, blue and green, as seen in Figure 4. These are the primaries of the
additive colour reproduction principle. Where two colours are overlapping,
they form cyan, magenta and yellow, respectively. We recognize these as
the primaries of the subtractive colour principle and the colorants used in
traditional printing. When all three are combined we get white. In monitors,
we have the so-called additive principle: and red, green and blue spots are
created from the phosphors used.

21 (95)

Marianne Klaman Royal Institute of Technology 2002

Figure 4. The principle for additive colour mixing.

The starting position for the subtractive process is white (in printing the
substrate, mostly white paper and illuminated by white light) and the
subtraction of red, green and blue to achieve black, Figure 5. To be able to
subtract these, colorants that are the opposites of red, green and blue must
be used i.e. cyan, magenta and yellow, CMY. With the use of these and the
white of paper, it is possible to reproduce eight colours, i.e. the overprints of
the primaries. Adding a black colorant K will produce a better black.

Figure 5. The principle for subtractive colour mixing.

22 (95)

Marianne Klaman Royal Institute of Technology 2002

5.2 Halftoning
To be able to render more colours, the fraction of surface coverage by each
of the four colorants must be varied on a microscopic scale. By
superimposing four images, one for each of the four colorants, and each
split into small dots, of a size under the resolution level of the eye, a whole
range of colours can be reproduced. This is called halftoning. The process
of converting a continuous tone image into a dot image is called screening.

There are different modes of screening techniques where the conventional

screening, also called amplitude modulated screening (AM screening), is
periodic in the sense that the distance between the dots (the centre of the
dot) is equal and the dot size is varied to obtain the different steps of the
tonal scale. With frequency modulated screening (FM screening) the
distance between the dots is instead varied and the dots are equal in size.
Finally, the conventional and the frequency modulated screening can be
combined in different ways in order to obtain the advantages of each
technique. For instance, the AM screening can be used in parts of an image
where the halftones have to be smooth and the FM screening can then be
used in areas of fine detail.

Regarding the AM screening, the geometry of the dot is one of the factors
that will certainly influence print quality and specifically colour quality.
The initial round or square dot is mostly exchanged to the elliptical and
chain dot as well as improvements of these and they have become the most
frequently used dot shapes. Dependent on how the dots interact in the
middle tones, that is the critical area due to linking of the dots, the colour
rendering will probably vary.

In digital printing the dot shape is mostly specific and linked to the special
technique used. Dots from an ink-jet printer are dissimilar to dots from an
electro-photographic printer or press and thus assumed to have different
effect on colour rendering.

A number of equations have been proposed to represent the relationship

between reproduction density and amount, or dot area, of ink. Best known
are the Neugebauer equations (Neugebauer, 1937). They are based on the
microstructure of the printed halftone pattern. The equations predict the
reflectance of red, blue and green from given dot areas in print.

The amount of CMYK, respectively, needed for every single colour spot in
an image can be calculated. Depending on whether any kind of UCR or
GCR is used, the combination of CMYK will vary according to the degree
23 (95)

Marianne Klaman Royal Institute of Technology 2002

of UCR or GCR used. UCR is defined as Under Colour Removal and
implies that the cyan, magenta and yellow amounts that form the neutral
part of the image can be replaced by black ink instead. Analogous to this,
the parts of cyan, magenta and yellow that form the neutral and the
complimentary parts in an image can be substituted by black, GCR (Grey
Component Replacement). To exchange the coloured inks with black will
have a positive impact on the grey balance, as well as on the drying of the
ink-layer, since less ink is delivered to substrate. Added to that are the
economical benefits, since the black ink is less expensive than the coloured
ones.

5.3 Dot gain

Since the screened image is composed of small dots, the dots will increase
(in some process steps also decrease) in size from the original values,
through the different steps in the reproduction process and result in what is
called dot gain. Depending on factors such as plate, substrate, ink and press,
the dot gain will vary. In the printing process dot gain together with ink
density is a parameter to be controlled and held consistent. There are two
kinds of dot gain.

Mechanical (or sometimes called physical or geometrical)

Optical

The Murray-Davies equation (Murray, 1936) includes both types of dot gain
and is expressed as:

DMD = - log [1-a(1-10-Ds)] [Eq. 14]

where:

DMD is the optical density of the halftone print, a is the dot area and Ds is the
optical density of the solid print

The optical dot gain is also called the Yule-Nielsen effect (Yule, Nielsen,
1951) and is expressed as:

DYN = -n log [1-a(1-10-Ds/n)] [Eq. 15]

where DYN is the predicted optical density of the print, a is the dot area and
n is a factor regarding the amount of light diffusion into the paper and
24 (95)

Marianne Klaman Royal Institute of Technology 2002

dependent on the type of paper and the screen ruling and finally, Ds is the
optical density of the solid print.

More recent research in this field has been performed by (Gustavsson,

1997), and others.

5.4 Proofing
An important tool used in the chain of reproduction processes is proofing.
Proofing is the simulation of a print made before the production run for
control of the prepress work and also used for customer approval. There are
analogue and digital proofing devices, where proofs are made either from
the reproduction film sets or from the digital file. Today most proofing
systems are digital.

The technique is based on the same principles as many printers and copiers:
electrophotography or ink-jet. Hard proof then refers to a copy from any of
these devices. Soft proofing is the use of the monitor as a proofing device
(must then be a high quality monitor, calibrated and characterised, ie
calibrated to standard setting and then characterised according to colour
performance).

The definition, and at the same time the aim of proofing, can be stated as:
The control of the prepress or premedia work with the inclusion of
layout, text quality, image quality including the colour matching of
proof with original.
The simulation of the optical appearance or impression of the print.

However, the proof shall not be used as an original to which the print is
targeted. Printing must be performed with established standard values for
the solid tone density and dot gain.

A comprehensive document on the basics of colour proofing is the work of

(Bruno, 1986).

25 (95)

Marianne Klaman Royal Institute of Technology 2002

 26 (95)

Marianne Klaman Royal Institute of Technology 2002

6 Evaluation of colour
Colour can be evaluated by two means:
Technical measures
Perceptional assessment

Both methods are used in the work involved in this thesis, whereas one
approach is to find the relationship between technical measures and
perceptional judging.

6.1 Technical measures

Measuring colour in the media industry becomes more and more essential
whereas the demand on high colour quality increases and since colour is an
inter-disciplinary science and “colour, therefore, involves not only material
sciences, such as physics and chemistry, but also biological sciences, such
as physiology and psychology; and, in its applications, colour involves
various applied sciences, such as architecture, dyeing, paint technology, and
illuminating engineering. Measuring colour is, therefore, a subject that has
to be broadly based and widely applied” (Hunt, 1987, p.17).

According to colorimetry, the basics of colour measuring include certain

definitions explained shortly in this chapter.

The CIE X, Y, Z tristimulus values can be calculated from spectral data.

The crucial factors when obtaining spectral data is to use standard
illuminants and light sources, define the colour temperatures, define
standard reference whites and define the geometries and viewing angles of
measuring devices to be used.

Illuminants are, according to CIE, defined in terms of relative spectral

distribution while sources are defined as the physical producers of radiant
power. The distribution temperature of a source is defined as the absolute
temperature in Kelvin (K) of the blackbody radiator for which the spectral
radiant power distribution is proportional (or approximately so) to that of
the source considered. The term colour temperature is applied to highly
selective radiators, such as electric discharge lamps, when the light of this
radiator has the same (or nearly the same) chromaticity coordinates as a
blackbody radiator at a certain temperature (Wyszecki & Stiles, 1982). The
blackbody radiator, also called a Planckian radiator, is a thermal radiator
imagined to be capable of providing a spectral power distribution that
depends only on the temperature.

27 (95)

Marianne Klaman Royal Institute of Technology 2002

Standard illuminants D represent daylight and, for the media industry, D50 is
used: daylight having a correlated colour temperature of 5000 K. In the
paper industry, D65 is used, recommended in 1963 by CIE, to represent
better the daylight in the ultra-violet region, due to the use of optical
brightening agents in the paper production and since the older standard
illuminants had too little power in the UV region.

According to CIE, the reference white is a perfect reflecting (or

transmitting) diffuser, easy to use in colorimetric calculations but of course
not available in practical measurements. Using working standards that are
calibrated against a perfect diffuser in a national standardizing laboratory
has often solved this. These working standards have been prepared in
different ways, such as with barium sulphate pressed to a cake or ceramic
tiles (Hunt, 1987).

The spectrophotometers designed for printed media applications have 45/0

or 0/45 illumination and viewing angles. The paper industry often uses
diffuse illumination provided in instruments by integrating spheres.

Measures used in the work involved in this thesis are based on the
colorimetric standards and the colour space preferably used is CIELAB.
The specific methods used are described in each paper separately. Besides
spectrophotometric methods, methods based on densitometry, image
analysis and some methods in connection with microscopic analysis are
used.

(Schultz, 1987) discussed the main advantages and disadvantages of

densitometry versus colorimetry in order to evaluate the properties of
printing inks as well as the visual appearance of colours and their variation
in the many steps of the printing process. Also (Wilkinson, 1985) took up a
fundamental discussion of the use of colorimetric evaluation of the match
between pre-proofs and gravure prints.

Up until then, the visual assessment or the densitometric measurement were

the most common methods in use. But since that time, the use of
colorimetry and spectrophotometric measurements are more frequent in the
production of printed media. In research work concerning colour, it is
fundamental.

28 (95)

Marianne Klaman Royal Institute of Technology 2002

6.2 Perceptional assessment
Since the quality, both the print and colour quality, of a printed product is,
in the end, always judged by an observer - the consumer of the printed
product - the measured quality must always be validated against the
perceptually perceived quality. Concerning the quality of colour different
persons see colours slightly different, but certain generalizations can be
made, such as:

We are more sensitive to deviations in chroma and hue than to these in

lightness
The motif and kind of colours in an image are probably more or less
sensitive to colour variations

The visual assessments made in the studies of the work for this thesis are
mainly based on the pair comparison method defined by (Bristow and
Johansson, 1983) and on methods specifically designed for each test, and in
that case described in the specific paper. The visual tests are often
performed in comparison with some kind of a reference that serves as the
truth. Visual assessments based on proscale evalutaion and opinion ratings
(Eidenvall et al., 2001) and (Norberg et al., 2001) were also used in the
investigation according to Paper VI. The principle of the proscale
evaluation briefly is explained by: each one of the test panel may group the
test samples into categories according to some characteristics for each
category, and then also decide whether the group defined could be
acceptable or not referred to a consumer. The opinion rating is described as:
the test sample is compared with a reference and rated according to the
reference.

Perception assessments have to be conducted under standardized

circumstances such as standardized viewing conditions, D50 and neutral
surroundings. The surroundings are both the area immediately adjacent to
the test subjects (images or others) to be assessed and the sphere of the
room or viewing box around the place of judgement.

29 (95)

Marianne Klaman Royal Institute of Technology 2002

 30 (95)

Marianne Klaman Royal Institute of Technology 2002

7 Colour control
The control of colour involves the different process steps in colour
reproduction. It includes conversion of the colour values from RGB values,
probably in different steps (at least scanner and monitor), to CMYK values.
Since different digital cameras and scanners have different filters and CCD
cells, monitors have their different characteristics, and printers and presses
their specific characteristics and specific colour properties, there are many
crucial decisions and calculations to be performed in the chain of colour
control.

7.1 Limitations of colour gamut

When converting colours from the gamut of the original or of the proof to
the gamut of the print, there is often a limit both in size and in shape. The
available colour gamut is often smaller and the conversion of colour values
comprises some kind of decision as to what to do with those colour values
that fall outside of the available or reproducible gamut.

When reproducing an original image for printed media, there are certain
limitations in the press that will influence the available colour gamut. The
press, with its certain characteristics, will limit the possibilities in
reproduction; the substrate used will limit the gamut, as will the inks or
toners used. The variations in the printing process itself heavily influence
the print quality, not least the colour rendering.

If we take the substrate into consideration, we have the whiteness of the

paper as a limiting factor. It certainly limits the gamut since it is never
possible to gain a “whiter white” than the existing one, due to the white of
paper. The ink limits the gamut in the other end, since we cannot gain a
blacker black than the ink or the toner gives. Concerning all the colours in
an original, there are many colours that are not reproducible with the
available pigments of inks.

A literature review on colour gamuts in the printing process by (Rydefalk,

1997) summarizes many of the fundamentals in this area.

The expanding of the reproducible colour gamut is a matter of great interest

and research in this field has been performed, among others, by (Paul,
1994), who compared ideal inks with real inks and testing what he called
“best inks” to expand the gamut and calculate the amount of distinguishable

31 (95)

Marianne Klaman Royal Institute of Technology 2002

colours in the gamut. He showed that with these best inks, 1.379 million
distinguishable colours could be achieved as compared to 1.012 million
colours with standard inks in the CIELUV system. Ideal inks should give
4.5 million distinguishable colours.

7.2 Colour Management and ICC

With the digitalization of the workflow in media production, there arises a
demand for a more effective colour control. The development of different
colour management systems became the solution. A control, based on
mathematical equations, through the different conversion steps became
central. The development of such colour management systems very soon
created the need for a standard to be able to communicate colour profile
data. The first systems developed were all vendor specific. (Steiger, 1993)
reported over a dozen of such systems at that time. During the early 90’s
there seemed to be a lot of activities concerning device-independent colour
reproduction (Clippeleer, 1992; Johnson, 1992), among others.

The International Colour Consortium (ICC), established in 1993 by several

of the leading companies in the prepress colour industry, created, promoted
and encouraged the standardization of an open, vendor-neutral, cross-
platform colour management system architecture.

The standard defines a format (ICC, 1996) for the characterization of

devices, whether they are scanners, monitors or out-put devices. This
characterization is gathered in the device profile which provides the Colour
Management System, CMS, with the information necessary to convert
colour data between native device colour spaces and so-called device
independent colour spaces. The CMM, Colour Management Module, is the
colour space transformation model and is the engine performing the
conversion of image data from one colour space to the other. PCS, Profile
Connection Space, is the device-independent colour space used for the
transformations and CIELAB is one of the spaces recommended by ICC.
Four rendering intents were recommended for the standard. The latest
revised Format Specification came in 2001 (ICC, 2001).

ICC offers information on their website about news, and also the last format
specifications can be found there. Furthermore, additional links to vendors
with specific and valuable information for users, not only on colour
management but also, for instance, on colour models, can be found (see
some of them under References/Links).

32 (95)

Marianne Klaman Royal Institute of Technology 2002

In the beginning of the introduction of the ICC standard, there were
limitations and a lack of tools for measurement and software for the creation
of the profiles. These limitations of colour management and the current
practices in industry were discussed by (Has and Newman, 1995). They
also discussed the value of ∆E , whether reported values of visible
differences of 1.0 ∆E versus other reported values of visibility of
2.5 ∆E were to be stated. They pointed out that colour patches and images
subtend different fields of vision and that the incongruous results were not
necessarily inconsistent. This problem is due to investigation in Paper III
where colour patches and images are assessed, as well as in Paper VI, where
images are judged, and the results are discussed in the discussion part of the
thesis.

(McDowell, 1998) declared in his status report of standardization work that

colour data produced by the printing process is dependent on many
variables. “This has led to a three-tiered approach in standards development.
First, the colour of the ‘ink-in-the-can’ is defined, based on the use of
reference substrates and laboratory procedures. Then, specific printing
process control procedures are defined. These include paper, solid ink
densities, tone value (dot value) curves, overprint colour etc. With these
variables defined, a standard test pattern can be printed and measured to
provide the characterization data.

These characterization data can then be used as the reference for prepress
colour separation aims, such as the aim and control for both analogue and
digital proofing systems, and as the reference for printing process control”.

McDowell also provided information about something that was seriously

being considered: the establishment of reference printing conditions that
could be used as the basis for both proofing and data exchange referring to
CGATS.6 for process control and TR001 for the characterization data.

There are certain limitations in the ICC-Profiles Standard that are doubtless,
at least part of the reason why different colour management systems yield
different results. The main reasons that according to (Kohler, 1999) can be
categorized into three main problems: Narrow PCS, no CMM specification,
and no guaranteed interoperability. This includes the PCS not being defined
well enough in every respect of the reference space. A whitepoint as well as
a relevant appearance model have to be built into the PCS. The lack of
CMM specification can lead to a situation where two CMMs, given
identical input, can yield very different results. Included in the term lack of
interoperation, Kohler points out the problems with private tags, the issues
of rendering intents and gamut mapping. The rendering intents are not very
33 (95)

Marianne Klaman Royal Institute of Technology 2002

well defined and the definitions of the relative and absolute colorimetric
intents do not coincide with the CIE definitions. Since ICC has allowed
private tags, this inevitably causes differences in the profiles. Last, the
gamut mapping, also left open, yields different results. The CIE appointed a
committee to define a “baseline” gamut mapping method and the hope
Kohler expressed was that the ICC would quickly adopt the CIE
recommendations for such a baseline method. These facts, gained by
investigation of different colour management systems, were also considered
in the experiments performed in the thesis work and resulted in the choice
of certain systems that had shown a good capacity in the preliminary tests
preceding the empirical studies performed later on (Paper II, IV, V and VI).

A crucial task is to compare the potential of different colour management

systems. To perform a scientific, systematically accomplished study is
difficult since the different systems are based on vendor specific algorithms,
which are not available, and the comparison can easily be irrelevant.
However studies concerning the usability and user-friendly aspects are of
interest to the industry. Tools were developed by (Schläpfer and Widmer,
1996) with a checklist to give the user the opportunity to identify the basic
requirements for a given application environment and a tool for assessment
of the quality of a colour management system for a given rendering intent.
In terms of colour difference values, the colour rendering quality is
considered.

Research has continued in different fields of the colour management area, as

well as with the development and improvement of gamut mapping, colour
appearance modelling, implementation and improvement of different colour
imaging software. Lately, there has been an increase in research activity on
the workflow issues.

7.2.1 Gamut mapping

Though gamut mapping is not the field of research for the work involved in
this thesis, none-the-less this subject is of high interest. Gamut mapping is
perhaps the most crucial part of colour management and research in this
field is extensive.

(Green and Luo, 2001) presented a new gamut mapping algorithm for
newsprint. The algorithm is derived from analysis of high quality colour
reproductions. Testing of the algorithms were performed in a transparency-
to-newsprint workflow, assumed to be the most severe form of gamut
compression in commercial use. For each physical device in the colour

34 (95)

Marianne Klaman Royal Institute of Technology 2002

imaging chain, there is a finite range of colours that can be imaged on a
given substrate, depending on factors such as the colorants used, the
densities at which they are imaged, the imaging characteristics of the
device, and the surface properties of the substrate. This combination of
device and substrate is referred to in the paper as a medium. An example is
given of three media with a large difference between them. The challenge is
to reproduce a given original image onto these different media in a pleasing
and consistent way. Most gamut mapping algorithms operate on the
cylindrical coordinates of lightness, chroma and hue angle. Since colour
media do not have hue limitations, the problem of gamut compression can
be addressed in the two dimensions of L*lightness and C*chroma instead of
the three dimensions L*, a* and b*.

As Green says, the precise methods which will give the most pleasing
mapping between two gamuts are the subject of continuing research, and the
CIE has established a Technical Committee to consider the subject and
recommend one or more mapping algorithms. The newsprint field seems to
be fairly much untested from the perspective of gamut mapping, although
there is the smallest colour gamut of any reproduction medium in
commercial use. In his paper, Green discusses the different ways to perform
gamut mapping with respect to different parameters. Linear scaling or
scaling with an s-shaped lightness scaling each has its advantages as well as
disadvantages. Most studies have found that a simultaneous mapping of
lightness and chroma performs better. In these algorithms of Green and
Luo, lightness of neutrals was mapped linearly in the gloss-coated proofs,
but with an S-shaped curve in the newsprint proofs.

(Morovic and Luo a, 2001) gives a survey of the fundamentals of gamut

mapping , with a terminology list and explanation of different methods used
as the main types: gamut clipping and gamut compression. The same
authors (Morovic and Luo b, 2001) reports of an evalutaion of different
gamut mapping algorithms and they prove that the new developed gamut
mapping algorithm performs better than the initial methods on which it is
based. This new algorithm was developed during the study. They also
demonstrated that the method performed well both with CIELAB and
CIECAM97s.

7.3 Colour Appearance Models

Colour is of a complex nature. If regarded as a mere physical phenomenon
spectrophotometry could be used to characterize all of its nature. This could
be possible if certain objects or certain wavelengths of light had their given
35 (95)

Marianne Klaman Royal Institute of Technology 2002

colour. But since it is more correct to say that certain stimuli are perceived
and viewed under specified conditions in which they obtain a certain colour,
it is also necessary to include other factors to be able to describe colour
correctly. As soon as the lowest level of response of a human observer is
considered “in order to better describe our perceptions of stimuli, one is
within the domain of colour appearance modelling” (Fairchild, 1998 p.XV).
Colour appearance modelling is necessary to take into consideration when
phenomena like surrounding colours influencing the colour observed have
to be described. Merely measuring the colour would give the same value,
although it looks different depending on the adjacent colour. In digital
colour reproduction, the application of relevant colour appearance models
are crucial, since in reproducing images the image colours are not only
physical but always perceived in a certain way due to factors influencing
each single colour. There is also a need for relevant colour appearance
models in developing colour management systems.

Intensive research is also going on in the field of Colour Appearance

Modelling and improvements are searched for all the time. (Fairchild,
2001) describes the improvements made in CIECAM97s, which was
proposed by CIE TCI-34 in 1997 in response to the needs of the imaging
industry for a single, practically applicable, colour appearance model for
device-independent colour imaging applications. The model has, however,
been less successful in achieving the goal of being practically applicable
since it remains very complex. This complexity allows the model to
describe and predict various phenomena, at the same time as it makes the
model’s resource (computer craft) demanding and not user-friendly.

(Xu and Levkowitz, 1996) presented a model called Uniform Color

Appearance Model, just to mention one of the later attempts to find a model
for colour reproduction applications in the graphic arts field of media. The
model incorporates a simplified chromatic adaptation transformation with a
generalized tonal reproduction and colour correction.

One aspect of the achievement of reliable colour reproduction, not least

when concerning cross-media colour reproduction, is the need of a reliable
colour appearance model capable of predicting the colour appearance across
a variety of imaging devices under different viewing conditions.
Psychophysical experiments were carried out to assess colour model
performance in terms of colour fidelity by comparing soft-copy and hard-
copy images. The CIECAM97s-type models performed better than the other
models. In addition, input parameters for each model had a distinct impact
on model performance (Sueeprasan, Lou, Rhodes, 2001).

36 (95)

Marianne Klaman Royal Institute of Technology 2002

8 Contemporary research in related areas
Reported in this chapter is research performed in related areas to the subject
of this thesis. The survey does not claim to cover the entire area but rather
to report on the kind of research going on, and in particular research with
implications on the results of this thesis work. Some discussion related to
the work conducted and reported in the papers is included in this chapter but
the discussion is mainly placed in the chapter succeeding the summary of
papers.

8.1 Colour management applications

Research for improving colour workflows and improving routines for
colour management continues in many fields within printed media. (Lind
and Adams, 2001) concluded that by using CMS, a flexographic printer
could increase the quality to the point where it competed favourably with
gravure but with the advantage of having shorter pressruns and lower costs.
Soft proofing proved to be a viable solution, thanks to colour management
software. Similar results were assumed or predicted in Paper II, and partly
shown in Paper VI. The soft proofing advantage is crucial.

The creation of ICC profiles is subject to discussion and (Chung et al.,

2001) report debates about whether an ICC press profile is better
constructed from one press sample or from the average of a number of
samples. In a recent project they reported excellent colour matching
performance with the single sample approach. In work involved in this
thesis the one press sample approach has mostly been used but in this case
the control of the printed sheets becomes even more crucial and the need to
be careful increases to secure that the printing has been conducted
according to established press parameter values.

(Johnson, 2000) discusses the background of colour management and how

the current ICC architecture may be improved to help meet media workflow
and quality requirements. Johnson proposes a flexible colour management
architecture for cross-media reproduction that separates appearance
matching, gamut mapping, image editing and characterization. In the work
reported in Paper IV, V and VI one approach, particularly, is to find the
characteristics of the out-put part. The focus is especially on the substrate.
In future colour management systems it could be a solution, according to
flexibility, to separate the components of the output part (and along with
that also other parts of the process colour information) in order to be able to
put all the links together to a “colour chain” with the components desired in

37 (95)

Marianne Klaman Royal Institute of Technology 2002

the specific application. This should also give a flexibility according to the
demands of cross-media publishing.

(Collet, Chagas, Baudin, 1999) studied colour management in order to

evaluate the effects of using ICC-profiles and their ability to minimize
colour variations. The original was the reference and the objectives were to
evaluate the capacity of an input device, several output devices (an ink-jet
printer and a combined thermal transfer/sublimation proofing device) and a
colour monitor. They investigated the ability of these devices to render the
colour data supplied. In the experiment, different treatment of data for
stability/repeatability and difference from original of a given colour were
established. In the first step without using ICC profiles and in the second
with ICC-profiles applied in order to assess their influence on the
reproduction process. The authors pointed out that each unit in the process
chain generates its own specific colour deviation, which in turn seems to
induce a propagation phenomenon increasing the errors along the workflow.
One result was that the ink-jet printer yielded better colorimetric fidelity
than the thermal/sublimation device. They also showed that the efficiency
of a profile is strongly dependent on the exactness of the calibration and
characterization process and the importance of verifying the process
periodically. They moreover reported different behaviour of the digital
output units – the improvement of colour fidelity of the thermal/sublimation
device as opposed to the decreased fidelity of the ink-jet printer. In all the
cases, the colour gamut (defined as the volume of the colours reproduced in
the L*a*b* space) was enlarged by the application of a profile. Another
interesting result obtained from their study was that colour fidelity seemed
to be more difficult to obtain by stochastic screening than by conventional
screening.

A panel discussion concerning colour management at the TAGA/ISCC

bridge symposium was summarized in the proceedings (Chung, 1999). Tim
Kohler concluded the following about the current state of ICC and its
direction:

There is a good success with ICC-based colour management if the

devices are calibrated and the profiles and the CMM come from the
same vendor. The goal of ICC is, however, a vendor-neutral colour
management solution. Since only the format of the device profiles is
specified, there are problems in interoperability caused by lacking
definitions concerning the Profile Connection space and the Colour
Management Module. But the PCS works well for printed conditions
under D50 illumination, worse, however, for monitors. This in turn

38 (95)

Marianne Klaman Royal Institute of Technology 2002

 seems to be due to the fact that ICC does not specify a chromatic
adaptation or colour appearance model.
Since there is no explicit CMM specification, the ICC is contemplating
two kinds of CMMs. One is smart and the other dumb. The latter one
simply takes data in the profiles and interpolates them and nothing
more while a smart CMM would take also viewing conditions, white
point and other conditions into consideration.

8.2 Proofing
Schmitt reported a new test chart for proofing (Schmitt, 2001). The paper
handles the question of proofing and how to verify that a colour control
target, placed on the page in order to get it proofed effectively, reflects the
rendering of the total gamut. A specific colour control target, highly
condensed, concerning the patches used (compared with the ISO 12642
which is considered to have too many colour patches for this kind of use)
was developed and used in the tests. The distribution of the ∆E values of
this test chart and of the ISO chart were found to be similar. The main
results for the conventional printing process were reported as:
The test chart developed, effectively represents and simulates the
colour space for offset printing.
There was a difference in the reaction between the test chart and
the ISO one, concerning tone values or inking changed during
printing.

For digital proof printing Schmitt concluded that:

The test chart effectively represented the whole gamut under
various conditions such as different adjustment gradation curves.
It represents the whole gamut when different paper grades are
used.
The results also showed that the method used in the calculation of
the ∆E value with the chi∧ 2 distribution (Dolezalek, 1990) was
preferred in this case.

The fact that the colour patches being spread evenly over the whole gamut,
as is the case with the ISO chart, probably gives the key of success. In Paper
I it was shown that at least 30 colour patches, distributed in the gamut, was
one of the crucial requirements for a relevant assessment of the ability of
the proofing system to simulate the print correctly.

In a study by (Shimamura et al., 2001), ICC-based digital proofing

performance with an actual press run was conducted to test the colour
39 (95)

Marianne Klaman Royal Institute of Technology 2002

matching performance between the digital proof and the press sheet. They
found that with the modifications used (one-ink-zone profiling target and a
modified one-ink-zone IT8.7/3 basic target as a means to minimize the
influence of press variability), the ICC-based digital proofing works well in
simulating the press sheet in actual production environment. They also
found the use of the cumulative relative frequency curve, CRF curve, to be
a valuable tool. But still, visual assessment of targets showed noticeable
colour difference between proof and print.

(Chan et al., 2000) describe a strategy for evaluating digital proofing

performance under optimal testing conditions. In order to eliminate the
press variability, instead of press runs, the ANSI CGATS TR 001-1995 –
Type was used, and an improved digital performance was gained. Besides
the elimination of the press run, a stable digital proofing device with a
sufficiently large colour gamut together with improved ICC profiling and
colour management tools were used.

In remote proofing, the use of ICC-profiles implies a stable process. This

was shown in a study concerning remote proofing (Klaman, 2000). The
choice of an appropriate paper for the remote hard copy proofing unit was
also important in order to obtain a good colour rendering. There was a
problem in deciding the tolerance level for accepted proof compared with
the approved print, since the proofs oftentimes showed good agreement in
some colours but differed in others.

(Hulsman and Notermans, 2000) claim that many high end users demand
the average ∆E<1 and maximum ∆E<2 for contract proofs versus print. An
interesting and useful solution that the authors predict is a small inexpensive
spectrophotometer to be placed within the proofer to provide auto-
calibration.

8.3 Colour gamut

(Nurmi and Sivonen, 2001) showed how much the colour gamut is changed
over time and how much the printing on the opposite page affects the
gamut. Test images were compared with prints without double-sided prints
and ∆E values were calculated. The most sensitive colours on the opposite
side were the yellowish and reddish ones and the smallest changes were
obtained in blue and green. The most obvious changes appeared in the most
chromatic colours. The decrease in gamut volume was found to be up to
18% in three days. The largest changes were also found when there was no
print on the opposite side or when maximum ink coverage was printed on
40 (95)

Marianne Klaman Royal Institute of Technology 2002

that same side. The reduction of gamut size was caused by print on the
opposite side and changed the colour values with 2-4 ∆E with largest
change in chromaticity.

(Nordström et al., 2001) calculated the size of colour gamut for offset (in
this case waterless offset) compared with a digital toner transfer press, the
latter being approximately 24% less.
.
In a study performed by (Bolanca and Mrvac, 2001), the results indicate that
digital liquid toner gives almost the same gamut size irrespective if the
substrate is a glossy or matte fine art paper. The prints in offset give
increased size on glossy compared with matte coated papers. The authors
also accuracy in using seven-colour printing for high quality products
despite increased process costs. Conversely, six-colour printing in digital
offset improves the gamut only negligibly and the opinion of the authors is
that it is difficult to find the economical justification for a six-colour process
in this case.

Based on a model for ink penetration, (Yang and Kruse, 2001) showed that
the effect on colour rendering for multichromatic (halftone) images were
simulated with consideration taken of optical dot gain. The result from the
simulations showed a remarkable colour deviation, both hue shift and
saturation reduction on the colour image rendering. Also, the effect of
colour gamut reduction was shown. In an earlier paper, (Yang, Kruse,
1999), the ink penetration was modelled with respect to the density of
penetrating ink and the colour gamut was also here shown to be reduced by
penetration.

A suggestion of classification of colour gamuts is based on an investigation

of numerous combinations of process/substrate and colour measures of
these (Schläpfer, 2000). The intention is to use these gamuts in the
comparison of different processes as conventional and digital, as a
quantification tool for reproduction processes. Such a case can be the
utilisation of more than four primary colours where the intention is to
enlarge the colour gamut and the need for a quantification tool is obvious.

The difference of the colour and light scattering properties of the stocks is
one of the main factors in the colour gamut differences showed by (Zawacki
et al., 1999).

Green (Green, 2000) describes a new target defining media gamut

boundaries. He claims that the use of measured boundary data, provided by
making use of this target, is capable of improving the accuracy of gamut
41 (95)

Marianne Klaman Royal Institute of Technology 2002

boundary prediction in both device model and gamut mapping
computations.

In research work by (Berns, 1999), he shows that the CIE94 formula implies
that differences in colour gamut between multimedia devices may not be as
large as it appears when looking at CIELAB gamut volumes. In a
comparison of two gamuts, one for a display and the other for a printer, he
obtains a noticeable difference in gamut volume. He claims that the large
differences in Euclidean distance are in fact less noticeable visually. Berns
also reports the “considerable interest in the colour difference research
community to develop a new colour space that is corrected for this chroma
position dependency” (1999, p. 121).

8.4 Colour and colour difference measures

Continuous and active research is taking place to find even more accurate
colour difference methods, entirely new or those based on existing
formulae.

The last one presented is the CIE 2000 colour-difference formula,

CIEDE2000 (Luo, Cui and Rigg, 2001). This equation “has been officially
adopted as the new CIE colour-difference equation. It is based on CIELAB
and includes not only lightness, chroma and hue weighting functions, but
also an interactive term between chroma and hue differences for improving
the performance for blue colours and a scaling factor for CIELAB a*-scale
for improving the performance for gray colours” (Luo et al., 2001 p. 340).
The equation was tested together with other advanced CIELAB based
equations. For the test, special colour discrimination data sets, based on
object colours, were used. The authors showed that the CIEDE2000 formula
“outperformed CMC and CIE94 by a large margin” (2001, p. 340). The
equation is not included here but for those interested it is found in the
reference or in (CIE, 2001). The CIE is at the time developing a draft
standard based on the report.

(Johnson and Green, 2001) and (Johnson and Green, 1999) have performed
experimental work to evaluate the magnitude of colour difference that can
be accepted for critical comparison in graphic arts media. The CMC
formula is discussed and though it is still widely used in the textile industry,
it has found little application in media industry. It appears to give
substantially better uniformity for small differences in colour. Also
discussed is the DE94 by which the main uniformity error is corrected;
namely, that pairs of colours of high chroma produce far larger numerical
42 (95)

Marianne Klaman Royal Institute of Technology 2002

differences than their appearance difference justifies. DE94 is non-
Euclidean and dependant on which sample is taken as reference. This has
been corrected in CIE2000. But the authors in fact clearly proved that
neither CIE94 nor CIE2000 offer any significant improvement in producing
a uniform visual scale over CIELAB. They conclude that there is no or very
little indication as to whether the acceptability weighting factors introduced
by CIE94 could be useful in the graphic arts media industry. Results show
that perceptible differences are obtained at around 1 to 2 ∆E ab units. For
acceptability CIE94 proved far more useful. This result can be compared
with (Paul, 1998) who found a low correlation between ∆E ab and visual
ranking but improved this significantly when using ∆E 94 . (Bassemir,
1995) on the other hand found reasonable correlation between perceptibility
differences and ∆E ab . These results indicate that further research in this
area is required. (Johnson and Green, 2001) also address the fact that
relatively few studies have investigated acceptability in colour difference. In
a study of acceptability and perceptibility in complex images they found a
mean perceptibility tolerance of 3∆E compared with 6∆E for acceptability.

(Chung et al., 2001) claim that there is no easy way to assess colour
difference quantatively between two complex images. In their study, they
used a standard press characterization target (IT8.7/3 basic block) to
quantify pictorial colour image difference by colorimetry. The paper shows
a way to use the Cumulative Relative Frequency (CRF) of the ∆E
distribution for the quantative analysis. A visual match could be specified
by means of the CRF curve and its derived statistics. In addition, a visual
match between two colour images imposed a tighter colorimetric tolerance
than process conformance as specified by solid ink density and dot gain.

(Sørensen, Jørgensen, 1993) showed in a study that there was a significant

correlation between colorimetric ∆E-values and the percentage of persons
accepting the certain printed specimens. They also showed that smaller
visible differences were to be expected if the test images were placed side
by side as opposed to when they were placed with a space of 2 mms in
between, leaving a white line separating them.

43 (95)

Marianne Klaman Royal Institute of Technology 2002

8.5 Factors influencing colour
(Nordström, 2000) reports from a study concerning colour shifts caused by
substrate, ink composition and halftoning. The coated substrate affected
very little in comparison to halftone and ink. The ∆E value for glossy/silk
was below 1.5, which Nordström gives as the visible value. Through
spectral measurements, we may still determine how different variables
induce colour shifts; however, we cannot clearly determine the actual
reason. Nordström claims that a trained pair of eyes may detect differences
of ∆E ≈ 1,5.

In a study performed by (Baudet and Rousset, 2001) the influence of paper

properties on print quality in ink-jet printing was investigated. They
proposed a potential simple method to calculate ink penetration into the
paper by print-through measures.

A survey on subjective print quality was conducted by (Gast and Tse, 2001)
also in terms of colour rendition and tone reproduction. They used a method
that induced certain adjustments to the original in order to cause controlled
deviations in the parameters chosen. The panel then had to choose which of
four proofs they preferred. Concerning colour rendition, most of the
observers disliked a greenish cast in the image, and thereafter magenta.
Cultural difference was indicated to be very strong in colour preference.
When it comes to lightness, most of the observers preferred a darker image,
which was assumed to give an impression of having higher contrast and
sharpness.

Paper properties influencing print quality is the subject discussed

concerning the relationship between a high-speed continuous ink jet system
and the printing substrate (Heilmann and Lindqvist, 2000). Non-absorbent
materials, with different topography of the surface, were used to evaluate
the importance of the droplet impact and spreading. Commercial papers
were also used. The results showed that the roughness and the absorption
properties of paper were found to be of great significance to the print
quality in continuous ink-jet printing. It was found under varied process
conditions that the roughness and the orientation of the printing surface had
a strong impact in the final size and the shape of the dot.

(Heilmann and Lindqvist, 1999), showed that with coated paper grades, the
interrelation between the drop size and the magnitude of the surface
structure was found to be important, compared with the final shape of the
dot. With uncoated paper, the growth of the dot is dependant on the size of
44 (95)

Marianne Klaman Royal Institute of Technology 2002

the drop. The final intensity of the print was found to be the outcome of the
characteristics of the ink and the absorption properties of the paper.
The aim was to:
Analyse and model the interaction between ink and paper.
Develop testing method for ink-jet substrates.
Discussion of macroscale and microscale; small microscale differences
of the topography of the paper strongly affect the size and the shape of
the dot, when the printed dot is small.
Properties, especially absorption, can strongly reduce the darkness of
the final print.

The difficulty in determining print quality using technical measurements

alone was addressed in a work concerning ink-jet printing and paper quality
also by (Lindqvist, Heilmann, 1999). The authors pointed out the problems
of the objective and subjective measurements compared and agreements not
found. They gave an example of two printers giving the poorest detail
rendering but despite that, producing the best visual images. They found,
which is important, that paper dependence could probably be measured by
technical means and that visual tests are then not needed for this purpose.
Which means that this attribute may be used to characterize the printers as
well as the product development of digital printing paper.

The Neugebauer equations have provided an accepted basis for theoretical

simulations of the formation of multicolour images in various printing
processes. (Juntunen, Jäntti, Lindqvist, 1989) showed some new ways of
estimating the colour values with different screen, dot structures (those used
in conventional or in non-impact printing) and with some disturbance
factors, such as the errors in the amounts of ink and misregister introduced
and tested. With the model, they could understand the effects of errors on
the geometric factors of colour formation and no distinction was needed
between the conventional and non-impact printing.

(Juntunen and Virtanen, 1991) also showed that experimental results

indicated that the effects of roughness dominate the formation of print
quality in non-impact printing. Absorption affects the fastness of the dots
and controls the final dot size in ink-jet printing. According to the results,
they suggested that paper roughness is the first factor to be minimized in
order to reach an optimal printing result in non-impact printing. They did
not, however, analyse the colour quality of prints, but other print quality
factors.

45 (95)

Marianne Klaman Royal Institute of Technology 2002

 46 (95)

Marianne Klaman Royal Institute of Technology 2002

9 Summary of original work

9.1 The author’s contribution to papers

All the work performed and presented in the different papers have been
initiated and formulated by the author. The author has also raised funds,
arranged network cooperating and been project manager in the projects. The
papers have mainly been written entirely by the author, with the exception
of Papers III and IV where the author has performed the analysis and
written all text and I. Anderson conducted most of the experimental work,
produced the diagrams and tables and contributed with valuable discussions.

Paper I - A Test Model for Proofing Systems

Klaman, M., 1st Joint TAGA/IARIGAI Technical Conference, Paris,
September 17-20, 1995. Proceedings, Advances in Printing Science and
Technology, vol. 23, pgs. 163-175.

Paper II - Improved Productivity and Image Quality by using Colour

Management Systems
Klaman, M., IARIGAI Conference, London, September 1997. Proceedings,
Advances in Printing Science and Technology, vol. 24, pgs. 3-14.

Paper III - Colour Shifts in Four Colour Printing

Klaman, M., Anderson, I., (1999), paper presented by Klaman at TAGA,
Vancouver, May 1999, TAGA Proceedings 1999, 15 pages.

Paper IV - The Influence of Paper on ICC-Profiles

Klaman, M., Andersson, I., (1999), paper presented by Klaman at TAGA,
Vancouver, May 1999, TAGA Proceedings 1999, 21 pages.

Paper V - The Influence of Paper Whiteness and other parameters on

the creating of ICC-profiles for digital colour printers and conventional
offset presses
Klaman, M., paper presented at the 26th IARIGAI conference, Munich,
September 1999, Proceedings, Advances in Printing Science and
Technology – Advances in Digital Printing, 13 pages.

47 (95)

Marianne Klaman Royal Institute of Technology 2002

Paper VI - Colour Rendering Aspects in Digital Printing
Klaman, M., paper presented at TAGA, San Diego, May 2001, TAGA
Proceedings 2001, 23 pages.

Appendix – The Comparison of Colour Gamuts

Bristow, J. A., STFI, Klaman, M., GFL, Hoc, M., STFI, IARIGAI
Conference, 1993. Proceedings Advances in Printing Science and
Technology, vol 22, pgs. 311-332.
This paper was written by J. A. Bristow in cooperation with the author and
M. Hoc.

9.2 Paper I – A test model for proofing systems

Proofing is an important tool and step of colour reproduction in order to
achieve correct print quality, included colour. The assessment of different
proofing systems addressing especially the potential of colour simulation,
demands a suitable test model. Routines must be established to compare and
evaluate the print and the proof and to judge the potential of different
proofing systems.

The aim of the work presented in this paper is:

The development and validation of a test model, primarily aimed for testing
digital proofing systems, but with the intention of also making it applicable
for other colour comparisons.

The test model developed combines densitometric and spectrophotometric

measurements. The solid tone density, the density deviations between print
and proof and the density of the tone curve are discussed as measures of the
ability of the proofing system to simulate the print result when different
grades of printing paper are used. The importance and significance of the
assessment of dot gain is also considered.

The colour gamut and its effect on colour rendering are extremely important
for the ability to achieve a good colour matching between proof and print.
Besides the ∆E*-values, the significance of the areas and volumes of the
triangles, hexagons and dodecahedra formed by the colour coordinates in
the CIELAB-diagrams of the print and the proofs are considered.

48 (95)

Marianne Klaman Royal Institute of Technology 2002

The proof must be produced with the capability of predicting what the print
will look like and as (Bruno, 1986 p. XI) says, the proof “must be capable
of being produced consistently, reproducibly and predictably".

The research approach taken here is the development of a test model with
the desired qualification to compare proofs and the final print in order to
analyse the proofing systems ability to simulate the print and to fulfil the
requirement stated above.

Prints were made on both coated and uncoated paper and the different
proofing systems were used to produce proofs simulating both these
printing situations. In this paper, results obtained with three representative
proofing systems are selected, together with the offset prints to illustrate the
merits of the parameters discussed.

The values obtained from different measurements are compared with visual
judgements of prints and proofs to establish the most relevant methods for
the test model.

The solid tone density is one of the crucial parameters analysed. The result
indicates strongly the importance of the density levels for the proof and the
print to be the same, and the sequence for the density values of the different
colours also to be the same. Moreover, the density of the proof should vary
according to the type of paper that the proof seeks to simulate.

Additionally, for a high quality result, the density of the tone curve of the
proof has to follow the density of the print and deviations must be
proportional to that in the solid tone. Proof B was the proof chosen from the
investigation being the one that performed the best result. Figure 6, a and b,
illustrates the ability of the proof to simulate the behaviour of the tone curve
on coated and uncoated paper.

49 (95)

Marianne Klaman Royal Institute of Technology 2002

Figure 6, a and b. The best result for simulation of proof with print is obtained with
Proof B performing a density tone curve in accordance with the tone curve of the
print.

In the case of the dot gain, there is no immediate correlation between the
visual appearance of the proof and the print and the difference in dot gain
values between them. This is assumed due to the fact that fairly different
techniques are used in the different proofing systems. Although the absence
of real dots in many systems, due to technique, presumably affects the so-
called dot gain, it seems to be important that the dot gain value is flexible.

50 (95)

Marianne Klaman Royal Institute of Technology 2002

Another parameter to be taken into account is the available colour gamut,
here described by CIELAB. In the first place, the two-dimensional gamut in
the chromaticity plane is considered. This simplification is regarded as
relevant, because the eye is more sensitive to differences in chroma than to
differences in lightness. People accept an image that is slightly different in
lightness rather than an image that differs in chroma when colour rendering
is the main issue.

*
The ∆E ab value is used for the definition of differences between the proofs
*
and print. In the test model, the individual ∆E ab -values for the six colours
C, M, Y, B, R and G are used. These six colours prove not to be sufficient
to decide whether the proofing system has the potential to simulate colour
*
rendering. About 30 more colours with the ∆E ab -values for these colours,
*
have to be included in the model. Then the ∆E ab -values for these
approximately 30 colours give a good correlation with visual judgement and
thus provide a good measure of the degree of agreement between the colour
gamut of the proof and that of the print.

As the result of an attempt to break the gamut information down into

smaller fragments, it is found that the areas of the triangles formed by the
coordinates for C, M, Y, B, R and G and the paper white O, see Figure 7,
are useful parameters to show the degree of similarity between proof and
print.

The differences in lightness ∆L* are less important. In the simulation of the
print, colour rendering is important. As the eye is more sensitive to
differences in the chroma and less sensitive to differences in lightness, it is
may be reasonable to simplify the calculation for the areas to the two-
dimensional projection on the chromaticity plane. The areas according to
this simplified procedure were calculated, using the same values for the
paper white as to the print and the proofs.

51 (95)

Marianne Klaman Royal Institute of Technology 2002

 +b*
Y

R
G
O M
-a* +a*

B
C

-b*

Figure 7. The triangles formed in the chromaticity plane by the projection of the
coordinates for CMY and RGB and the paper white O.

If instead the areas are calculated as they appear in the three-dimensional

gamut, the values obtained were slightly different, and the extent of the
discrepancies between various types of paper depended on how much the
lightness differed. The conclusion from the calculations made are that, as
long as the ∆L*-values are not too large, the simplified model can
justifiably be used to calculate the areas.

The volume of the gamut was assumed to be a useful parameter for

evaluation and comparison.

It was, however, found that the total volume of the gamut does not correlate
well with the visual appearance, due to the fact that deviations in the
coordinates for a* and b* can be compensated for by deviations in L*. It is
possible to have the same volumes for the total volume gamuts but still have
a very large difference in colour appearance. The volumes therefore seem to
be too insensitive to be a good measure for assessing the potential of the
proofing system.

For calculating the grey balance the ∆E ab distance in the chromaticity

plane can be used, where:

∆E ab = (a *)2 + (b *)2 [Eq. 16]

52 (95)

Marianne Klaman Royal Institute of Technology 2002

The advantage of this approach is that, instead of two curves, only one
curve needs to be analysed for each system, see Figure 8.

Figure 8. The . ∆E ab curves for the proofs compared with the print.

The conclusion is that most of the parameters discussed have to fulfil the
stated requirements if the proofing system is to have a high potential to
predict the print result. The solid tone density, the sequence for the density
values of the different colours and the density of the whole tone curve are
important. The gamut of the proof has to perform the same size, shape and
variability as the print. The total area as well as the total volume of the
gamut is not useful, since even though they have the same area or volume,
two gamuts still can differ in shape.

This is due to the fact that coordinates can compensate each other and still
give the same size but with a significantly different achievable gamut. The
comparison of areas of the separate triangles proved to be a useful tool.
*
Concerning colour rendering the ∆E ab values for at least 30 colours must
be calculated and compared. These 30 measured colours give a good corre-
lation with the visual assessment and thus provides a good measure of the
degree of agreement between the colour gamut of the proof and that of the
print.

The comparison of grey balance values must be included and added with a
total quality analysis of text, vignettes and detail rendering.

53 (95)

Marianne Klaman Royal Institute of Technology 2002

9.3 Paper II – Improved Productivity and Image
Quality by using Colour Management Systems
Discussed in this paper are the possible benefits or disadvantages of
working with accurate colour reproduction with separations made with
colour management systems and profiles based on the ICC standard.

The development of commercial systems for Colour Management and the

development of the ICC standard are discussed in terms of the benefits of
working with controlled colour management based on the ICC standard
versus working the traditional way.

The International Colour Consortium (ICC) defined the basics for colour
management systems and a standard format (ICC, 1995, 1996) for device
profiles, which to a high extent increased the opportunities for graphic arts
companies to implement a functional solution for colour control. The
colour space transformation, gamut mapping and the creation of device
profiles are mathematically calculated and based on measurements of the
actual colour behaviour for each specific device.

Compared with conventional work, with more simple calculations, it is

certainly an advantage to work with a colour management system, but there
is reason to be aware that the resulting colour depends on the quality of the
software. The production workflow and organisation probably have to be
changed with the implementation of a colour management system. For
example, a number of device profiles have to be created and then handled in
a proper way. This will lead to the need for some kind of profile
management.

In order to gain high image quality with a conventional prepress system

without colour management, the normal and correct way to collect
information for the prepress work is to determine the print curve, which
includes the dot gain at the optimum density level. At this level and with
correct dot gain, the grey balance is determined. Correct grey balance is a
requirement also for correct colour balance. This way to work is based on
one-dimensional transfer curves and also depends on the skill of the
operator. The result is often good.

The algorithms in the different Colour Management Systems, both on the

operative system level and in the software, will be crucial for the image
54 (95)

Marianne Klaman Royal Institute of Technology 2002

quality. The essential parameters are the colour space transformation
algorithms, the creation of the device profiles and the gamut mapping
algorithms.

Considered in the paper is the fact that different vendors have very different
numbers of colour patches in their test charts designed for the creation of
the ICC-profiles. Some claim that a limited number is sufficient and others,
to the contrary, that a large number is needed. Hypothetically, the number
of patches required depends to a high degree on how the patches are chosen,
i.e. where they are situated in the colour space. If they are chosen in an
appropriate way and the algorithms used are of high quality the number can
probably be reduced.

The colour gamut for different devices differ not only in size but often also
in shape, and this means that the mapping of colour values which fall
outside the reproducible colour gamut is not a simple linear transformation
but also requires a compression and a distortion of the colour gamut. The
gamut mapping is assumed to be perhaps the most crucial part of the colour
managing calculation.

This is considered in the case with a proof going to simulate a print and
where the colour values of the proof are falling outside the gamut of the
print and having to be mapped into the gamut. The way in which this is
done greatly affects the final quality. Different vendors solve the mapping
in different ways, and the algorithms used are crucial for the success of
achieving higher image quality. The colour values falling outside the
reproducible gamut are sometimes merely moved inside the boarder of the
gamut and this can cause an odd appearance of the image. Difficulties occur
when the compression has to be large, as for instance is necessary for the
compression to the gamut of newsprint. In tests where differences in image
quality due to the software used were evaluated, these differences were
assumed to depend specifically on the gamut mapping calculations.

As one of the benefits working with ICC characterization of the monitor,

soft proofing is possible to use and processing time could be reduced.

The importance of a stable, repeatable and calibrated process that also

includes the viewing conditions (illumination and surrounding) is
considered to be high.

The differences in the number of processing steps between conventional

colour control and ICC-based are calculated. There is a difference, although
not a large one. On the other hand, it is important to consider how often it is
55 (95)

Marianne Klaman Royal Institute of Technology 2002

necessary to make corrections after producing a proof and also how flexible
the operator is when it comes to correcting images reproduced by someone
else. Some of the benefits of the production workflow included a Colour
Management System are analysed to be:

Any operator can make changes to images reproduced by someone

else.
Incoming light (when working with a monitor) is controlled.
Monitors can be calibrated individually, according to each other
and according to customer’s monitors.
Calibrated digital proofs are accurate enough to be used as proofs
to be sent to the customer.
Only the final changes to the job produce a set of films.
Soft-proofs can be sent over the network to customers with
calibrated environment.
High quality can be obtained even though the operator has little
knowledge of colour.

9.4 Paper III – Colour Shifts in Four-Colour Printing

The importance of a stable, predictable and repeatable process is a basic
demand in colour reproduction. Since the printing press to a high degree
determines the end quality of the printed product, it is of great value to
examine to what degree variations in the inking level affects the colour.
This is also highly relevant since, however controlled the printing press is,
natural fluctuations will always occur in the inking.

The aim of the work was to study variations in the inking level and the
colour shifts caused by these variations and to state the relationship between
these colour deviations perceptionally assessed and technically measured.

When printing with four colours or more, colour shifts depending on natural
variations in the level of inking in the printing press can occur. This
variation will have a negative influence on the grey balance as well as on
the colour balance, since the variations of the process inks occur randomly
and individually, i.e. some of the inks can vary in one direction and some in
another. This will lead to the increase or decrease of density and dot gain
and it will occur independently for each colour and thus cause a colour
shift.

A test form was printed at the optimum level and then with changed density
levels, downwards and upwards, for cyan, magenta and yellow, with the
56 (95)

Marianne Klaman Royal Institute of Technology 2002

ink density changed one at a time. Test targets and photographic images
were analysed at the different levels. The test charts used are parts of the
IT8-testform (ISO12640, 1997).

In an earlier investigation (Klaman, Anderson, 1992) it was analysed the

degree to which the inking level could deviate without causing an
unacceptable grey balance. That study gave a tolerance limit of the solid
tones density of ± 2.5 %.

In investigations made by NPIRI Task Force on Color Measurement

(Bassemir, 1995, a and b) colour patches were analysed according to
correlation between visual evaluation and measured colorimetric data.
Results from these investigations show good correlation between the
parameters where the colorimetric data were mostly related to hue changes.

Since the optimum value for each colour existed not only in proof Zero but
also in all other proofs where some colour had been changed, the small,
unintended density mean deviation ∆D (due to the difficulty to keep the
press inking level completely unchanged during the test-run) was calculated
initially. The conclusion drawn from this test is that even small density
deviations (unintended changes) cause colour differences with an
*
approximate value of ∆E ab ≈ 0,7 . About 40 % of the panel estimated the
proofs as different.

The colorimetric data were calculated according to CIELAB. The colour

*
difference was expressed as ∆E ab and as ∆E CMC . The former parameter is
mostly used in the graphic arts literature as a standard (ISO 12647-1, 1996).
∆E on the other hand, is preferred by the textile and plastic industries,
CMC

but can be of interest for the media industry as described by e.g. (Pike,
1993), (Clark et al., 1988) and (McLaren, 1986).

Colorimetric values for cyan, magenta and yellow as well as for red, green
and blue, were measured. Ten patches of two- and three colour overprints
distributed within dark, middle- and light tones were chosen and measured.

Visual judgements were carried out to analyse the influence of the density
changes on visually experienced colour. A panel consisting of 10 persons
evaluated images and colour patches. The proof without density changes
was the reference and the other proofs were compared with the reference for
being alike or different. Alike in this case means no visual difference. The
difference was also described as more bluish, darker or similar. The images
57 (95)

Marianne Klaman Royal Institute of Technology 2002

that were classified as different also were estimated with respect to be
acceptable or unacceptable.

The colour differences between proofs at the optimum level, and the proofs
with density changes were calculated as follows.
For the proofs at the optimum level, the values of L*, a*, b*, C*ab and hab
were calculated. In the same way the values for the proofs with density
*
changes were calculated and the differences ∆L*, ∆a*, ∆b*, ∆C ab and
∆H ab as well as the values of the total differences ∆E ab
*
and ∆E CMC were
measured for the solid tones of cyan, magenta and yellow and the
corresponding two-colour combinations blue, green and red.

The following conclusions are drawn: Density changes give different

*
magnitudes to the colour difference value of ∆E ab . A density change of ±
* *
0,2, for instance, give the values ∆E ab ≈ 3 for cyan, ∆E ab ≈ 5 for
*
magenta and ∆E ab ≈ 7 for yellow. The most obvious change in colour is in
*
the chromaticity components with a change of the values of ∆C ab and
∆H ab .

The outcome of the perceptual judgement produced the result that according
to the initial tests where 40 % of the test panel estimated the small
unintended density deviations to cause a difference, an acceptance limit was
decided to 50 %, i.e. a limit with a clear marginal to the 40 %. This 50 %
limit is the level where 50 % of the panel perceived the proof as being
different from the reference.

The density changes of cyan, especially for decreased values were

considered more acceptable by the panel than those for magenta and yellow.
Density changes of yellow were the ones least acceptable, i.e. the test panel
did not fully accept any of the images. For magenta, the tolerance limit was
dependant on the image character. Images with skin tones had a more
limited tolerance interval than other images. It could be concluded that the
*
same change in density gives different values for the ∆E ab . Density
changes of yellow gave the largest visually perceived difference and cyan
the smallest. The values of ∆E∗ab showed the same results, i.e. the highest
value for yellow and the lowest value for cyan. The measured data and the
visually judged data show a good correlation where a negative value, for

58 (95)

Marianne Klaman Royal Institute of Technology 2002

instance of ∆b ∗ with a larger part of blue component also gave a visual
appearance of more bluish than the reference.

The level of unacceptance clearly stresses the greater influence of changes

in yellow as opposed to cyan (and magenta), se Figure 9, a and b.

Density change, cyan

100
Musicians
Perceived difference%

80 Fruit Basket
Portrait
60

40

20

0
-0,2 -0,1 0,1 0,2
∆D

Density change, yellow

100
Musicians
Perceived difference%

80 Fruit Basket
Portrait
60

40

20

0
-0,2 -0,1 0,1 0,2
∆D

Figure 9, a and b. Visual judgements of the three images included in the test at the
different density levels. The black staples indicate the percentage of the panel that
had judged the image quality as unacceptable in comparison with the reference,
which is the image at the density level without change.

59 (95)

Marianne Klaman Royal Institute of Technology 2002

A tolerance limit of the solid tone density of ± 0.1 could be considered
relevant for an acceptable quality. It must be concluded, however, that it
would be preferable to keep the deviations of yellow lower than the limit of
± 0.05 since a deviation of yellow of ± 0.05 up to 0.1 is comparable with a
deviation of cyan of ± 0.2D.

A density change of ±0.2D gave the magnitude of colour difference with the
* * *
values ∆E ab ≈ 3 for cyan, ∆E ab ≈ 5 for magenta and ∆E ab ≈ 7 for
yellow.

9.5 Paper IV – The influence of Paper on ICC-

profiles
As is pointed out in paper II, a large amount of profiles to be handled in the
workflow can be the result of introducing work with colour management
and ICC-profiles. There is an evident risk that the amount of profiles that
have to be created and used can easily become too large to be handled in the
workflow in a practical way.

In the work presented in Paper IV, the attempt is therefore to reduce the
required amount of profiles needed by a systematically performed analysis,
including the development of a simple test model to show the tolerances for
ICC-profiles concerning the substrate. The question asked: can an ICC-
profile created for a specific substrate be used for a wider range of
substrates with preserved quality and to what extent, expressed in terms of
paper quality tolerance or certain paper property tolerance? The different,
most crucial paper parameters are investigated in terms of influence on
colour rendering.

An ICC-profile created for a printing press or a printer includes all the

parameters that influence the colour rendering. These consist of the
influence of the press and the materials used, such as paper, ink, blanket and
plate. Many graphic arts media companies use different paper qualities as
well as different printing presses.

In this study, an evaluation is performed in order to characterize the

influence of different paper parameters on the ICC-profiles and on the print
quality. The difference in colour rendering between different paper
∗
categories was expressed as ∆E ab and in terms of being alike or different
determined by visual judgement. The correlation between these values and
paper parameters such as surface roughness, whiteness and absorption, was
60 (95)

Marianne Klaman Royal Institute of Technology 2002

studied. These parameters were assumed to have a strong impact on colour
rendering. The influence of the whiteness (used here as an expression for
the shade) was studied as well as the surface roughness, representing here
the structure of the paper. The absorption was not measured but studied
with microscope methods.

The model to study the tolerance of ICC-profiles was based on the

crosswise use of the ICC-profiles. They were applied in such a way that
each profile was used for the paper that the profile was created for as well
as for the other papers. This was performed, with the exception being that a
profile created for a coated paper was not used for an uncoated paper and
vice versa, since the colour rendering tests showed a large difference
between these categories of paper.

To study the influence of the paper quality, L*, a* and b* were measured
for a number of light and dark tones and in tone scales for the primary
colours cyan, magenta and yellow. In order to be able to compare the colour
values from the colour patches, they were divided into light and dark tones,
the definition of dark and light according to the total amount of primary,
secondary and tertiary colours. ∆E*ab according to (ISO 12647, 96) was
used as comparison value for the colour rendering variation.

The proofs were divided into groups where the paper with its own profile
was the reference with which the other papers with the same profile were
compared. The image was evaluated according to its identity with the
reference. The term identical in this case was defined as no or a very small
visual difference. The evaluation was made separately for skin tones and for
bright or vivid colours.

Prints were also analysed with an ESEM (Environmental Scanning Electron

Microscope) and with light microscopy.

Results
The difference in colour rendering increased, following the order of glossy
coated paper, silk and matte papers when compared with a glossy coated
paper. The difference between the glossy coated paper and the uncoated
papers was considerable.

It was shown that the motif of the images had a strong impact on the
possibility to use “incorrect” profiles. When using “incorrect” ICC-profiles,
it is less suitable to use it for images with skin tones than for images with
saturated colours. Listed below are the main results regarding the tolerance
for using ICC-profiles for a category of papers:
61 (95)

Marianne Klaman Royal Institute of Technology 2002

 Glossy paper can be grouped together and the same profile used.
Matte coated paper with silk finish can be included in this group as
well.
A profile created for a glossy coated paper should not be used for a
common matte-coated paper, since it will probably not give a satisfying
image quality. A profile created for a matte-coated paper is not
recommended for use on a glossy coated paper.
When ∆E* > 2 a majority of the persons of the panel seemed to judge
the images as different.
Less than 50 % of the panel judged the images to be alike when the un-
coated papers were evaluated. Since uncoated paper can be very
different, the recommendation for uncoated papers must be, at least
when high quality is aimed for, to create a specific ICC-profile for each
paper.
If there is a demand for a single ICC-profile that can be used for glossy
coated as well as for matte and silk-coated paper, a profile created for a
silk-coated paper is to be recommended. It will be optimal as a
compromise profile for these groups of paper.

The study with the ESEM microscope showed that:

The glossy coated paper had a smooth surface with a relatively even
dot edge. The dots were round with an even surface. The matte coated
paper had a somewhat less smooth surface. The dots were still round
but had a little unevenness at the edges compared with the glossy
coated paper. The uncoated paper had a very rough surface with the
fibres very clearly exposed. It was difficult to find out where the dots
were located. The edges were uneven and the ink did not cover the
unevenness of paper. Deep cavities could be seen in the surface of the
paper.

The study with the light microscopy focused on cross-sections in the Z-

direction of the printed paper, and the result given was:
The glossy coated paper had a very smooth surface and the ink was
distributed evenly on the surface and did not penetrate into the
structure.
The matte coated paper did not have as smooth a surface as the glossy
coated paper. The surface was somewhat wavy and the ink had
penetrated the structure here and there. In some areas the ink was
missing.
The uncoated paper had a very rough structure where the ink penetrated
into pores and cavities.

62 (95)

Marianne Klaman Royal Institute of Technology 2002

In order to analyse the surface roughness and its influence on colour
rendering the ∆E*-value for unprinted paper and for dark and light tones of
the colour patches were calculated and plotted as a function of the paper
surface roughness values. The reference was chosen as one of the coated
papers. It was shown that the roughness would influence the colour
rendering with the largest influence in the dark tones, i.e. when more ink is
printed on the paper. If there is a large difference in roughness values
between two kinds of paper, the possibility of using the same ICC-profile is
diminished. The roughness values and whiteness values were also compared
with the visual evaluation values by plotting these as a function of each
other. The values were expressed as ∆Roughness and ∆Whiteness. It could
be concluded that the more the roughness or the whiteness differed from the
reference, the more the persons of the panel judged the images as different.

According to the microscopy studies, it has to be stressed that the absorp-

tion of ink must be considered to have an important influence, considering
the roughness parameter. The ink distribution is, due to the different surface
roughness and absorption characteristics, different for different paper
categories and in consequence the colour appearance becomes changed.

9.6 Paper V- The Influence of Paper Whiteness and

other parameters on the creating of ICC-profiles
for digital colour printers and conventional
offset presses
The study performed and presented in Paper V is expanded from the study
presented in Paper IV to a digital colour printer and, accordingly, a
comparison between offset and this digital printer.

Crucial paper parameters expected to influence the colour rendering of

images were, as previously, the colour, the surface structure and the
absorption. In the tests, attention was given primarily to the influence of
shade and surface structure. The influence of the colour was studied in
terms of the whiteness of the paper and the structure as the surface
roughness.

Papers with different whiteness values as well as different surface

roughness values were used. The different papers were characterized with
respect to optical properties and surface roughness.

63 (95)

Marianne Klaman Royal Institute of Technology 2002

The papers were printed without ICC-profiles and then with individual
profiles for each of the papers. The colour rendering, depending on the
paper whiteness and paper surface roughness, was evaluated by a visual
assessment of images and by colorimetric measurements of colour patches.

ICC-profiles for the different papers used in this study were created and
used cross-wise for each of the paper categories in order to investigate the
tolerance level of acceptance for an incorrect ICC-profile. The method was
also used in order to assess the possibility of grouping paper into categories
and using a single ICC-profile for each category. The cross-wise use of
ICC-profiles meant that each profile was used with the paper that the profile
had been created for as well as for other types of paper.

To study the influence of the paper quality

L*, a* and b* were measured in the test form used for 74 light and
dark tones, and
L*, a* and b* were measured for 17 or 10 tones in a tone scale
between 0 and 100% for cyan, magenta and yellow.

To compare the colour values of the 74 colour patches, they were divided
into 50 light and 24 dark tones. The light tones were defined as tones where
the maximum total amount of one separate colour is 50 %, for a two-colour
combination 60 %, for a three-colour combination 70 % and for a four-
colour combination 90 %. Similarly, the dark tones were defined with the
minimum values of 90, 180, 260 and 320 %. ∆E*ab according to (ISO
12647, 1996) was used to obtain a comparison value for the colour
rendering difference.

For the visual assessment the prints were divided into groups where in each
case, the paper printed with its own profile was the reference against which
the other papers printed with the same profile were compared. The image
was assessed with respect to its identity with the reference. Identical in this
case was defined as having no or only a very small visual difference. The
assessment was made with respect to total image quality.

It was observed that paper roughness and paper whiteness, not

unexpectedly, in some aspects influence the colour rendering differently
when printed in a digital colour printer and in an offset press. The
roughness seemed to have the largest influence on the colour rendering of
the laser prints in the light tones, i.e. when the paper is not covered by toner.
For offset prints, the result was the opposite, i.e. the influence of surface
roughness seemed to be greater when more ink covered the paper surface.

64 (95)

Marianne Klaman Royal Institute of Technology 2002

An increasing difference in whiteness gave the largest difference for the
unprinted paper when it came to the laser printer. The colour rendering,
expressed as ∆E*, gave the smallest difference on the dark colour patches.
In summary, it can be expressed as: the influence of the paper increases as
the amount of ink decreases and decreases when more ink is printed on the
paper. Offset prints produced a similar result, but it was not as pronounced
as it was for the laser prints.

The results also showed that it is possible to use the same ICC-profile for
categories of paper printed in an offset press and to some extent also in a
digital laser printer.

From the tests using different ICC-profiles, it could be concluded that

profiles created for glossy coated paper can be used for other glossy coated
and matte-coated paper with a silk finish. Profiles created for silk paper can
be used for other silk paper and for glossy coated paper. This is true as long
as the whiteness or the surface roughness or both are similar for the
different paper. This is especially accentuated when a profile created for a
paper with lower whiteness value is used on an image for laser printing on a
paper with higher whiteness value. In such cases, it is recommended that a
specific profile be created for each paper.

9.7 Paper VI – Colour Rendering Aspects in Digital

Printing
As a continuation of the studies performed and presented in Papers IV and
V, the study presented in this Paper, VI, was performed with a wider range
of digital printing presses.

Colour control in digital printing is an area not yet as well developed as

colour control or colour management in conventional printing in offset. In
response to the need of a broadened knowledge of the mechanisms behind
colour, and the rendering of colour in digital printing, the study reported
here was conducted. The influence of paper quality and different digital
press techniques on colour rendering and on the colour gamut is the main
issue of this Paper VI.

Colour Management and the establishment of ICC-profiles have progressed

during the last few years. The European Color Initiative (ECI, 1999) has

65 (95)

Marianne Klaman Royal Institute of Technology 2002

drawn up guidelines for how colour data must be saved and processed
device independently. The guidelines give recommendations for the
introduction of efficient ICC-workflows in reproduction technology based
on the ICC-standard. Furthermore, the guidelines give advice on the
automation of device-independent colour processing and how the safe
exchange of ICC-capable colour data can be organized. Improved
workflows have been the target issue in several publications as (McDowell,
2000) and (IFRA, 2000).

A total of 41 different substrates, paper and paperboard, were printed and

tested for aspects of print quality. In order to evaluate the sensitivity of each
printing process to substrate, all levels of surface treatment, from uncoated
to fully coated glossy paper, were included in the study. To achieve optimal
print quality, the substrate was adapted to each printing process as far as
possible, and ICC-profiles were created for every combination of paper and
printing press.

Eight presses were used: two electrographic high quality digital presses, one
liquid toner high quality digital press, one common colour copier and one
large format ink-jet printer. Moreover as a reference, a sheet-fed offset press
and two flexographic high quality presses were used.

The methods used to characterize colour rendering are presented as follows.

Colour gamut coordinates were measured and the gamut volume, V, was
obtained by adding the volumes of tetrahedrons where the co-ordinates were
CMY, RGB and the white and black co-ordinate.
For one tetrahedron the volume is calculated by:

L0 a0 b0 1
L a1 b1 1
V= 1 [Eq. 17]
L2 a2 b2 1
L3 a3 b3 1

To evaluate the reproduction of difficult colours within the gamut certain

colours of the IT8.7 test chart were chosen and measured. These colours
were considered critical to be reproduced or sensitive to influencing
parameters. To calculate the colour difference, ∆E*, for these values on
different substrates, the value of the highest ranked sample in the visual
perception study was used as reference.

66 (95)

Marianne Klaman Royal Institute of Technology 2002

The calculation function included in the profiling software was used for the
calculation of a total ∆E*-value between different groups of ∆E*-values.
These groups represent some of the presses and the total ∆E*-value has
been used to compare the colour rendering between them.

In order to analyse the influence of dot gain in digital presses, a simulation

test was made in one of the electro-photographic presses where the dot gain
values were increased and decreased from the “normal” setting. The L*-,
a*- and b*-values were calculated for the different levels and the ∆E*-value
was calculated with the normal setting as reference.

The influence of certain paper properties such as whiteness and surface

roughness were evaluated according to the investigations made and reported
in Papers IV and V. The relationship between assessed print quality and
assessed colour failure is one of the relationships addressed.

An initial test to analyse dot reproduction and influence on colour rendering

was conducted and an outline for a test model proposed.

For the perceptual evaluation, two images were used, one of which was a
still life “silverware” representing a low-key image with a homogeneous
neutral grey background. The other motif was a portrait of a girl
representing a medium- to high-key image for reproduction of skin tones
and fine details in hair and dress.

Results
The most obvious result concerning the gamut was that for offset, the size
of the gamut increased on highly coated paper and decreased on uncoated
paper. For the electro-photographic presses and the liquid toner press, the
size of the gamut decreased only slightly when changing from coated to
uncoated paper.

Some of the presses gave prints with significantly larger gamuts. However
the correlation between colour gamut (expressed as the volume) and the
corresponding print quality, evaluated perceptually, was very weak. The
relationship within each category of presses was further investigated and
found to be either very weak or at the level of r = 0.70 – 0.80 for offset,
flexography and one of the electro-photographic presses.

Difference in colour rendering of the colour patches in the IT8.7 seems to

be evenly distributed, in the colour space, as in the example, Figure 10,
where a*-values from offset prints and electro-photographic prints are
plotted.
67 (95)

Marianne Klaman Royal Institute of Technology 2002

 80

60

40

20 Offset
a*

0
106 El.press 2
127
148
169
190
211
232
253
274
22
43
64
85
1

-20

-40

-60
Colour patches

Figure 10. The rendering of colours, expressed in a* and b* of prints from one
electro-photographic press and one offset press, is shown for all the colour patches
included in the IT8.7test chart. The print substrate was a medium coated paper.

Concerning difficult colours, the blue colour was in general the most
difficult colour to reproduce. The most dramatic deviation, expressed as
∆E*, was obtained for the uncoated papers printed in offset. For the prints
from the electro-photographic presses the difference between highly coated,
medium- and uncoated papers were much smaller than for offset in
accordance with the results of the gamut. The skin tone was best reproduced
in offset for all paper types and least acceptable from the copier. Next to
blue, orange was the colour most difficult to reproduce.

The comparison of colour rendering differences, expressed as ∆E*, for

different presses and different paper grades gave the result that the colour
deviation between the presses and papers increased from a rather small
value for highly coated paper compared with the reference highly coated
paper to a deviation of about ∆E* = 5 for the uncoated paper. For one of the
electro-photographic presses, this difference is not very obvious regarding
the colour gamut

Prints on three different paper grades, highly coated, medium coated and
uncoated, printed in electro-photograhpic process and offset were compared
with respect to a*-values. Figure 11 –12 show very clearly how dependant

68 (95)

Marianne Klaman Royal Institute of Technology 2002

offset is on substrate and, on the contrary, the independence of the electro-
photographic process.

Electrophotographic press

60
40
20 Medium coated
0 Uncoated
a*

1
37
73
109
145
181
217
253
-20 Highly coated

-40
-60
Colour patches

Figure 11. The a*-values for three different paper grades prnted in an electro-
photographic press.

Offset

60

40

20 Medium coated
0 Highly coated
a*

Uncoated
1
30
59
88
117
146
175
204
233
262

-20

-40

-60
Colour patches

Figure 12. The a*-values for three different paper grades printed in an offset press.

69 (95)

Marianne Klaman Royal Institute of Technology 2002

The best results concerning perceptually assessed print quality were gained
from the conventional processes, with offset still producing the best overall
quality, followed by the flexographic process. The electro-photographic
processes, although performing in the mid-quality range, appeared to
perform equally well on both assessed images and also on all paper grades
from glossy to uncoated. The samples printed in the copier and with liquid
toner showed an overall lower quality.

An almost complete linear relationship between the difference in dot gain

and the achieved colour difference for the test in the electro-photographic
press was found.

According to the studies in Papers IV and V, paper surface roughness, paper

whiteness, measured colour gamut, print quality and assessed colour failure
of skin tones were analysed and compared with each other in order to find
the existing relationship, if any. For all the presses, with the exception of
the electro-photographic, there is a trend that with increasing paper surface
roughness the colour gamut decreases. The electro-photographic presses do
not seem to be at all influenced by the paper roughness.

Concerning paper whiteness, the paper with the highest whiteness values
were also those with the highest roughness values. The paper surface
roughness seems to be the parameter with the highest impact on the colour
gamut in that it seems as if papers with higher whiteness values influence a
decrease of the colour gamut. Due to the fact that they also have the highest
surface roughness values, this factor seems to be the most crucial and gives
the greatest impact of these two on the colour gamut.

The colour difference measure gave the best correlation between a separate
measure and assessed print quality, although not higher than (r = 0.53).

An attempt was also made to categorize and grade the different

“screenings” and dots from these according to the form, edge sharpness and
distinctness of the dot and the impact of dot reproduction on colour
rendering. An outline for a test model linked to dot reproduction and aimed
at being a part of a quality colour test model is proposed.

It should be pointed out that although colour management is used the

differences in colour rendering are sometimes fully visible. The causes are
assumed to be very complex. The processes themselves influence the colour
rendering by the technique, the toners and pigments involved, as do the
screening technique used and the dot reproduction caused by these
70 (95)

Marianne Klaman Royal Institute of Technology 2002

parameters. The dot gain, as has been shown in one test, has a very strong
influence on the colour.

The limits in colour management can certainly be improved by even more

strictly defining the settings of each printing press (density level, dot gain
and other parameters) before creating the ICC-profiles. The algorithms for
profiling can possibly be improved. The result from this study indicates
strongly that in digital printing technology, there are other demands on the
creating of ICC-profiles compared to the traditional printing technology.

Although some techniques seem to have a large colour gamut, this is not a
guarantee for a good printing result. The size of the gamut depends of
course on factors such as ink or toner density, which in some cases may
have been too high, as it seems to have been for the copier. This in turn
means that the inner part of the gamut is influenced with incorrect “dot
gain” as a result.

The following main conclusions were drawn:

Offset print quality on coated paper is still ahead of what is possible in
digital print technology today.
Although ICC-profiles were used, the result showed slightly discernible
to more accentuated differences between prints from the different
presses. This was true both for measured colour data and perceived
colour.
Offset quality is to a large extent dependant on the paper quality,
following, in order, highly coated, medium- and uncoated.
No relationship was found between large colour gamut and print
quality in respect to digital print technologies. However, for offset and
flexography, there was a clear relationship between a large gamut and a
high print quality.
The most critical colour for reproduction was blue. Skin tones were
best reproduced in offset.
Colour gamut increases with decreasing paper surface roughness.
Colour gamut increases with increasing paper whiteness if not the
paper surface roughness increases at the same time, in which case the
roughness dominates and the gamut will instead decrease.
The importance of establishing relevant standard settings for each
process is crucial if the goal is to obtain similarity in colour rendering.
Digital printing, as is strongly indicated, requires routines and standards
for colour management and the creation of ICC-profiles specially
adapted to the printing technique.

71 (95)

Marianne Klaman Royal Institute of Technology 2002

9.8 Appendix - The Comparison of Colour Gamuts
Bristow, J. A., STFI, Klaman, M., GFL, Hoc, M., STFI, The Comparison of
Colour Gamuts, IARIGAI Conference, 1993, Paris, Proceedings Advances
in Printing Science and Technology, vol 22, pgs. 311-332.

This paper contains much of the initial discussions concerning further work
in the colour field of printed media and discusses the fundamentals
concerning colour gamut issues in the world of colour reproduction. It
serves as a good starting point for the papers included in the work involved
in this thesis.

The importance of developing reliable methods to ensure that the degree of

fidelity of reproduction from one stage to another in the graphic arts process
can be satisfactorily assessed, has been stated. The need to make an accurate
assessment of the available gamut arises in different situations, not only
when final prints are compared with each other and with an original, but
also in the comparison with the potential of different sets of inks on a given
paper, of a given set of inks on different papers, the comparison of different
colour copiers and the comparison of different proofing systems.

The desire to achieve a faithful representation of an input target gamut may

be limited by the available colour gamut of the output process. In the
graphic arts field, there are many factors such as dot gain, grey balance
adjustment, trapping, misregister, etc., which will affect the fidelity of a
print to the original and can lead to colour deviation. The colour gamut,
however, is not defined in relation to these factors, even though they can
have a slight influence on the colour gamut itself.

When talking about colour gamuts, the discussion also concerns

boundaries. It is clearly pointed out in the paper the importance of not
forgetting that what a person sees is the ultimate truth when trying to define
the gamut in terms of instrumental measurements. At the same time, only
perceptual gamuts, based on purely psychometric systems, are not enough
when the need for communicating data becomes evident and the
requirement of measures arises. It is important to emphasize that
instruments do not measure the colour, but rather the physical stimulus –
usually the reflected radiation at different wavelengths – that reaches the
human eye and gives rise to the perception of a colour.

It is stated here that not only the colour or the whiteness of paper is of
importance but also the absorption capacity, the structure of the paper and
the interaction between ink and paper. Added to that is the importance of
72 (95)

Marianne Klaman Royal Institute of Technology 2002

illumination and viewing conditions which have to be standardized before
measurements are made.

The secondary colours, red, green and blue, are more susceptible to the ink
balance, the order of printing and similar factors.

Possible ways for the quantitative comparison of gamuts are pointed out in
the paper, and are briefly:

Agreement of the vertices of the hexagon may be determined by

calculation of ∆ E * for each vertex or, which is suggested possibly to
be a better method, both the chromatic difference ∆E ab and the
achromatic difference ∆L * , which for the latter also gives directional
information.
The total deviation ∑ ∆E * can be calculated, and for small values,
good agreement is obtained. There is, however, an uncertainty when the
values are large and the interpretation becomes unclear.
Total gamut areas may be compared.
The areas of the six component triangles within the hexagon may be
compared.
The degree of gamut mismatch may be determined by considering the
degree of overlap.
Determination of the maximum common gamut.

In the paper, the limitation and the risk of only considering the two-
dimensional gamut is discussed.

Concerning the three-dimensional gamut, the following important facts

were shown by experimental findings:
The RGB points are always in a lower plane than the CMY points.
The CMY points do not lie in a horizontal plane, since yellow is always
lighter than cyan and magenta.
Similarly, the RGB points do not lie in a horizontal plane, since blue is
always darker than either red or green.
The last two points indicate that the distance between the two planes is
also of interest.

The black printer, besides giving a better black to prints with the increase of
contrast, etc., also lowers the black apex of the gamut.

73 (95)

Marianne Klaman Royal Institute of Technology 2002

 74 (95)

Marianne Klaman Royal Institute of Technology 2002

10 Discussion and conclusions
During the thesis work from the first paper produced to the last one,
development of new technique meantime naturally changed some of the
intentions stated at the start but also provided new possibilities for the work,
although sometimes in another direction than the one outlined.

The first test model developed (Paper I) was based on the thoughts and
some of the experiences drawn from the work in (Appendix) and of course
earlier work in that field. In parallel with the work performed, the ICC
standard Profile Format was defined and gave new tools for the
achievement of colour fidelity. This logically had an impact on the work of
the thesis and the first action was to investigate the quality and productivity
aspects of colour management systems. (Paper II) was based on
investigation and trials made with different colour management systems
(Klaman, Caretti, 1997). Since variations in the press inking level is one of
the crucial factors in the printing press that certainly will cause changes in
colour rendering, the next step was the investigation of colour shifts caused
by these variations (Paper III) and trying to connect the visual perception
with the measured. To be mentioned, but not reported in this work was an
earlier investigation conducted by the author (Klaman, Anderson, 1992)
concerning the grey balance that showing that considerably small changes
in the inking level caused fairly large deviations in grey balance.

The next three papers (IV-VI) all in a sense deal with the output part of the
reproduction process, the printing press or printer and the influence that the
materials and different parameters have on the colour rendering and
available gamut of the process. The visionary goal here was to try to
separate the different influencing parameters, as substrate, and in the end
have sub-profiles, characterizing colour behaviour of every single
parameter, in order to have the possibility of chaining them together to an
entire profile for a certain application. The different approaches outlined in
the objectives and the conclusions are summarized along these lines.

10.1.1.1 Models developed

The development of a test model for proofing (Paper I), was a matter of
high urgency since the proofing area has always been subject for discussion
whether one proofing system, one technique or approach is superior to
another. The work gave as a result a comprehensive functioning test model
for proofing. What is very clearly pointed out is that reliable methods to
ensure that the degree of fidelity of reproduction from one stage to another
75 (95)

Marianne Klaman Royal Institute of Technology 2002

in the media process can be satisfactorily assessed, have to be complex. The
model to be used must be comprehensive and include different measures.
Comparing it with the comprehensiveness that the human viewing
represents it may well be expected. There is research going on in different
fields to find one simple measure expressing the total quality, but a more
complex model must be considered most plausible not least when colour is
the subject. The development of an appropriate test target to be included in
the proof, or attached to, as reported by (Scmitt, 2001) can be regarded as a
product development based on similar assumptions and findings as those
gained in Paper I.

The test model developed with the crosswise use of ICC-profiles (Paper IV)
for test of the usability potential of the profiles concerning different
substrates and qualities can probably be modified to make the assessment
procedure refined to facilitate an entirely technical measuring method and
with that also a faster method. However, the method proved to differentiate
the tolerance levels of the profiles according to usability. For the industry,
although experienced operators assumingly can have a feeling of how
different profiles can be used also for other substrates than for the one used
for profiling, the results of the systematically conducted evaluation are
beneficial since the trial and error mode can then be excluded.

What is not taken into consideration is the reproducibility of the test

models. This is an important parameter but it has not been possible within
the resources for the projects to analyse it with repeated tests.

An initial attempt was also made to develop a test model according to dot
reproduction. This model is aimed to be part of a quality colour test model
where the screening and the quality of screen dots are considered. The work
is not finished but indeed crucial for further investigation.

10.1.1.2 Theoretical studies

Paper II was presented in the very beginning of the ICC work and viewed
from that horizon and maybe now can be felt dated, but it has to be stressed
that there still is value in it. For the industry it was also at the time
important to be aware of the weaknesses of the standard, since part of it was
not a standard in the full meaning of the word. Only the profiling format
was defined as a standard and rest of the components not. In our tests the
lack of ability of the gamut mapping from large gamuts to smaller ones was
striking.

76 (95)

Marianne Klaman Royal Institute of Technology 2002

The dissimilar and not standardized output targets were another issue as
were the private tags. Concerning the number of colour patches the colour
management area has progressed. For the creation of ICC-profiles the best
choice would be to use the IT8 7/3 target also for output (some vendors use
it). This IT8 target is developed by means of a standardized approach and
consideration has been taken of many of the facts pointed out as a question
in Paper II. Also considered is the distribution of patches in the colour
space, the number is approximated to be relevant and the design is
considered. It was assumed in Paper II that a quality manager would be
necessary to handle the workflow, especially with ICC-profiles. Here
instead the work-flow-management tools and automation has been in focus
for development and a control system of this kind will then be suitable for
the handling of ICC-profiles too. Since the paper was presented gamut
mapping algorithms have developed. The specific problem mentioned in
Paper II regarding mapping from a wide colour gamut to a small one, where
compression of the gamut as well as distortion is needed, has been partly
solved, due to improvements in gamut mapping methods.

10.1.1.3 Colour difference metrics and acceptability of colour differences

The question of what colour difference metrics to be used and how they
correlate with the human vision and their ability to fulfil the requirements,
as a tool for comparison of different images, is important. In the studies
carried out small differences with the magnitude of 0.5-0.95∆E were
recognized (Paper III). In Paper IV was shown a clear breakpoint at ∆E>2
above which a majority of the panel considered the images to be different.
The first comparison was with colour patches and the second one with
images. (Has and Newman, 1995) discussed whether reported values of
visible differences of 1.0 ∆E versus other reported values of 2.5 were the
limit values and why they differed. The assumption that colour patches and
images subtend different fields of vision may be an explanation to the
difference achieved in these two studies. Further studies must however be
performed to state the assumptions. There are also values reported as ∆E of
1.5 as the limit of visible differences (Nordström, 2000).

In research work by (Berns, 1999) is shown that the CIE94 formula implies
that differences in colour gamut between multimedia devices may not be as
large as it appears when looking at CIELAB gamut volumes. In a
comparison of two gamuts, one for a display and the other one for a printer,
Berns obtains a noticeable difference in gamut volume. He claims that the
large differences in Euclidean distance are in fact less noticeable visually.
Berns also reports the “considerable interest in the colour difference

77 (95)

Marianne Klaman Royal Institute of Technology 2002

research community to develop a new colour space that is corrected for this
chroma position dependency” (Berns, 1999, p. 121).

(Johnson and Green, 2001) performed experimental work to evaluate the

magnitude of colour difference formulas that may be accepted for critical
comparison in media issues. They refer to DE94 by which the main
uniformity error is corrected, namely that pairs of colours of high chroma
produce far larger numerical differences than their appearance difference
justifies. In their research, however, they claim that neither CIE94 nor
CIE2000 offer any significant improvement in producing a uniform visual
scale over CIELAB.

This is partly contradictory to the outstanding performance reported by

(Luo, Cui, Rigg, 2001) concerning the new CIE2000 formula.

∗
In this thesis work mainly CIELAB and ∆E ab have been used together
with the calculations of gamut volumes and gamut areas. Combinations of
these measures have proved useful. The CMC has also been investigated in
∗
one case, but no advantage compared with ∆E ab could, however not, be
concluded from the study. Further research work is clearly needed with
focus on CIE2000 for media applications, especially printed media.

There is a distinction in what degree deviations from a reference are

detected as a difference and what difference is accepted. (Johnson and
Green, 1999) address the fact that relatively few studies have investigated
acceptability in colour difference. They noticed in a study that the
acceptability limit was about two or three times that for perceptibility. This
seems to correspond well with the results reported in Paper III. Concerning
their results of a mean perceptibility tolerance of 3∆E compared with 6∆E
for acceptability the opinion is rather that, due the results shown in Paper
III, these figures have to be more differentiated due to the various
sensitiveness according to colorant. The yellow colorant was much more
sensitive than the cyan. 6∆E as an acceptability limit sounds as too high a
value, at least if the variations due to colorant should be taken into
consideration.

10.1.1.4 Colour matching

It is fundamental to have reliable methods for comparison of the deviations
of colour that occur in a process, but also to have control in the
establishment of the ICC-profiles in order to be able to achieve colour
fidelity or colour matching. In the study with digital printing technique

78 (95)

Marianne Klaman Royal Institute of Technology 2002

(Paper VI) the process was optimized with appropriate paper for each
different technique, a profile for each paper and settings for each printing
press optimized. Still there are obvious colour differences. The physical
limitations cannot be ignored. But the establishment of a stable, repeatable
and predictable process is most important.
The achievement of colour matching by means of the method presented by
(Shimamura et al., 2001) with the one-ink-zone profiling target and a one-
ink-zone IT8.7/3 worked well as did the use of the CRF curve. Still visual
assessment of targets showed noticeable colour difference between proof
and print. According to the acceptability limit the value of ∆E ≈ 5, or lower,
is suggested for the difference between print and proof (Klaman, 2000).
Between them the tolerance level must be allowed to be wider than the
tolerance limit within a printing session.

(Collet, Chagas and Baudin, 1999) obtained the result in their study, which
is interesting, that colour fidelity seemed to be more difficult to obtain by
stochastic screening than by a conventional one. A possible reason among
others, might be that the control of the small dots is not sufficient and, if
not, will cause larger colour fluctuations than the conventional process.

In the attempts to achieve a stable process for the creation of an ICC-profile

for a printing press it is central to design the printing trial in such a way that
the density, the dot gain and, according to that, the inking level is stable and
even so across the print sheet or web. (Nordström, 1999) designed a test
form with a “built-in-design” to facilitate the printing trial.

10.1.1.5 Colour gamut

Concerning the importance of the size of the colour gamut it has to be
stressed that sometimes it is equal to higher print quality but as shown in
Paper VI this can also be very misleading. The prints from the copier had
large gamuts but a very low print quality. As is hinted in the paper this is
probably due to the fact that the density setting and the density level have to
be adjusted to exactly the same level as for the print. This is also in
accordance with the results of the test model for proofing (Paper I). The
density, certainly, is not the only factor but together with other quality
parameters it strongly affects the gamut size. Besides that, at the same time
it will have a strong impact on the overall print quality, since with
increasing density the middle tones, too, are affected. The skin tones were
best re-produced in offset for all paper types and least acceptable for the
copier. Since skin tones are generally considered to be the most difficult to
reproduce, this result probably shows that offset still is the process where
the possibilities to optimize and render these tones in a qualified way are the
best. Consequently, digital techniques do not always seem adapted to the
79 (95)

Marianne Klaman Royal Institute of Technology 2002

demands of every aspect of colour reproduction. The transition of tones is
not smooth enough.

In Paper VI the difference in size of the colour gamut between offset and
electro-photographic presses is shown. (Nordström et al., 2001) calculated
the difference between waterless offset and digital laser print with a
decrease for the latter one to be 24%.
The question is how much the gamut of the laser or electro-photographic
press can be increased with help of, for instance, papers developed very
specially for digital printing. We have seen that the paper grade in itself
does not influence the size of the gamut very much but perhaps the
improvement of other parameters, press or paper or others, can increase the
colour gamut, it will say if it is needed? It is of importance to have in mind
that the digital printing techniques basically compete in other fields than the
quality, although the quality is important. Factors such as possibility of
personification, customisation, distributed media, and print-on-demand are
the characteristics of the digital printing techniques. The analysis of to what
degree the quality demand versus the market demands, are to be prioritized
are of vital interest.

The approach to compare colour gamuts by use of total ∆E-values (the

mean value of all the ∆Es of two gamuts compared) seems to be a feasible
way. However, this method has its limitations, since there is a risk that large
deviations and very small ones will cancel each other out and therefore not
characterize the deviations in the different parts of the gamuts. But it has a
value as a simple tool to compare colour gamuts and also to have a measure
that shows the trend and magnitude of differences. Since it is often built into
software for colour management it offers a handy instrument for operators
and others in need of conducting comparisons of available colours and
gamuts.

Similar to the comparison of gamuts is the comparison of complex images

and (Chung et al., 2001) claim that there is no easy way to assess colour
difference quantitatively between two complex images. In their study they
used a standard press characterization target (IT8.7/3 basic block) to
quantify pictorial colour image difference by colorimetry. The paper shows
a way to use the cumulative relative frequency (CRF) of the ∆E for the
comparison. Accordingly, the approach with comparison of total ∆E-values
can be an alternative.

80 (95)

Marianne Klaman Royal Institute of Technology 2002

10.1.1.6 Methods for characterization
Methods for characterization of crucial parameters in the printing stage
influencing colour is one of the main objectives in the thesis. Parameters of
the substrate influencing the colour have been one of the main subjects.
Different press techniques and some press parameters are the others.

The absorption of the substrate is not measured, only visualized in the thesis
work as ESEM (Environmental Scanning Electron Microscope) and light
microscopical images, but these images show in a very illustrative way that
the ink is distributed both in the plane and in the vertical direction. It would
be possible to use such kind of information and to establish measures linked
to the images. In relation to the absorption findings the attempt made (Paper
VI) to categorize and grade the different “screenings” and dots can be
linked together with absorption assessment. Regarding the form, edge
sharpness and distinctness of the dot and the absorption factors the impact
of dot reproduction on colour rendering can probably be even better
characterized. The method proposed by (Baudin and Rousset, 2001) might
be a suitable one to use for the investigation of relationship between
absorption mechanisms due to ink penetration and colour rendering.

Concerning the absorption it is stated among others by (Field, 1988, p. 86)

“from the color reproduction viewpoint, it is desirable to have substrates
with low absorbency, hence resulting in minimal degradation of the printed
ink film color. However, other printability concerns such as drying and ink
transfer ultimately impose the minimum absorbency requirements”. And
further (p. 85) that “the absorbency of paper has been shown to cause shifts
in the color of the printed ink film”.

All the results concerning absorption and the influence of paper surface
roughness and whiteness (Papers IV-VI) demonstrate the utility of these for
characterization of paper, from the viewpoint of their influence of colour
rendering. There certainly are differences between conventional printing
and digital and the characterization might be differentiated, or better, a
general comprehensive method, including the differences, can be used for
all printing methods.

Also (Lindqvist, Heilmann, 1999) discuss the problems of subjective tests

being the best method for comprehensive evaluation of print quality (in
their case not colour), but being slow and expensive to execute. This is also
the case in the evaluation of colour fidelity. It is easy to assess visually and
clearly determine the colour fidelity but with technical measurement we
need a group of measures to express the complexity of colour. At the same
time the perception trials are very time-consuming and impractical.
81 (95)

Marianne Klaman Royal Institute of Technology 2002

But according to paper characterization (Lindqvist and Heilmann, 1999),
found, which is important, that paper dependence could probably be
measured by technical means and that visual tests are then not needed for
this purpose. If so, it would be an efficient tool in the investigation of colour
printers and presses as well as in the product development of digital printing
papers.

The relative independence of substrate quality in terms of coated - uncoated,

related to colour gamut, of the digital printing technology, gives the
opportunity to design paper qualities for the digital printing technique, with
focus on developing maybe other and more vital parameters. Since the
coating in itself does not seem to improve the printed colour quality in any
higher degree, there seems to be a potential improving other factors, or it
might perhaps be possible to design the coating or surface otherwise
instead. The independence of substrate of the colour gamut for digital
printing and the dependence for offset was also shown earlier by (Lie et al.,
1999).

Concerning the influence of whiteness and the eventual change of colour

due to the optical brightening agent (OBA, or as it is also called, fluorescent
whitening agents FWA) this is not at all addressed in the investigations
performed. The assumed effect is dependent on the kind of illumination
used. However, this influence seems to be limited (Rydefalk, 1998). He
shows that cyan and magenta, which are assumed to be able to use the light
emission from the OBA in the blue range of light, do not so to any greater
∗
extent. Observed ∆E ab -values seem to depend on the unprinted paper
between the printed dots and Rydefalk makes the interpretation that the
printing ink absorbs most of the UV-radiation with the consequence of only
minor light emission under the printed areas. This can, however, be the
subject for future work concerning colour rendering and paper whiteness.

In the last papers (IV, V and VI)is stated that the need of standardization
for the digital printing processes is striking. Only by optimizing each
process there is no possibility to gain a complete colour fidelity or colour
matching between separate processes or separate devices. There have to be
set up standards for the different digital printing techniques in the same way
as has been done for offset. This was already indicated in the first paper
where the issue was proofing. Proofing devices, however, are basically built
on the same techniques that the digital presses, and the need to target all
process values is the same. In offset the standardization work has

82 (95)

Marianne Klaman Royal Institute of Technology 2002

progressed during many years and certainly the same has to be worked out
for the different digital printing technologies.

10.1.1.7 Relationship between technical measures and visual assessment

Paper III gave a good agreement between assessed colour deviation and
measured. Also when an image had a colour shift measured with a negative
a*-value (more blue), the visual assessment was also defined more bluish.
In the work performed, measured parameters have been found correlating
with perceptionally assessed parameters. Important to notice is the impact
of the image type on the correlation. Skin tones are often sensitive to a
much higher extent and give also a better correlation than saturated colours.

10.1.1.8 Relationship between quality parameters and colour

The results show the importance of the stable, repeatable and predictable
process not only for the total procedure but also for every single step
included. Starting with the printing, and the printing press itself, the control
of the inking level, is of course fundamental concerning printed media. This
is already known, but unexpected is perhaps that the relatively small
variations that arise in a controlled level influence the colour rendering
visibly. The uncontrolled changes of density were of the magnitude 0.06 to
0.08 from the target value and rendered a colour deviation, expressed in
∗
∆E ab , between 0.5 and 0.95. These small changes in colour deviation were
estimated by 40% of the panel. Since the standards for tolerance of density
are often considerably larger, the actual deviations will thus be larger and
clearly visible. Added to that, the result obtained with the influences of the
process inks varying in magnitude depending on which ink is fluctuating, it
seems as if the tolerance limits ought to be more tightened. The result that
deviations in yellow gives numerically larger impact on colour shifts, is of
importance and should be further investigated in order to verify the standard
values in press control if needed.

In the light of this the results obtained by (Nurmi and Sivonen, 2001) are
interesting, although from another point of view; that the yellow tones are
more sensitive (in their case to print on the reverse side) to changes of the
optimal level. They also concluded that after yellow it was the red tones and
then blue and green ones that were most insensitive.

(Green and Luo, 2001) found in their work with the algorithms for gamut
mapping that hue shifts in yellow were somewhat sensitive and close to a
new boundary or name).

83 (95)

Marianne Klaman Royal Institute of Technology 2002

This sensitivity, declared in different cases, can be a reason for continued
analysis of the yellow colorant more in detail. Also from the view that the
yellow colorant, although its importance for correct rendering of skin tones
(the varied effect due to type of image is also essential) is always pointed
out, often is considered of lower importance since it is paler and regarded as
having smaller influence.

Conclusions to be drawn from this are the need for and importance of
relevant measuring during printing to avoid fluctuations in the inking level.
According to the findings of reverse side printing and the yellow colorant,
the layout of the printed product is obviously crucial. Further research
seems essential to investigate the sensitivity of the yellow, together with the
other pigments and inks.

10.1.1.9 Some of the main findings

Some of the main findings are listed below.

To be able to analyse colour in a way that is relevant for the printing

process including proofing, rather comprehensive test models have to
be used.
Relationship between technical measures and perceptional assessment
are found but according to the first statement, also here, several
parameters are mostly needed.
∗
For the measurement of colour differences in printed media the ∆E ab
has proved useful and the CMC did not show any advantage compared
∗
with the ∆E ab . In some applications the total dE-value proved reliable.
For the characterization of the available gamut the simplified
calculation of areas within the gamut proved to be feasible for use in
comparisons of gamuts.
Even small variations in press parameters will influence colour
considerably.
The different colorants, i.e. process inks, have different properties with
a certain change in density unit giving different dE-values.
There is, according to the previous statement, a need for tightened
tolerances.
It is possible to use standard parameters for substrates to characterize
the substrate concerning colour rendering. These parameters are mainly
paper surface roughness, paper whiteness and absorption.
It is possible to create ICC-profiles for a group of paper grades, but also
other factors have considerable influence and for digital print
technology these factors are maybe more important and must be

84 (95)

Marianne Klaman Royal Institute of Technology 2002

 considered at the creation of ICC-profiles. Also the motif of the images
had a strong impact on the possibility to use “incorrect” profiles.
The relationship between deviation in dot gain and deviation in colour
is almost linear.
The crosswise model for evaluating profiles is a suitable tool.
Improved standards for digital printing techniques are needed. They are
supposed to differ from those for offset. Probably specific routines and
special algorithms are needed for the different digital printing
techniques.
The establishment of a stable, repeatable and predictable process is the
basic demand for the achievement of a colour-controlled process.

10.1.1.10 Suggested Future Work

The findings achieved have lead to the following possible areas or subjects
of future research:
A deeper analysis of the yellow colorant and its sensitivity in different
ways to changes.
Further analysis of which colour different metrics are the most suitable
for printed media according to investigation of the CIE2000.
Analysis of the importance of OBA in the characterization of print
substrate in terms of colour quality.
Improvement of the crosswise tool with a stronger connection to
technical evaluation only.
A comparison of the use of total ∆E-difference and CRF calculation
between complex images.
Improvement of the characterisation methods for substrate
characterisation.
Further development of a dot reproduction quality test model.
Development of standards for digital printing techniques.

85 (95)

Marianne Klaman Royal Institute of Technology 2002

 86 (95)

Marianne Klaman Royal Institute of Technology 2002

11 Acknowledgements
In a research work of this kind there are many persons involved and many
to remember. First of all, however, I would like to thank professor Nils
Enlund, who became my supervisor and who has always been encouraging
and supporting. You inspired me.

Further, special thanks to Gunnar Rosin, former manager of the Institute of

Media Technology. You initiated the thought to continue my research work.
Thank you for your always brisk and friendly attitude.

I am also obliged to colleagues and friends, within my institute as well as

outside it. Thanks for all cooperation and for good events that we have
shared. It has been a long journey within the media field and some of you I
have known for a very long time. Thanks to all of you, nobody mentioned,
and none forgotten. Special thanks, though, to Ingrid Anderson for
cooperation through the years and for performing experimental work in
many projects.

Part of my work was carried out within the PFT research programme and
part of it is subsidized by T2F. I am most grateful for collegial and financial
support.

And warmest thanks to my family, beloved Torsten, Josefin and Gustav, all
of you supporting and encouraging me (and sometimes not disturbing me),
and Torsten also for persistent and inventive checking of my manuscript.
Thank you my dear mother, for always supporting me, and being there. And
not least my warm thoughts to our dearest Lucas, accompanying me and
working with me. Thank you for sometimes strolling over the keyboard,
with a demanding miaow, when you thought I needed a break (and when
you needed a rest it was on my books).

Marianne Klaman, Stockholm, April 2002.

87 (95)

Marianne Klaman Royal Institute of Technology 2002

 88 (95)

Marianne Klaman Royal Institute of Technology 2002

12 References
Bassemir, R., Costello, G., DiBernardo, A., DiPiazza, J., Kuna, D., Puilius, K.,
Rybny, C. and Zawacki, W., A Comparison of Visual and Spectrophotometric
Evaluations of paired color prints, TAGA Proceedings p. 558-578, 1995.
Bassemir, R., Visual and Spectrophotometric Evaluations of Paired glossy prints
having selected ∆E* values, TAGA Proceedings. 1995
Baudin, G., Rousset, E., Ink-jet Printing: Effect of Paper Properties on Print
Quality, Proceedings of IS&T’s NIP 17: International Conference on
Digital Printing Technologies, 2001
Berns, R., S., Challenges for Colour Science in Multimedia Imaging,
Colour Imaging – Vision and Technology, John Wiley and Sons LTD ed. by
MacDonald, L. W., Luo, M. R., p 119, 1999.
Bolanca, S., Mrvac, N., Reproduction Objectivity of Four Color and Seven
Color Conventional Printing and Digital Offset Printing Technique, 28th
Research Conference of IARIGAI, September 2001.
Bristow, J.A., Johansson, P-Å, Subjective evaluation by pair comparison:
Pitfalls to avoid and suggestions for the presentation of results, Advances in
Printing Science and Technology, Proceedings of the 17th International
Conference of Printing Research Institutes, Saltsjöbaden Sweden, Pentech
press, London, June 1983.
Bruno, M., Principles of Color Proofing – A Manual on the measurement
and Control of Tone and Color Reproduction, GAMA Communications,
1986.
Chan, C.J., Chung, R., Cheung, W., Digital Proofing with ICC CMS, TAGA
Proceedings 2000, p 143-159.
Chung, B., Color Management Systems, A Panel Discussion, TAGA
Proceedings 1999, p 623-628.
Chung, R., Shimamura, Y., Quantative Analysis of Pictorial Color Image
Difference, TAGA Proceedings 2001, p 333-345.
Chung, R., Shimanura, Y., Conducting a Press Run Analysis, Proceedings
of the 28th Research Conference of IARIGAI, September 2001.
CIE (79) (Commission International de l’Eclairage (1995) Industrial Color-
Difference Evaluation, CIE Publication 1162. Vienna: CIE
CIE Publication 142-2001.

89 (95)

Marianne Klaman Royal Institute of Technology 2002

CIE, Industrial Colour Difference Evaluation, CIE Tech. Rep. 16, Vienna,
1995
Clark, F. J. J., MacDonald, R., Rigg, B., Modification to the CMC colour-difference
formula, Text. Chem. Color, 1988. 20(6), p 128-132. 1988
Clarke, F.J. J., McDonald, R., Rigg, B., Modification of the JPC colour
difference formula, J. Soc. Dyers Colourists 100, 128-132, 1984.
Clippeleer, J. De, Device Independent Color Reproduction, TAGA
Proceedings 1993.
Collet, S., Chagas, L., Baudin, G., Colour Management: A User’s Point of
View, Advances in Printing Science and Technology, Proceedings of the
21st Research Conference of the International Association for the Graphic
Arts Industry, Munich, September 1999.
Eidenvall, L., Eriksson, L., Norberg, O., Pauler, N., Klaman, M., Westin, P.,
Fahlcrantz, C-M., Johansson, P-Å., Lindberg, S., Digital Print Evaluation
2000/2001, Framkom Report 2001:06, 2001
Fairchild, M. D., A Revision of CIECAM97s for Practical Applications,
Color Research and Application, Volume 26, Number 6, December 2001, p.
418-491.
Fairchild, M. D., Color Appearance Models, Addison Wesley Longman
1998.
Field, Gary G. Color and Its Reproduction, Graphic Arts Technical
Foundation, 1988.
Gast, G., Tse, M-K, A Report on a Subjective Print Quality Survey
Conducted at NIP16, Proceedings of IS&T’s NIP 17: International
Conference on Digital Printing Technologies, 2001.
Giorgianni, E. J., Madden, T. E., Digital Color Management, Addison-
Wesley Longman, 1998.
Green, P., Defining colour gamut boundaries with a test target, Colour
Imaging Group, December 2000.
Green, P., Luo, R., The development of a new gamut mapping algorithm for
newsprint, 28th Research Conference of IARIGAI, September 2001.
Gustavsson, S. Dot Gain in Colour Halftones, Doctoral thesis no 492,
Linköping Studies in Science and Technology, 1997.
Guyton, A. C., Textbook of Medical Physiology, 7th Edition, W. B.Saunders
Company, Philadelpia, 1986

90 (95)

Marianne Klaman Royal Institute of Technology 2002

Has, M., Newman, T., Color Management: Current Practice and the
Adoption of a New Standard, Advances in Printing Science and
Technology, Proceedings of the 23th Research Conference of IARIGAI,
September 1995. (Joint conference with TAGA).
Heilmann, J., Lindqvist, U., Significance of Paper Properties on Print
Quality in Continuous Ink Jet Printing, Journal of Imaging Science and
Technology, Volume 44, Number 6, Nov/Dec 2000.
Heilmann, J., Lindqvist, U., The effect of the Drop Size on the Print Quality
in CIJ Printing, IS&Ts NIP15: 1999 International Conference on Digital
Printing Technologies, 1999.
Hulsman, W. J., Notermans, J. T. E. C., Color Management Requirements
in Contract Proofing, IS&Ts NIP16:2000 International on
Digital Printing Technologies, 2000.
Hunt, R.W.G., Measuring Colour, 2nd edition ed. 1991, John Wiley & Sons,
1987.
ICC.1:2001-12, File Format for Color Profiles (version 4.0.0),
www.color.org
IFRA: Winsløw, S., (The Graphic Arts Institute of Denmark), Colour
Management Workflow Analysis, , Ifra Report, 2000.
International Color Consortium, ICC Profile Format Specification, version
3.3 November, 1996
ISO/DIS 12640, Graphic technology -Prepress digital data exchange- Standard
colour image data (SCID).
Jarnvall, Torsten, Private talk, 2002.
Johnson, T., Green, P., The CIE2000 colour difference formula and its
performance with a graphic arts data set”, 28th Research Conference of
IARIGAI, September 2001.
Johnson, T., An effective Colour Management Architecture for Graphic
Arts, TAGA Proceedings 2000, p 88-111.
Johnson, T., Device Independent Colour … is it Real?, TAGA Proceedings
1992.
Johnson, T., Green, P., Issues of Colour Measurement and Assessment,
TAGA Proceedings 1999, p 518-534, 1999.

91 (95)

Marianne Klaman Royal Institute of Technology 2002

Juntunen, S., Jäntti, M., Lindqvist, U., Colour Imaging in various Printing
Process-Modelling and Simulation in a PC-Environment, Advances in
Printing Science and Technology, Proceedings of the 20th Research
Conference of the International Association of Research Institutes for the
Graphic Arts Industry, September 1989.
Klaman, M., Anderson, I., Densitetsförändringars inverkan på gråbalansen,
GFL Projektrapport nr 75, Juli 1992.
Klaman, M., Caretti, M., ”Färghanteringssystem – Metodik och teknik för
ökad bildkvalitet”, IMT Projektrapport nr 87, Januari 1997.
Klaman, M., Hard, Soft and Remote Proofing, TAGA Proceedings 2000.
Kohler, T., ICC Achievements and Challenges, Paper presented on the joint
TAGA – ISCC Conference 1999.
Lie, C., Short-run Printing – Influence of Paper on Print Quality, Adv. In
Printing Science and Technology – Adv. In Digital Printing, Proceedings of
the 26th Research Conference of the International Association of the
Research Institutes for the Printing, Information and Communication
industries, 1999.
Lind, J.T., Adams, R.M., Advances in colour management”, 28th Research
Conference of IARIGAI, September 2001.
Lindqvist, U., Heilmann, J., The Paper Dependence of the Print Quality in
Drop-on-demand Ink-jet Printing, Advances in Printing Science and
Technology – Advances in Digital Printing, Proceedings of the 26th
Research Conference of the International Association of the Research
Institutes for the Printing, Information and communication Industries,
September 1999.
Lindqvist, U., The Restructured Media Field Sets New Demands on
Technical Research, Graphic Arts in Finland, 29(2000)2, p 3-5.
Lou, M. R., Cui, G., The Development of the CIE 2000 Colour-Difference
Formula: CIEDE2000, Color Research and Application, Volume 26,
Number 5, October 2001.
MacDonald, L. W., Luo, M. R (edited by), Colour Imaging – Vision and
Technology, John Wiley and Sons LTD, 1999.
Mc Dowell, D., Color Management; What’s needed for Printing &
Publishing, IPA Standards Committee, March/April 2000

92 (95)

Marianne Klaman Royal Institute of Technology 2002

McDowell, D., Graphic Arts Standards – A Current Status Report,
Advances in Printing Science and Technology, Volume 25, - Advances in
Offset Printing, Proceedings of the 25th Research Conference of IARIGAI,
September 1998.
McLaren, K., The colour science of dyes and pigments, 2nd ed. 1986,
Bristol: Adam Hilger. 209 p. 1986.
Morovic, J., Luo, R., The Fundamentals of Gamut Mapping: A survey,
Journal of Imaging Science and Technology, May/June 2001.
Morovic, J., Luo, R., b, Evaluating Gamut Mapping Algorithms for
Universal Applicability, Color Research and Application, Volume 26,
Number 1, February 2001.
Murray, A., Monochrome reproduction in photoengraving, J. Franklin Inst.,
1936.
Neugebauer, H. E. J., Die theoretischen Grundlagen des
Mehrfarbendruckes, Z. Wissenschaftlisches Photograph., 1937.
Norberg, O., Westin, P., Lindberg, S., Klaman, M., Eidenvall, L., A
Comparison of Print Quality between Digital and Traditional Technologies,
DPP 2001 Eye on the Future, Proceedings from The International
Conference on Digital Production Printing and Industrial Applications,
2001.
Nordström, J.-E., STFI-JEPTM , STFI-Justified and Enhanced Print, Patent
Application STFI, 1999.
Nordström, J-E. P., Color shifts in waterless offset, - trapping or dot gain in
halftones?, TAGA Proceedings 2000, p 218-245, 2000.
Nordström, J-E, Karlsson, S., Ståhlberg, J., Defining the size of a four-
colour ink sequence colour gamut: - waterless offset compared to toner
transfer printing process, 28th Research Conference of IARIGAI, September
2001.
Nurmi, O., Sivonen, J., The Effect on the Colour Gamut of Newspaper
Printing on the Reverse side, and how this effect changes with time,
Proceedings from the 28th International Conference of IARIGAI – Advances
in Colour Reproduction, 2001.
Paul, A., Colour fluctuations in printing, J. Prepress & Printing Tech. 2 p.
18-28, 1998.
Rydefalk, S., Inverkan av optiskt vitmedel på färgtryck – inledande studie,
PFT-report no 44, 1998.

93 (95)

Marianne Klaman Royal Institute of Technology 2002

Rydefalk, S., Wedin, M., Literature Review on the Colour Gamut in the
Printing Process – Fundamentals, PFT-rapport nr 32, 1997.
Schildgen, T. E., Pocket Guide to Color with Digital Applications, Delmar
Publishers, 1998.
Schläpfer, K., Classification of colour gamuts of printing processes, TAGA
Proceedings 2000, p 112-129.
Schläpfer, K., Widmer, E., How to test a Colour Management System,
TAGA Proceedings 1996, p. 253 - 266.
Schultz, U., Density Measurement and Colour Measurement – Advantages
and Disadvantages in the Printing Industry, Advances in Printing Science
and Technology, Proceedings of the 19th International Conference of
Printing Research Institutes, June 1987.
Shimamura, Y., Chung, R., Sigg, F., Further Study of ICC-based Digital
Proofing, TAGA Proceedings 2001, p 381-398.
Siljander, R.A., Fisch R.S., Accuracy and Precision in Color
Characherization, TAGA Proceedings 2001, p. 57-79.
Sørensen, P.H., Jørgensen, H.O., Colorimetric Tolerances on Packaging for
Medicine, Advances in Printing Science and Technology, Proceedings of
the 22nd Research Conference of the International Association of Research
Institutes for the Graphic Arts Industry, September 1993.
Steiger, W. F., Device-independent Colors in Office Technology and in the
Graphic Arts Industry, Advances in Printing Science and Technology,
Proceedings of the 22nd Research Conference of the International
Association of Research Institutes for the Graphic Arts Industry, September
1993.
Sueeprasan, S., Lou, M.R., Rhodes, P.A., Investigation of Colour
Appearance Models for Illumination Changes Across Media, Color
Research and Application, Volume 26, Number 6, December 2001.
U. Schmitt, U., Do Colour Control Devices Monitor the Whole Gamut?,
28th Research Conference of IARIGAI, September 2001
Wågberg, P., Lindberg, S., Klaman, M, Johansson, P-Å, Report from the
International Workshop on Print Quality, June 1997, PFT-report no 35,
1998.

94 (95)

Marianne Klaman Royal Institute of Technology 2002

Wilkinson, A., Colorimetric Evaluation of the Match between Pre-proofs
and Gravure Prints, Advances in Printing Science and Technology,
Proceedings of the 18th International Conference of Printing Research
Institutes, 1985.
Wyszecki & Stiles, Color Science – Concepts and Methods, Quantitative
Data and Formulae, first edition 1967, second 1982, Wiley & Sons, 1982
Xu, L., Levkowitz, H., Uniform Color Appearance Model, TAGA
Proceedings 1996, p222-234.
Yang, L., Kruse, B., Chromatic Variation and Color Gamut Reduction due
to Ink Penetration, TAGA Proceedings 2001, p 399-407.
Yang, L., Kruse, B., Scattering and Absorption of Light in turbid Media,
Advances in Printing Science and Technology – Advances in Digital
Printing, Proceedings of the 26th Research Conference of the International
Association of the Research Institutes for the Printing, Information and
communication Industries, September 1999.
Yule, J. A. C. ”Principles of colour reproduction”, John Wiley and Sons,
Inc., 1967.
Yule, J.A.C., Nielsen, W.J., The penetration of light into paper and its effect
on halftone reproduction, TAGA Proceedings, 1951.
Zawacki, W. F., Steele, L. C., Color Characterizarion of Various Ink,
Substrate and Printing Process Combinations, TAGA Proceedings 1999, p
135-147, 1999.

12.1 Links
http://www.adobe.com/support/techguides/color/colormanagement/

http://www.color.org/

http://www.creoscitex.com/products/color/profile_wizard/

http://www.srgb.com/

95 (95)

Marianne Klaman Royal Institute of Technology 2002