File: tubtiv_gene.htm
1. Title and Author
Historic developments of colour-image technology between 1998 and 2024
included in the TUB text_image_vocabulary (tubtiv)
for colour education and standardization
Prof. Dr. Klaus Richter, Technical University Berlin (TUB),
Department of Lighting Technology, Einsteinufer 19, D-10587 Berlin
email: klaus.richter@mac.com
2. Content and search text within the tubtiv
This vocabulary of the Technical University of Berlin (TUB)
combines text (three lines) with colour images.
Therefore any search text within the three lines
is connected by links to a special colour image.
This colour image (or a colour table)
of the size A4 may consist of 16 sub images
The same search text usually appears in many
other of the about 36.000 lines of 6.000 pdf-colour images.
For example there are 283 hits for the search text: hue circle.
3. File names for different Sections
Meaning of different letters and numbers in a file name tubtiv_ahe-2401:
tub: Technical University Berlin
tiv: Text image vocabulary
a: Section: a (1998-2024), 0 (1998-2010), 1 (2011-2018), 2 (2019-2024)
e: Language: e (English), german g (German)
2401: Version Date (Year & Month)
htm: File name extension .htm, .txt.
Note: the file name without version date
(tubtiv_ahe instead of tubtiv_ahe_2401)
always contains the current version.
4. Start or new Start (e. g. after use of too many links)
For a (new) start go to one of the two TUB websites
(a new window will open)
http://farbe.li.tu-berlin.de/index.html or
http://color.li.tu-berlin.de/index.html.
All the following htm links then work as relative links.
tubtiv_ahe.htm in e (English) or
tubtiv_ahg.htm in g (German).
5. Creation, content and application of pdf files
The TUB colour image series are based on PostScript vector graphics.
All files in the programming language PostScript
(File extension; .eps or .ps) have been converted by the software
Windows Adobe Acrobat Distiller 3.1 (2005) into pdf files
(File extension; .pdf), and without metadata.
Therefore according to IEC 61966-2-1:1999 (sRGB colour space),
the default standard output will be produced on any colour device
without metadata in the pdf file.
All pdf files therefore use the defined PostScript operators,
for example, for the conversion between rgb*- and cmyk*- data,
and for converting the gamma values.
5.1 Languages
The colour image series serve as the basis for colour education in the
fields of colourimetry, colour image technology, layout, design and art.
Many images are available with text in six languages: English (E),
German (G), French (F), Spanish (S), Italian (I) and Norwegian (N),
see for example color/index.html.
This page shows links to a paper of Klaus Richter with the title
Colour, Colour Vision, and Elementary Colours in Colour Image
Technology.
This paper is also available on the ISO standard Maintenance Server
in English, German and French. For six languages the files of this
paper with many colour figures are optimized for the display, offset,
and printer output.
5.2 Standardization
The colour-image series for colour-image reproduction have been
designed for colour education and standardization. Many proposed BAM
and TUB images have been included in documents of the standard
organizations ISO, IEC, CIE, CEN, and DIN, see in a new window
http://standards.iso.org/iso/9241/306/ed-2/index.html or
http://standards.iso.org/iso-iec/15775/ed-2/en.
5.3 Publications
Many examples of applications of colour images in colour technology,
design and art are published, between 2010 and 2024, see
XY91FEN.html or between 2000 and 2009, see
A/XY91AE.html.
Other known vocabularies of the standard organisations ISO, IEC, CIE, and
DIN usually combine usually only text information.
However, the tubtiv combines the textual information with
the visual colour information.
The tubtiv contains important developments in colour image
standardization between 1998 and 2024.
Since 2005, many of the advantages of old standards for example
according to the 1MR definitions, and without metadata
are increasingly no longer available, and not any more used.
6. Applications of rgb and/or cmyk data,
problems and solutions
Since about 2005 in applications the historic equal display and print
output with rgb data, and in addition with the three alternative
PostScript-cmyk data has continuesly disapeared.
However, this connection seem essential for a workflow between
display and print and backwards.
Fig. 1: Relation of rgb and cmyk data for the output
by the 1-Minus-Relation (1MR)
For the download of this figure in the vector-pdf format, see
AEB11-7N.PDF.
If in any colour file the rgb* and cmyk* data are connected
by the 1MR definitions, equal output on a printer shall be
produced. In 2024 a colour chaos seems present compared to many historic
applications.
6.1 Application, and interpretation change of the 1-Minus-Relation
Until about 2010 for example an OKI-laser printer for the consumer
area produces output with two different options
without or with colour management.
The option without colour management produces equal output according
to the 1MR definitions. The ICC profiles seem to be the
source of the output change of this printer and many others until 2024.
Therefore, especially for the consumer area the question arises, if
for example ICC profiles (and metadata) have advantages compared to
the historic solutions with the 1MR definitions.
The 1MR definitions are based on colourimetry. For examples the tristimulus
values XYZ of the display primaries RGB and the tristimulus values
XYZc of the complementary (c) CMY colours are connected by the
1MR equations. However, until 2024 the equal display output for the
related rgb* and cmyk* data disappeared nearly completely.
Fig. 1 shows the user benefit of an ergonomic and sustainable print output.
This print output is called 100% Under Colour Removal (UCR).
All grey colours including black are always printed only by the black colorant
and not by an overprint of up to three chromatic colorants.
With free chose of the related rgb* and cmyk* data in the
colour files, the offset-pint method UCR has been used for the print of the
ISO/IEC-test charts according to ISO/IEC 15775:1999 and ed2:2022.
Since about 2005 more and more consumer-print devices ignore the 1MR
equations. For example the four values cmyk=1000 and the corresponding
three values rgb=011 do not print only Cyan C but a mixture of
C and some amount of M or some of Y.
Therefore the printer-device software includes device specific transfers
T (see Fig. 1) between rgb and cmyk data.
An example shows the output of the 1080 colours of the ISO-test chart
AE49 according to ISO 9241-306. The larger squares are defined by
rgb data and the inner squares by cmyk data.
Both are related by the 1MR definitions. The output is not equal
and very different, see
http://standards.iso.org/iso/9241/306/ed-2/AE49/AE490-7N.PDF.
The standard series DIN 33872-1 to -6:2010 includes many test of the
1MR topic, see part 3 and 4 on the page
A/33872E.html.
See for example the pdf file of 11 pages with output questions
on page 11:
B/De14/10L/L14e00NP.PDF.
Now the consumer has nearly no chance to steer an ergonomic, efficient,
and sustainable output on both printers and displays. One reason is the
ignore of the 1MR definitions by the image processing software.
However, equal print output is possible in the professional area
of offset printing, see the example Fig. 1. Also professional printers
for proofing allow to realize main user wishes: equal output according
to the 1MR definitions, and equally spaced according to
ergonomics.
All printers in the consumer area seem to accept cmyk data.
However, the value cmyk=1000 does not print only Cyan C.
An exception seems an available PostScript printer of the
company HP. Usually consumer printers and multifunctional devices
seem to use a printer specific transfer from rgb to cmyk.
The cyan case shows the blocking of main user wishes: application of the
1MR definitions, and towards an UCR output with equal
ergonomic spacing.
The software Adobe Illustrator includes a specific transfers between
rgb and cmyk data in both directions which are not in
agreement with the 1MR definitions of Adobe PostScript.
The user can work either with rgb or cmyk data,
and not with a mixture. If the user has both types of data, then
the software Adobe Illustrator uses Adobe specific transfers
of the file data.
If a user decides to work with rgb data, then for example the
software Adobe Illustrator uses a specific transfer from the
user-cmyk data to the rgb data.
This is similar, if the user decides to work with cmyk data.
For example the user data rgb=011 of cyan are not transferred
to the data cmyk=100 of cyan as expected by most users.
Similar cmyk=100 is not transferred to rgb=011. These
properties of the software Adobe Illustrator and others
are in conflict with colourimetry and the 1MR definitions.
It is proposed to use the 1MR definitions as default
option on any device. If necessary other possibilities shall be
clearly defined similar as the 1MR definitions.
Then for example an ergonomic, effient, and sustainable
100%-UGR output is possible on print and copier devices. The ISO goal to
increase the user-net benefit by ISO standards is then reached.
6.2 Application and ignore of the Gamma value in ps files
The value Gamma=2,4 is used to describe the contrast between the
device white W and the device black N (French=noir). According to
the standards ISO/CIE 11664-4 (CIELAB colour space) and
IEC 61966-2-1:1999 (sRGB-colour space) the rgb* data have
for W and N the values 000 and 111.
The ergonomic rgb* data are spaced proportional to the CIELAB
lightness L* between N and W. Therefore in this case a
star (*) is added to the rgb data. The interpretation is
"visually equally spaced proportional to L*"
For example the different reflections of the ambient light on the display
require for visually equally spaced grey series on SDR and HDR displays
a Gamma change between 1,8 and 3,0. The L* function is
therefore in any application case different.
Fig. 2: Three values Gamma = 2,0, 2,4, and 3,0 for the surrounds white,
grey, and black
For the download of this figure in the vector-pdf format, see
eea00-4n.pdf.
Fig. 2 shows Gamma values between 2,0 and 3,0 which are used for surface
colours in colorimetric applications. For the mean value the relation to
ISO/CIE 11664-4 (CIELAB) and IEC 61966-2-1 (sRGB) is given.
There are two visual effects which also require a change of Gamma.
Both effects are described in the following.
6.21 More light or a more dark colour field compared to whole
viewing field
Visually the contrast decreases or increases, if the viewed
colour field is lighter or darker compared to the whole viewing field.
The colour field includes the viewed colours and the next surround.
Therefore in the more whitish colour field the black threshold increases
and the viewed contrast decreases. The Gamma decreases from 2,4 to 2,0,
see Fig. 2.
<
Therfore in a more blackish colour field the black threshold decreases and the viewed
contrast increases. The Gamma increases from 2,4 to 3.0, see Fig. 2.
6.22 Black threshold increase and contrast degrease with increasing
age
With increasing observer age the stray light in the optical eye media
increases. Therefore the black threshold increases and the viewed
contrast decreases with age. If for a standard viewing field
of surface colours the value is Gamma = 2,4 for the age of 20,
then a decrease to the value Gamma = 1,5 for the age of 60 may
be appropriate.
For the age of 20 the black threshold may be near Y=1,8 compared
to the tristimulus value Y=90 of the paper white W. In 2024 a
professional display with a special coding of the display surface has
a reflection Y=1,8. Most displays may have a reflection near
Y=3,6.
Therefore, the age of the observer, and the display reflection of the
ambient light produce at least a visual black threshold near Y=3,6.
Compared to the paper white it is valid for the contrast:
YW : YN = 90 : 3,6 = 25 : 1. The white W is 5 times
lighter and the black N is 5 times darker compared to the
mean grey with Y = 18 and L* = 50 in CIELAB.
According to the ergonomic standard ISO 9241-306 the white paper and
the white display shall have the equal luminance L=142 cd/m^2.
Then both the white paper and the display have the tristimulus value
Y=90.
Fig. 3 Reflection of CIE-test colours no. 9 to 12 (R, Y, G, B)
For the download of this figure in the vector-pdf format, see
ME040-2N.PDF.
In figure 3 for matte surface colours the reflection 4% appears for all
wavelength compared to the ideal reflection 100% for all wavelength.
Therefore in summary for both the display colours and the surface colours
the Y-contrast range between white and black is approximately
YW : YN = 90 : 3,6 = 25 : 1.
Of course a local adaptation to more blackish or more whitish colours
is possible. This adaptation takes more time for dark compared to light
colours.
6.23 Historic changes of Gamma from 1,8 via 2,4 to 3,0?
Until about 2005 for example the operating system Apple OSX
has used the ergonomic value Gamma = 1,8 for the SDR-display output.
This Gamma value is recommended according to the ergonomic standard
ISO 9241-306 for the standard display reflection Y=2,5 in
offices.
Since about 2005 a new default value Gamma = 2,4 according to CIELAB and sRGB
has been used in OSX. However the ergonomic value Gamma=1,8 was
still possible with a display-output option in the operating
system OSX.
Since 2022 the use of the ergonomic value Gamma = 1,8 has disappeared
in OSX. Most standard images are designed for the value Gamma =2,4.
However, the standard display reflection Y=2,5 requires the value
Gamma = 1,8. The realization of this value is 2024 not any more
possible in OSX.
Since 2022 the default value is Gamma = 2,4 (or 2,2) within the
operating system OSX. No way is known how to change the Gamma.
Therefore often the two darkest grey steps of 16 steps are visually not
any more distinguishable.
This property appears also for new images, which may require
approximately the value Gamma = 3,0 on HDR-displays.
Then additional grey and colour steps of these images may not be
distinguishable on the standard SDR displays.
Some alternate solutions for Apple OSX are available.
Fig. 4: Gamma software and ISO-test charts which include 16 Gamma
values
For the download of this figure in the vector-pdf format, see
eei20-7n.pdf.
Figure 4 shows software which allow to change the Gamma of the
image output. For OSX the GammaAdjuster changes the whole
display output for both still images and video.
This software may replace the GammaSlider on OSX which
disappeared in 2022. The display capture, transfer, and store of a
changed image is not possible.
The software GraphicConverter can change the Gamma of still
images in many file formats. It allows to store new images in many formats
with one or 16 different Gamma values. This allows to choose
the ergonomic one which is visually approximately equally spaced.
Both Gamma-software applications use relative Gamma values.
The relative values between 0,5 and 2,0 correspond to the Gamma range
1,2 <= Gamma <=4,8. The user can decide the appropriate gamma for
his application. This Gamma value depends on the default Gamma of the
display, the display reflection of the ambient light, and the
age of the observer. Yes/No questions are used in ISO 9241-306 to
determine the a special Gamma out of 16 Gammas for the ergonomic
equally spaced output.
Remark: The software GraphicConvertor can transfer an image
in the format eps or ps to a pdf-image. This property of the
software MacPreview on OSX disappeared in 2022.
The software GraphicConvertor can change many other file
formats in both directions.
6.24 Change of Gamma in the PostScript file
The programming language PostScript includes the PS-operator
{ } setcolortransfer. Usually a potential function is used to change
the values rgb in the range 0 to 1, for example by the equation
r' = r^2,4 with the exponent 2,4.
If a file with the extension .eps or .ps in tubtiv
includes a line { } setcolortransfer near the beginning
no change is produced.
However, the use of different gamma values is possible by a
PostScript-frame file. For example the exponent 1,8 in the
frame file is transferred to the image file and applied.
The frame file technology has been used to include both the
1MR equations and the Gamma changes within the eps and ps images.
These properties are then included also in the pdf-file.
Many frame file examples are included in the tubtiv.
After change of a pdf file to a ps file, the frame file can apply
the 1MR equations and the Gamma to the ps file. Then a new
pdf file with the intended properties 1MR and Gamma can be
produced.
Remark: For many years the web browser Safari of OSX has
transferred automaticly all PostScript files in the format eps
and ps of tubtiv to the pdf-image output.
However in 2022 this property disappeared in the web browser
Safari of OSX.
6.25 Ergonomic equal spacing and image quality specification
ISO/IEC 15775, Annex G, defines the quality of the image input and output
by the regularity index g*. For the ergonomic output for example
the 16 step grey scale shall be equally spaced in CIELAB-lightness
L*.
ISO/IEC 15775 defines digital and analog test charts. For example
16 equally spaced grey steps with equal differences are specified by:
rgb* values: 0, 1/15, 2/15, ..., 14/15, 15/15 with delta_rgb*=0,066
L* values: 20, 25, 30, ..., 90, 95 with delta_L*=5,0.
The digital ISO/IEC-test charts with the above rgb* value
have been used to produce the analog ISO/IEC-test chart no. 3 with the
above L* values on offset paper. The analog test chart no. 3 and a
test chart no. 1 on photographic paper in the range 10 <= L* <= 95
are called the originals for a colour copier.
The quality of the copy will be studied for one example output paper.
The standard office paper has the L* range: 25 <= L* <= 95.
Fluorescent copy papers have a larger contrast range in L*,
and recycled copy papers have a lower contrast range.
ISO/IEC 15775 specifies the output quality by the regularity index:
g* = 100 delta_L*min / delta_L*max
If the ergonomic output appears in the copy by equal spacing,
then by this equation the value is g* = 100.
If delta_L*min = 0, then two steps are not distinguishable.
In this worst case the value is g* = 0.
The output value g* = 100 is called an very good copy.
The goal of ICC-colour management is very different. The goal is the
minimum colour difference between the original and the copy.
The minimum colour difference is produced on the standard paper,
if the two dark steps with the values L* = 20 and 25 are both
reproduced equal with L*=25 in the copy. The 14 other steps
shall be equal in L* compared to the original
to produce the minimum colour difference.
The mean ICC-colour difference is then according to the formula
delta_L*mean =
[ delta_L*1 + delta_L*2 + delta_L*3
+ ... + delta_L*14 + delta_L*15] /15
= [ 5 + 14 x 0] / 15 = 0,3
Therefore the quality specification of the copy is opposite:
"very good" by the colour difference delta_L*mean = 0,3.
This value is smaler compared to the visual threshold delta_L* = 1.
"very bad" by the regularity index g* = 0.
The reason is the value delta_L*min = 0 of the two equal dark steps of 15.
For the copy example the quality specification is opposite
"very good and very bad":
The quality value delta_L*mean = 0,3 specifies an very good quality
according to ISO 20677;2019 (ICC color management).
Conclusion:
For an ergonomic output an equal spacing is required. Otherwise important
colour image information is lost.
It is therefore required to specify the colour image quality of both the
original and the copy by the regularity index g*
according to ISO/IEC 15775:2022, Annex G.
In colour applications there are shifts of the L* range, for example
smaller L* ranges with lower contrast on recycled paper, and
larger L* ranges with larger contrast on photographic paper.
In any case teh quality shall be spezified by the regulaty indes g*.
The mean colour difference according to ISO 20677 (ICC colour management)
is not appropriate to specify the quality.
Future application for colour copiers:
For example both users and manufacturers of a colour copier on
photographic or high glossy paper wish, that the whole
contrast range of the copier is used.
Proposal for the specification of the image quality for copiers
1. The mean colour difference is not usefull and shall be deleted
to specify the quality.
2. The regularity index g* shall be used.
The new default option of colour management shall be the
ergonomic colour management. The quality specification is
based on the regularity index g*.
This index may be modified by a factor f which is defined
by the ratio of the device (d) range
delta_L*d and the standard offset (o) device range
delta_L*o = 75 according to ISO/IEC 15775:2022.
Then this modified index increase with the contrast range.
The image quality shall not be specified by the mean difference according
to ISO 20670 (ICC colour management). Since 2010, more and more
special image corrections, for example "black point adjustment" are used.
These corrections are used with metadata and together with ICC profiles
according to ISO 20670. However, every lightness step must be adjusted
and not only the steps near black.
Since 2010, many user application programs for consumers
have increasingly used their own interpretations of ICC profiles and
metadata or ignore these. This is another source for the colour chaos in the
consumer area. ICC profiles and metadata are not required for
the ergonomic colour management according to ISO/IEC 15775:2022
and ISO 9241-306:2018.
6.26 Example of historic prints of 16 step grey scales
The author received in 1999 a printed official output of the standard
ISO/IEC 15775:1999, and in 2019 a printed official output of the
standard ISO EN DIN 9241-306/ed-2:2019
Both include prints of 16 step grey and colour scales.
The 1999 output includes 16 equally spaced grey steps.
The 2019 output shows four dark grey steps, which are visually
not distinguishable.
The image quality shall be specified according to the regularity index
g*. The 1999 output is specified by a value near g* = 100
and may be called "very good". The 2099 output is specified
by a value g* = 0. This worst case may be called "very bad"
and is not acceptable. The reasons for this example case is unknown.
One possible (wrong?) explanation:
If the pdf-file data rgb* are specified for the range
0 <= L* <= 100 of a display output without ambient reflection,
then four steps with L* = 0,0 6,6, 13,2, 20,0 can not be reproduced
and are printed with L*=25. This output has then the regularity
index g* = 0. This output is neither ergonomic, nor efficient, nor
sustainable because too much toner is used.
However, the correct output shall be equally spaced in the range
25 <= L* <= 95 with delta_L* = 4,66 between all neighbouring
steps. This output has the regularity index g* = 100.
Then the quality is "very good". This case has been realized in 1999.
7. Colour loop, benefits, and applications without metadata
Fig. 5: Colour loop ISO-rgb* file - print - scan -
ISO-rgb* file
For the download of this image in vector-pdf format, see
eeb00-3n.pdf.
Figure 5 shows an ISO file with equally spaced digital rgb*-colour
data with 000 for the device black and 111 for the device white.
Within the colour loop, after printing and scanning, approximately
the same rgb* colour data are generated compared to the beginning.
This applies for all colour-devices combinations, also for example
for the display output and the photo scan of the display output.
For the realization of the colour loop in Figure 5 the manufacturers
or users linearize their colour devices.
The technical basis for the implementation can be found in the standard
documents which are included in Figure 5. A TUB-colour vision model
for chromatic and luminance adaptation allows the application in the
following display areas: Low, Standard and High Dynamic Range
(LDR, SDR and HDR).
The application of the colour loop of Figure 5 is ergonomic, efficient and
sustainable.
8. Summary
A new future TUB development entitled:
Ergonomic architecture of colour-image processing with
chromatic and luminance adaptation without colour image metadata
may use the following developments:
1. the TUB text_image_vocabulary tubtiv for education and
standardization.
2. the ISO colour loop with standard documents in Figure 5.
3. The "Antagonistic TUB-relativity model of colour vision for
a wide range of luminance and chromatic adaptation", see
eea_s.htm.
Annex A. Structure of the "TUB text_image_vocabulary (tubtiv)"
The colour-image series consist of three Sections: 0 (1998-2010),
1 (2011-2018) and 2 (2019-present).
The latest image series of Section 2 (2019-present) is at the beginning.
Older image series of Section 1 (2011-2018) and the
Section 0 (1998-2010) follow.
All three Sections can be downloaded separately.
To do this, the file "tubtiv_ahe" with the three Sections
is replaced by "tubtiv_2he", "tubtiv_1he",
or "tubtiv_0he" with one Section in each case.
All files contain text information with links to the pdf-image series,
to the pdf-image in the size A4, and often to 16 individual pdf-images.
Section 2 (2019-present) contains: 5x260 image series.
Both Sections 1 (2011-2017) and 0 (1998-2010) each contain
26x100 image series.
All images are available for free download on two TUB servers, see
http://farbe.li.tu-berlin.de/index.html or
http://color.li.tu-berlin.de/index.html
Annex B. Copyright
For free copyright see:
http://farbe.li.tu-berlin.de/CEV1/CEV10-3N.PDF
in English or:
http://farbe.li.tu-berlin.de/CGV1/CGV10-3N.PDF in German.
Acknowledgements
For discussions and proposals to optimize the content of the tubtiv
I thank especially:
Detlef Ruschin, German Representative in CIE Div. 1
"Vision and Colour", and
Prof. Florian Suessel, German Representative in CIE Div. 8
"Image Technology".