181212 and 030101: A/DE81/DE811.HTM

For the start page, see DE81 in English or DG81 in German.

For this page with MTL code for OL with:
CIEXYZ data of luminous colours, see IE811 in English and IG811 in German,
CIELAB data of surface colours, see DE811 in English and DG811 in German,
For additional Information, see IE841 in English and IG841 in German,
DE841 in English and DG841 in German.

1.0 Colour Workflow (CW) for many different input and eight different output PS operators (device dependent and device independent).

Remark: The transfer of an input PS-operator, which use relative cmyn* coordinates, to a different output PS operator, which use olv* (rgb*) coordinates, is a main part of the colour workflow.

It is intended to produce the same grey scales for corresponding colours defined colorimetry by CIELAB in 8 different corresponding colour spaces. The 5 and 16 step grey scales (Fig. C2 and C3 of ISO/IEC-test chart no. 3) are studied here to reach this goal.

Remark: One may study first the output of the 5 step grey scale (Fig. C2 of ISO/IEC-test chart no. 3) in DE80 (6 input PS operators) or DE83 (4 input PS operators)

The corresponding colours are defined by a PostScript MTL code (Measurement, Transfer and Linearization) which can be stored at different places in the colour workflow, e. g. the file, the PS printer, the printer driver (PPD file) or the Adobe Distiller Startup directory, compare the technical paper BAMGREY.PDF (2300 kByte, 16 pages) and others, see XY91

Any PS file which use many input PS operators can be changed in the same direction as the following example files if the PS MTL code is stored at an appropriate place (see above).

Remark: It is possible to produce from any application or any PDF file a PS file. By the computer operating system for the production of a PS file a PS printer must be selected, e. g. the black and white Apple Laserwriter PS/640. Only a PPD file (PS Printer Description file) must be selected, in reality this printer must not be present. So it is possible at any computer operating system to produce PS files.

2.0 Colour Workflow (CW) and files for Fig. C3 of the ISO/IEC-test chart no. 3 with 16 step grey scales which are equally spaced in CIELAB, see DE81

Remark: The device dependent coordinates cmyn* (of standard offset printing according to ISO/IEC 15775) are transferred to device independent LAB* coordinates in the CIELAB colour space by the MTL code. For every other device (different monitors and printers) there are equations which transfer from the device independent coordinates LAB* to the device dependent coordinates and vice versa. The colour difference accuracy is intended to be below 3 in CIELAB and therefore the device must be linearized in CIELAB. The linearization method is described at the end.

Remark: The files with the extension .DAT include the PostScript MTL code which is also included in the FP files and then used for the file output. The content of the six files with the extension .DAT is identical, only the name and one parameter i*ptrsc=5,6,7,8,9,4 differs. This parameter selects the one output PS operator (see second column of Table 1 of DE81 and around line 20 in the DAT file).

It is useful to study the NP file L81E00NP. PDF and the two FP files, e. g. L81E00FP. PDF and L81E02FP. PDF to realize some basic problems in image technology with Adobe Reader 3.0 or later.

Remark about the size of the files: The size of the NP.PS and NP.PDF files is about 130 kBytes, the size of the FP.PS and FP.PDF files is about 230 and 350 kBytes. The size of the following tiff files is about 350 kBytes.

The output of the four corresponding 16 step grey scales of the NP.PDF file shows differences on the monitor. The two PS operators nnn0* setcmykcolor and 000n* setcmykcolor used for output give very different results with the software Adobe Reader 3.0 or later.

The four corresponding grey scales of every FP.PDF file (see Table 1 of DE81) agree in output but the six L81E0iFP files (i=0,1,...,5) differ in output with the software Adobe Reader 3.0 or later.

L81R0NFP.tiff (250 kByte)

Fig. 1: Screen foto of the two files L81E00NP. PDF and L81E00FP. PDF are both viewed by the software Adobe Reader 5.0 (label R=Reader in filename on Macintosh OS X 10.2).

Remark about the tiff format: In application the software Mac Preview (see later) is a part of the software Apple Quicktime which comes with Microsoft Explorer 5.0 on Windows and Mac and produces the image of the tiff format by the web browser Plug-In.

The four corresponding grey scales in the NP file output are different with different software, e. g. Acrobat Reader, Mac OS X Preview.

The four corresponding grey scales in the FP file output (right) are equal with different software, e. g. Acrobat Reader, Mac OS X Preview.

Therefore any colour management system depends at least on the input PS operators used (here four possibilities) and the colour workflow (transfer from the input PS operators to the output PS operators) used. The colour workflow is defined here by the transfer between four different input PS operators to one output PS operator cmy0* setcmykcolor or 000n* setcmykcolor in the FP file.

Remark 1: The six NP files of Table 1 have the same content and differ only by the name (and one PS parameter i*ptrsc=0 to 7 at the beginning which is not used in the NP file but used to select the PS output operator in the FP file). The colours are defined by two different input PS operators, see the figure descriptions in the output.

Remark 2: The four corresponding grey scales look the same in each FP file but the different FP files look different and show the same effect as the different grey scales in the NP files.

Remark 3: It is intended to produce the same grey scales for corresponding colours defined by CIELAB colorimetry in corresponding colour spaces.

It is useful to consider again the NP file, e. g. L81E00NP. PDF to realize some more basic problems in image technology.

L81RP0NP.tiff (250 kByte)

Fig. 2: Screen foto of the same file which is viewed by the software Adobe Reader 5.0 (left, label R) and by the software Mac OS X Preview (right, label P in filename)

If one uses the following software:
Adobe Reader 3.0 or later, Adobe Illustrator 8.0 or later, Adobe Photoshop 5.0 or later
then the NP file shows very different output on the monitor for the PS operator 000n* setcmykcolor (first row in output) compared tonnn0* setcmykcolor (third row in output).

If one uses one of the following software:
MAC OS X Preview, Omniweb Web browser
then the NP file shows the same output on the monitor for the PS operator 000n* setcmykcolor compared to nnn0* setcmykcolor

According to CIELAB colorimetry for the corresponding coordinates in corresponding colour spaces the same output is expected. The software Adobe Reader is not in agreement with this basic requirement of colorimetry which is at the same time a basic user requirement. The equivalent software MAC OS X Preview agrees in this point with CIELAB colorimetry which is shown in Fig. 3.

L81P0NFP.tiff (250 kByte)

Fig. 3 Screen foto of the two files L81E00NP. PDF and L81E00FP. PDF which are both viewed by the software Mac Preview (label P=Preview in filename on Macintosh OS X 10.2). The four different PS operators used in the NP file give the same output and the output of the FP file is the same compared to the NP file.

Basic results:
1. There is software, e. g. Mac OS X 10 Preview, which produces equal output for corresponding input PS operators in different colour spaces, e. g. for the PS operators 000n* setcmykcolor, w* setgray, nnn0* setcmykcolor and and www* setrgbcolor.
2. If one compares the intended equal spacing of the 16 grey steps, e. g. by the Landolt-rings, tthen here are large differences on the monitor between different software, e. g. the software Adobe Reader 3.0 to 5.0 or Mac OS X 10 Preview.

3.0 Basic user requirements for image technology software

1. For the corresponding colours in different colour spaces the same output colours must be produced.

2. For 16 step colour scales with equal digital spacing in the file 16 step output colour scales which sre equally spaced in the CIELAB must be produced.

According to the first user requirement the following six PS operators mut produces the same grey output colour:
1. 0.0 0.0 0.0 0.5 setcmykcolor
2. 0.5 setgray
3. 0.5 0.5 0.5 0.0 setcmykcolor
4. 0.5 0.5 0.5 setrgbcolor
5. 0.5 0.0 0.0 setcolor (in relative CIELAB space)
6. 56.7 0.0 0.0 setcolor (in absolute CIELAB space, offset range 18.0 to 95.4).

Similar the following three PS operators must produce the same cyan colour
3. 0.5 0.0 0.0 0.0 setcmykcolor
4. 0.0 0.5 0.5 setrgbcolor
6. 56.7 0.0 0.0 setcolor (in absolute CIELAB space, offset range 18.0 to 95.4)

The first basic user requirement is to produce the same grey or the same cyan blue at least for the above PS operators no. 3 and 4. The NP file shows very different output on the monitor by Adobe Reader 3.0 to 5.0. In the case of monitor differences still about 10% of the PS printers on the market show the same PDF output. For about 90% of the printers the large differences on the monitor remain in the output. Often additional differences between equal scales on the monitor appear on the printer.

Result ot the test of the first basic user requirement
For the monitor and printer output e. g. the software Adobe Reader is not in agreement with the first basic user requirement.

Many of the Display PostScript Systems (e. g. Silicon Graphics DPS and Photoshop, Compac VAX Decwrite, Mac OS X Server: application Yap) and many of the Display PDF systems are in agreement with the first basic user requirement.

Remark ot the test of the second basic user requirement (compare clause 5.0)
The second basic user requirement is violated by most of the software. This requirement to produce a mean grey or mean cyan blue which is visually (and in CIELAB lightness L*) in the middle between the CIELAB data of white (L*w=95.4) and black or cyan blue is often not addressed. Instead of the relative whiteness w*=0.5 (which corresponds to the lightness L*=56.7) or relative cyan blue c*=0.5 the different software on the market produce a relative whiteness w* or relative cyan blue c* between 0.25 and 0.75. The agreed tolerance range according to ISO/IEC 15775 is 0.47 to 0.53.

Remark: One of the reasons for the large differences are the different devices as source of the imaging data. The raw data of most scanners show a linear relationship as function of the luminance of the original, the raw data of a professional video camera are proportional to the square root of the luminance and the raw data of the slide or negative film transmission data are proportional to the log of the luminance. Software and users must therefore consider very carefully the source of the data and the complex properties of the devices and the device software.

4.0 Different solutions which are in agreement with the two basic user requirements

It is necessary to use at least a 16 step grey scale to include the second user requirement of a linear relationship between w* input and L* output coordinates. This linear relationship makes the system compatible to CIELAB. Similar linear relationships are intended for the six 16 step colour scales.

The possible solutions may be first studied with the 5 step instead of 16 step grey scales for six different input and six different output PS operators in DE80

4.1 Solution to get the same output independent of the colour space (PS operator) used

All FP files show the same output which is independent of the input PS operator used. Therefore one can select an appropriate output PS operator. This selection is device and software dependent. Two examples for the output of the file L81E00NP.PDF for monitors and printers are given:

On monitors and with:
1. e.g. the software Adobe Reader 3.0 to 5.0 it is not appropriate to use the output PS operator no. 3 nnn0* setcmykcolor because of the brown appearance of the mean grey colours. One may select no. 4 which appears neutral.
2. e. g. the software Macintosh Preview one may select any of the four (there is no brown colour output).

On PS printers with the software Adobe Reader 3.0 to 5.0 and:
1. e.g. the PS laser printer Phaser 740 one may select one of the first two which looks neutral. The last two appear yellowish green.
3. e. g. the PS laser printer OKI 8e one may select no. 2 which is the only neutral and the other three look greenish brown and different.

For the 16 step colour series the choise is more difficult and no recommendation is given here.

4.2 Solution to get an equally spaced output independent of the colour space (PS operator) used

The output of the 16 step colours must be measured in lightness L* of CIELAB. Alternative the L* data may be determined visually. For this it is appropriate to use the reflective and transparent ISO/IEC-test charts as reference for the comparison for printers and monitors.

In a first step the 16 LAB* data of the start output must replace the corresponding 16 reference data in the file L81E00FP.PS at the beginning of the file. One must replace the appropriate LAB* data of the four 16 step grey series N­W or A­W which are produced by the output. One must search for this replacement the corresponding LAB* data lines, labels and PS operators.

LAB* data line, label, data type, and PS operators

no. 048 to 063: N-W L 000n* setcmykcolor
no. 112 to 127: A-W L nnn0* setcmykcolor (A=all three colours cmy*)
no. 176 to 191: N-W L w* setgray
no. 240 to 255: A-W L www* setrgbcolor (A=all three colours olv*)

The standard data type is L=Linear data for the files L81.., but for some special files Q81.. and S81.. which produce lighter or darker output the data type is Q=Quadratic data or S=Square root data (see clause 4.3).

It is easy to find the appropriate LAB* data lines lines for the PS operators with any editor in the FP file. The new file may be saved with a new PS file name. The next steps are the transfer of the PS file to the PDF file and the linearized output. One can check if the new spacing is linear by the determination of the new LAB* output data.

In general the CIELAB colour difference is reduced by a factor 3 to 5, e. g. from 9 to 3 or 15 to 3. The value 3 is similar as the value for the intended tolerance and the accuracy of roproduction across a printer or screen page.

4.3 Two examples, one example solution and results

1. In the file S81E00FP.PS we have replaced the L=Linear L* data by S=Square root L* data. The output of the file S81E00FP.PDF looks much darker. This shows the effect and may be of some value if the start output is too light.

2. In the file Q81E00FP.PS we have replaced the L=Linear L* data by Q=Quadratic L* data. The output of the file Q81E00FP.PDF looks much lighter. This shows the effect and may be of some value if the start output is too dark.

Ony possibility for a general use for any PS file is to make a copy of the Adobe Acrobat Distiller Directory.

Remark: The following method has been tested with Adobe Acrobat 3.01 and 4.0 on Macintosh 8.6 and 9.2. If this does not work with an other Acrobat Distiller version and an other computer operating system there are other methods instead of using the Distiller Startup directory. The following is only an example. Organize the Hard disk as follows:

Distiller directory ­ subdirectory Startup ­ includes one default file: Example.ps

Copy of the Distiller directory ­ subdirectory Startup ­ includes the default file and one user file: Example.ps and S84E00FP.DAT

The copy of the Distiller directory differs only by the file S81E00FP.DAT in the Distiller Startup subdirectory. For a screen shot with this kind of organization see

DISTILLER.tiff

Fig. 4: Original and copy of the Adobe Acrobat Distiller 4.0 directory. The copy differs only by the file S81E00FP.DAT in the Distiller Startup subdirectory.

Step 1: Produce a copy of the directory Adobe Acrobat Distiller 3.0 or later

Step 2: Copy the file S81E00FP.DAT into the subdirectory Startup of Adobe Acrobat Distiller 3.0 or later which includes the file Example.ps.

Step 3: Produce from the NP file L81E00NP.PS two copies X81E00NP.PS and Y81E00NP.PS. Delete in the second the lines between %BEG DE81 and %END DE81 (line 21 to 59). There is no PS MTL code in the second file.

Step 4: Use the PS file X81E00NP.PS to produce the PDF file X81E00NP.PDF and use the original Adobe Acrobat Distiller (in the subdirectory startup there is only the file Example.ps, see Fig. 3).

Step 5: Use the PS file Y81E00NP.PS to produce the PDF file Y81E00NP.PDF with the copy of Adobe Acrobat Distiller (in the subdirectory startup there is the file Example.ps and the additional user file S81E00FP.DAT).

Step 6: Compare on the monitor the output of the two files X81E00NP.PDF and Y81E00NP.PDF in two windows of Adobe Reader 3.0 or later.

Result: The PDF file X81E00NP.PDF includes four different PS operators and produces different grey scales Adobe Reader 3.0 or later. The PDF file Y81E00NP.PDF produces four identical grey scales and all are darker compared to the first one if viewed by Adobe Reader 3.0 or later.

XY81R0NP.tiff

Fig. 5: The two files X81E00NP.PDF and Y81E00NP.PDF are both viewed by the software Adobe Reader 5.0 (on Macintosh OS X 10.2).

Interpretation: The copy of Adobe Acrobat Distiller takes the PS MTL code which is in the corresponding Distiller Startup directory if the second PDF file Y81E00NP.PDF is produced from the second PS file Y81E00NP.PS. The change from four input PS operators to one output PS operator and a square root output (darker grey scales) is produced by the PS MTL code of the file S81E00FP.DAT which is included in the subdirectory Startup of Adobe Acrobat Distiller 3.0 or later.

The four input PS operators:

1. 000n* setcmykcolor
2. w* setgray
3. nnn0* setcmykcolor
4. www* setrgbcolor

are transferred by the PS MTL code (which is in the distiller startup directory) to one output PS operator

000n* setcmykcolor

The output of the four 16 step grey series is the same on the monitor with Acrobat Reader and the output is equally spaced (if the start output L* data are determined and included in the file S81E00FP.DAT).

Conclusion: The second PS-file Y81E00NP.PS does not include any PS MTL code and therefore any PS-file may be used with any mixture of the above four PS-operators no. 1 to 4 for the intended corrections.

5.0 Comparison of the above example solution with the two basic user requirements

The first basic user requirement is that the same grey is produced and the
second basic user requirement is that a mean grey is produced.

There is a linear relationship between the input w* data (or the n*=1­w* data) and the output lightness L* data of the 16 step grey scale. Therefore a mean grey L*z = 0.5 (L*w ­ L*n) is produced for w*=n*=0.5.

The above example solution fulfils both basic user requirements.

Final result: The Colour Workflow (CW) and the Output Linearisation (OL) of the PS MTL code is based on ISO/IEC 15775 and DIS ISO/IEC 19839-X. The PS MTL code is compatible to CIELAB and fulfils the basic user requirements in the field of image technology.

This result for grey scales is identical to the final result in DE841 for the chromatic colours.

-------
For an introduction (archive 2000-2009) to all folders, see
../DE.HTM in English or ../DG.HTM in German.

For information (archive 2000-2009) about International and National Standards, Technical Reports, and Meetings, see
../4STAE.html in English or ../4STAG.html in German.

For information (archive 2000-2009) about colour test charts and colorimetric calculations, see
../indexAE.html in English or ../indexAG.html in German.

For publications (archive 2000-2009) of the BAM research group, see
../XY91AE.html in English or ../XY91AG.html in German.

Back to the main page of the TU web site (not archive), see
../../index.html in English or ../../indexDE.html in German.

The last link may not work, if you have visited directly this archive site.
Then go to the main page of one of the two TU web sites in a new window, see
http://farbe.li.tu-berlin.de/index.html or
http://130.149.60.45/~farbmetrik/index.html.