
|
[argyllcms]
||
[Date Prev]
[01-2008 Date Index]
[Date Next]
||
[Thread Prev]
[01-2008 Thread Index]
[Thread Next]
[argyllcms] B2A mapping of out-of-gamut colors
- From: Thomas Baumann <thomas@xxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Wed, 02 Jan 2008 11:13:06 +0100
Hello list,
when I print a Lab test target through any of my Argyll generated printer profiles there is a
peculiarity, most pronounced with relative colorimetric intent: Dark out-of-gamut colors with high b
values turn into a light yellow - see file Screenshot_Softproof_Argyll.tif.
Example (B2A1, L = 9, a = 0, b = 80):
$ xicclu -fb -ir SP3800_PSPP_2880_ARG2.icc
9 0 80
9.000000 0.000000 80.000000 [Lab] -> Lut -> 0.982462 1.000000 0.000000 [RGB]
This corresponds to the inverted A2B table with CAM Jab mapping (switch -b):
$ xicclu -fif -ir -a -b SP3800_PSPP_2880_ARG2.icc
9 0 80
9.000000 0.000000 80.000000 [Lab] -> Lut -> 0.982441 1.000000 0.000000 [RGB]
(clip)
[Actual 90.854331 -3.834409 115.588589, deltaE 89.338580]
So the problem seems to lie in the gamut mapping 9 0 80 [Lab] -> 90.854331 -3.834409 115.588589
[Lab], from a very dark to a very light yellow. When I open the profile in ColorSync, I can see that
for positive b, the B2A1 curves for both R and G over L change slope and rise up to maximum with
decreasing L, thus producing a tonal inversion. What could be the reason for this mapping behaviour?
It does not seem right to me.
Another detail I do not understand: When I specify absolute instead of relative colorimetric intent
in Photoshop CS Soft Proof, the effect vanishes, although xicclu -ia ... indicates it should be present.
For reference, I used these command line parameters with 'profile' version 0.70
Beta 7 on Mac OS 10.5:
profile -qh -i D50 -f -S /usr/local/etc/argyll/AdobeRGB1998.icc -cmt -dpp
SP3800_PSPP_2880_ARG2
All relevant files can be found here:
http://www.tbaumann.de/public/argyll/SP3800_PSPP_2880_ARG2.ti3 (1593 patches,
measured with DTP20)
http://www.tbaumann.de/public/argyll/SP3800_PSPP_2880_ARG2.icc
http://www.tbaumann.de/public/argyll/Lab_Test_Image.tif (smaller variant of
Bill Atkinson's target)
http://www.tbaumann.de/public/argyll/Screenshot_Lab.tif (for easier viewing)
http://www.tbaumann.de/public/argyll/Screenshot_Softproof_Argyll.tif (relcol,
with BPC)
http://www.tbaumann.de/public/argyll/Screenshot_Softproof_Argyll_abscol.tif
(no problem here)
For comparison:
http://www.tbaumann.de/public/argyll/Pro38_PSPP.icc (Epson's canned profile)
http://www.tbaumann.de/public/argyll/Screenshot_Softproof_Epson.tif
Regards
Thomas
|

|