This mail had bounced so I try from a different mail account. That´s why my reply is rather late now, but since the thread is still alive...
Hello Matthew,
Am 21.02.2007 um 22:50 schrieb Matthew Robinson: I found something rather unexpected when converting a colour from sRGB to CMYK using this new ISOCoated_v2_eci.icc in Photoshop CS2.
the glitch you found in the new eci profiles is not the only one. I can send a profile analysis off-line, if you wish. Or you can use the test files you find on http://colormanagement.org and find out for yourself.
The good news is that there are better profiles available, which are based on the FOGRA39 characterization data (free download from http://colormanagement.org as well – explanations in German only, at the moment, sorry, working on that). Use these to convert the test files and compare the results. If you look at the gamut shapes of the eci profiles (e.g. in ColorSync utility) you will find them very different, depending on the RI, while the basICColor profiles retain the same shape across the RIs. And there is more in the eci profiles: color shifts (Look at 255 R, 255 G for examle), banding etc.
Here are the results of the same test you made, based on a conversion with ISOcoated_v2_bas.ICC : The below conversions were all done in Photoshop CS2 using Engine=Adobe(ACE) & Black point compensation on.
R=0 G=0 B=255
converted to:
Intent = Relative Colorimetric, ISOCoated_v2_eci.icc C=95 M=100!! Y=0 K=0
ISOcoated_v2_bas.ICC C=100 M=77 Y=0 K=0
The ISO version is turning the blue, purple! Where as the Fogra27 is strikingly more blue.
And for comparison the following differences show up when changing the intent.
Intent = Perceptual ISOCoated_v2_eci C=96 M=81 Y=1 K=0
ISOcoated_v2_bas.ICC C=100 M=76 Y=0 K=0
So with the Perceptual Intent the ISOCoated_v2_eci magenta content is less than Cyan creating a blue, slightly 'bluer' than even the Fogra27.
Intent = Absolute Colorimetric
ISOCoated_v2_eci C=87 M=100 Y=0 K=0
ISOcoated_v2_bas.ICC C=100 M=77 Y=0 K=0
Remember? Relative with ISOcoated_v2_bas.ICC is: C=100 M=77 Y=0 K=0 Perceptual is: C=100 M=76 Y=0 K=0
This shows that the perceptual rendering intent can now be used as intented by the ICC when they first defined the idea of profiled color conversions some 13 years ago!
Is this a bug in the new ISO profile!?
Yes, and it´s not the only one. Even if it's not a bug, I think it's rather concerning and is enough for me to put plans for changing over to using the ISO profile on hold until I've got more information and done more testing. (By the way, the same type of results are also encountered when using the uncoated and 300 versions of the ISO profiles)
No surprise here, same profiler.
We will continue to publish free profiles for all standard printing conditions. They may not be perfect, but still better than the so-called standard.
In order to be consistent with workflows using the eci profiles (not an official standard, BTW, just one of severlal interpretations of the FOGRA39 characterization data), you can embedd the these after separating your files with the basICColor profiles. The proofing tables (A2B) are nearly identical (average delta E 0.1). Doing so will avoid another conversion in a color managed workflow, that might occur when embedding the basICColor profiles.
Best regards,
Karl Koch |