Hi,

>Does LCMS always compute the gamut on the fly, ie ignore the gamut tag
>when it is there?

Yes. gamut tag is unreliable. Many other CMM does same, Adobe ACE
for example. I've reported that problem to ICC. v4 profiles are going to fix that.

>Does it work reasonably well with LCMS Profiler's profiles?

Humm... well, primary goal of gamut check are output profiles. It would be 
quite difficult to use it on input profiles, since there are no out of gamut colors 
in input direction. An example: You want to render Lab=(50, -120, -120) on 
output profile. This is out of gamut, so the output profile cannot give any RGB 
for this entry. It uses gamut remapping and gives the most saturated blue it can, 
for example RGB=(0, 162, 255).  But on input, you are giving the RGB values, 
so they are already in RGB gamut. This would never give Lab=(50, -120, -120) 
because no RGB combination results in such Lab. No remapping is needed in
input direction!

>Does it work better for certain types of profiles (CLUT / matrix shaper)?

CLUT works better. Matrix shaper doesn't work very well. Its gamut is 
theoretical, and limited only by RGB components. Also, black point is at zero, 
which is not physically feasible.

>You say LCMS uses the rel.colorimetric CLUT (?) tables for the gamut 
>warning. What do you do in case of matrix shaper profiles?

It uses the only intent matrix-shaper has. It is already non clear if that is rel. 
colorimetric or (as I believe) perceptual. Think on matrix-shaper as a 
rel.colorimetric plus black point compensation. 

>How would a profile have gamut remapping embedded?

See above. Lab=(50, -120, -120) cannot be represented on the output device, so
the profile has to remap these out-on-gamut colors to whatever inside device space.
It cannot simply clip values since that would block all gradients from, say, 
Lab=(50, -40, -40) to Lab=(50, -120, -120). Gamut mapping is a hard subject and
that is what makes output profiles such difficult.


>I have experimented with a highly saturated image (rich yellows) and I
>found that in this particular case LCMS' matrix shaper profile did a
>great job. There was no gamut warning. When converting to A98 using LCMS
>engine, there was still a little clipping. But when using ICM or ACE
>there was none. Therefore I thought that the gamut warning is at least a
>good indicator which profile to choose.

Ops, you mean you have found a combination of profiles that behaves 
differently on lcms and Photoshop? I would be glad to take a look on these. 
Can please email-me a copy? I would destroy the profiles after checking what 
the problem is.

Thanks
Marti.






-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
Lcms-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/lcms-user

Reply via email to