Graeme, thank You for paying attention for my questions. I've checked again. I've think there was a mistake in my first checks. But I do another check again and the results are the same: output images almost independent of tiffgamut intent. The difference is in diagnostic *.wrl, but for real out of printer gamut images is not noticeable. See workflow.txt and "25(a-p - r-p) difference.jpg" from archive. May be mapping algorithm as universal and robust so real white point mismatch is not a problem? -----Original Message----- From: Graeme Gill <graeme@xxxxxxxxxxxxx> To: argyllcms@xxxxxxxxxxxxx Date: Mon, 08 Sep 2008 17:03:49 +1000 Subject: [argyllcms] Re: Device link gamut mapping: how to do it right? Nikolay Pokhilchenko wrote: > I have doing linking with perceptual intent after "tiffgamut -ia -pj". I've > noticed > that the result of that linking is white point mismatch between source and > destination > gamuts, which seen in checking *.wrl. Yes, because that is effectively what you asked for : - an absolute space in appearance coordinates using the colorimetric table. It's probably not documented very well, but to support absolute reproduction while using appearance space, argyll supports the concept of an absolute appearance space. You're using -ia with -pj triggers this. It probably isn't what you want if you intend to use the resulting gamut with collink. [ Currently "absolute appearance" space uses uses a D50 white point for tiffgamut, but for collink it uses a "half way" white point between the source and destination color spaces. I'm not sure how this difference could be resolved in a clean way. ] > The source image profile was sRGB, destination > profile was printer profile for A-type illuminant. While run "tiffgamut -ir > -pj" before > collink, there is other result - no white point mismatch in diagnostic *.wrl. Right, because this triggers a "relative" appearance space, that uses the colorimetric table. ? But I > cannot distinguish the difference between profiles and output TIFFs with > different -i > keys in tiffgamut! The results are rather good, interdependent of tiffgamut > -i key. I > can't understand how must I do linking the best way, with what keys... Hmm. I'm not sure why that would be the case. Perhaps the white point difference isn't large enough to result in a visible gamut mapping change. Graeme Gill.