Hello Nikolay, I can see your point about AdobeRGB vs. eciRGB. It would be logical and I tried that one also. The result is in deed more correct -but AdobeRGB simply performs better, especially in high-saturated reds. There is a lot more modulation and structure visible. The skintones are slightly more yellowish, witch I don´t really see as a problem (the image will be colour-corrected in Photoshop anyway). The black generation curve I posted had to achieve 2 basic targets: 1. To generate a UCR - separation (as mutch as possible) 2. The separations should be smooth (eg. when converting a LAB gradient [100, 0, 0 to 0, 0, 0]) I tried also to generate profiles where K is rather a locus (witch complies to basiccolor or HDB-profiles). After running xicclu -v -g -fb -ip -o, the curves complied to the visual look of my separations generated from the previous mentioned LAB-gradient. That means that they were not as smooth as in the -kp-example I´ve posted allready. I´ll try some other black generation curves anyway. Thank´s for the link you posted -maybe I´ll find my answers there. Regards Vladimir 2010/3/26 Nikolay Pokhilchenko <nikolay_po@xxxxxxx> > Vladimir Gajic wrote: > > > Hi Graeme, > > > > after converting an RGB-image (roman16) from eciRGB_v2 or AdobeRGB to my > > fogra39 profile, magenta and cyan primaries appear far to dirty. Here > are > > some values: > > > > eciRGB (Photoshop) > > Cyan: 61/138/194 -> 83/10/5/32 > > Magenta: 174/59/134 -> 0/85/9/29 > > > > That s at least 10-15% too much black in both colours. Since the > transition > > is very smooth, I really don t have any problem correcting them. Anyway, > my > > profile params are: > > -l330 -L95 -kp 0.00 0.18 0.98 0.95 0.55 -al -s AdobeRGB -np -tp -dpp > > Hello, Vladimir, I suppose this is because the black generation in Argyll > colprof/collink/xicclu is "wide", the black penetrates into the saturated > area until it conforms the black curve settings. > Take a look into > //www.freelists.org/post/argyllcms/colprof-and-proofing-purposes > > Anorher thing: Yo've done the mapping from AdobeRGB, but do the conversion > from eciRGB. IMHO it will be more correct to do the mapping from eci.RGB (-s > eciRGB) in this case. > >