Vallo Kallaste wrote:
Now this is getting hairy. I've converted bunch of raw's and saved them as 16bpc TIFF's with embedded "Chrome 2000 D50.icm" profile. Are you saying that I should create one monitor profile for every source profile I possibly could come accross? In current case use "Chrome 2000 D50.icm" profile as the source one, as argument for the -S switch? I don't actually understand what the following constraint in colprof manual means:
Yep, if you want actual gamut mapping using ICC device profiles, that's what you have to do. It's the basic nature of them that they have static gamut mappings cooked into their B2A tables. [Not very convenient eh ? That's why WCS went for dynamic linking, although ICC V4 with the PRMG can partially solve the problem too.]
/--/ Note that a input, output, display or device colororspace profile should be specified, not a non-device colorspace, device link, abstract or named color profile. /--/
What kind of profile the "Chrome 2000 D50.icm" actually is? In the raw converter sense it's output profile.
Without examining the profile it's hard to say. (ie. use iccdump on it).
What about that bwtest.jpg image which has embedded "Gray Gamma 2.2" profile? Should I extract the embedded profile (or convert the file to TIFF leaving the profile intact) and use that profile (or TIFF) as argument for -S switch while creating cLUT profile?
I suspect that using a monochrome profile will not succeed, because I probably haven't made provision for that special case. [It wouldn't be reasonable to force such a display profile to convert everything to monochrome, but a monochrome source profile doesn't define a 3D gamut, so some fudge would be needed to partially gamut map.]
I think that I understand the need for PCS space, A2B/B2A tables and different mapping intents, but seems I'm not able to map that understanding back to real life. I'm not rocket scientist, only lone UNIX systems admin trying to get hang of it :)
There are a few basic concepts to get a grasp of. What a gamut is, and how the different intents typically work, and the limitations of the different ICC profile formats (matrix vs. cLUT) seems to be the next step. Graeme Gill.