Alexander wrote:
The other day I attached a Argyll XYZ LUT profile to the bug, and made the comment trying to confirm if qcms in firefox really does support XYZ LUT profiles.
It would certainly be good for them to fix this.
In case anybody is unaware, this problem arose from Mozilla deciding it wanted to write a new color management system from scratch, QuickCMS (aka qcms). Unfortunately qcms is currently limited in features & profile support, not as accurate as lcms, and still a work-in-progress. The work-in-progress part is good in a way since eventually (if eventually is 10 days or 10 years who knows...), it's features and support should be flushed out. This of course depends on developers willing to write and commit patches in a timely fashion.
Seems like a hard road given there are already 3 or more other alternatives: my icclib, lcms and the ICC sample code. It's hard to whip up enthusiasm for re-inventing the wheel.
If you don't feel the need to use the latest Firefox 3.5, use Firefox 3.0 instead. Firefox 3.0 uses the MIT Open-Source project LittleCMS (a.k.a. lcms | http://www.littlecms.com/ ). lcms is a full featured cms system and it supports Argyll LAB and XYZ LUT profiles, ICCv4 profiles, as well as pretty much everything else under the sun, in a much more accurate manner then qcms. Mozilla claimed they had their reasons for ditching lcms (something about security?) but it made me a little annoyed when they did so.
It just seems like a rush of blood to the head. Their security excuse is bogus. Their performance goals could have been met by either improving lcms or adding their own pixel engine back end to lcms. If sanity ever prevails again they will switch back to lcms.
That said, Graeme, I still hope you would consider looking into adding matrix tags to the XYZ LUT profiles like other profiling applications do, if it really doesn't turn out to be too difficult.
See my previous post - it will be in 1.1.0 RC3. [The chances of Microsoft fixing the similar problem in Vista/Windows 7 is slim, since it's probably in old Heidelberg code, and their color team is almost but not quite non-existent now.] Graeme Gill.