On Mar 29, 2016, at 9:57 PM, Graeme Gill <graeme2@argyllcms.com> wrote:
i have found that colorsync may fail to match a color it reported out-of-gamut, i.e. transforms the coordinates, but the transformed color remains out-of-gamut.
Practical gamut checks are typically not very precise. The ICC profile has a gamut tag, but it is a notoriously inaccurate representation, and seems to be rarely used.
Many systems use a "round trip check" instead. Take the PCS color, convert to device space via the B2A table and back to PCS via the A2B. If the delta E is over a threshold, then regard it as out of gamut. The use of B2A tables and an arbitrary threshold makes the check imprecise.
If ColorSync is either using the gamut tag or doing a round trip check, then it would be no surprise for some colors to be regarded as "out of gamut", even after converting it to device space and back to PCS, since B2A tables have a degree of inaccuracy, particularly near the gamut boundary, which could lead to a noticeable round trip delta E.
Doing a precise gamut check is more work. The ArgyllCMS xicclu tool for instance will do a precise gamut check when doing an "-fif" lookup (inverse A2B table lookup). This is "accurate" in the sense that it determines whether the color is in the A2B table somewhere to a high numerical precision. A subtlety for > 4 color profiles is that a color may not be reproducible in the B2A table, even though it is present in the A2B table, due to smoothness/interpolation continuity constraints. (For the purposes of the discussion I've also ignored any consideration of ink limits.)
thanks for your detailed reply, & making us aware of gamut-check w/ xicclu . the scenario i mentioned occurs in the context of user specified/edited synthetic color, & this incurs frequent gamut-check calls from a ui control; but i’ll [try to] set up a batch test on sample data for evaluation of xicclu ‘-fif’, & look forward to seeing the results.
Graeme Gill. _______________________________________________ Do not post admin requests to the list. They will be ignored. Colorsync-users mailing list (Colorsync-users@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/colorsync-users/etaffel%40me.com
This email sent to etaffel@me.com