Alastair M. Robinson wrote:
/path/to/1.2.0/bin/colprof -cmd -dpp -S <srgb> 120_md /path/to/1.2.0/bin/colprof -cmt -dpp -S <srgb> 120_mt
Note the minimal difference between 120mt and 120md - and that 120md is marginally the darker of the two.
Hi, I'm not sure precisely what the explanation is, but some experiments using collink -ir rather than colprof perceptual lead me to (handwavingly) explain it this way: The purpose of the perceptual or saturation gamut transform is to basically morph the source gamut into the destination gamut. Changing the viewing conditions changes the shape of the gamut, but a perfectly operating gamut mapping will simply adapt to the different shape and map to the same destination gamut. ie. the gamut mapping tends to cancel out changes introduced by the change in the viewing conditions. My speculation would therefore be that V1.2.0 shows less difference because the gamut mapping is more accurate than V1.1.1 :-). Of course any changes within the source gamut (rather than changes affecting it's shape or volume) introduced by differing viewing conditions are probably not counteracted by the gamut mapping, so some changes will still be evident. [The collink -ir experiment confirmed that there is no change at all in the CIECAM02 behaviour between V1.1.1 and V1.2.0. The difference you notice must therefore be a property of the change in gamut mapping. ]
Apart from that the new mapping is impressive, and looks like it'll be well worth the vastly increased calculation time!
Yes, the performance change is unfortunate, but my priority was to "do it right", and worry about performance another time. Graeme Gill.