-
Bug is related to “Improve precision” feature which in fact causes banding when inverse profile is used and nonlinearity is present. With it disabled, the inverse curves looks like monitor’s output and banding is not present, as well as gamut not being artificially clipped.
This means the optimizer is fine, just the rspl or xicclu messes up.
-
This reply was modified 5 years, 12 months ago by
AstralStorm.
Well, DisplayCAL is selecting too smooth curves increasing deltaE unnecessarily in highest saturation/brightness region, but that’s tunable with an extra command line option.
What the heck, using -qu for colprof and -qu -r64 for collink produced much better results. There’s a systemic problem – seems like response correction “misses” step nonlinearities, causing something like nearest neighbor choice. Excess resolution works around it very effectively.
Smoothing needs a fix badly.
Display Calibration and Characterization powered by ArgyllCMS