-
I found out that matrix profiles are fine for my application and use the instead of LUT profiles for faster refreshing.
I came to conclusion that it is most effective to make a testchart from:
– 256 neutral patches
– several white patches to correct white point drift
– several iterative patches
– several single-channel patches
This should be fine as long as subpixels of LCD are not influencing adjacent subpixels.
I have done profiling and verification (with no simulation) and results are fine – except that the maximum error is detected on (198,198,198) (delta=3) and (0,0,255) (delta=4) both of which are obviously included in profiling chart.
What can go wrong? Why can there be so big error on samples included in chart?
I have redone profiling several times, results are almost the same – mostly fine with with spikes ay blue and gray patches.
Does it mean that there is indeed a non-linearity in LCDs which cannot be expressed with curves and matrix? Does it mean that I should use LUT?
Does it mean that there is indeed a non-linearity in LCDs which cannot be expressed with curves and matrix? Does it mean that I should use LUT?
Possibly, this depends on the particular display.
Display Calibration and Characterization powered by ArgyllCMS