Hey,
As a follow up to generating my own CCSS for my i1d3 using a ColorMunki
Photo as reference.
The following are 100 patch dispreads, checked using colprof against the
ColorMunki Photo as reference:
dispread -v -yl -P 0.5,0.5,1.5 new_i1d3-lcd
Verify results:
Total errors: peak = 63.301272, avg = 33.768038
Worst 10% errors: peak = 63.301272, avg = 50.042637
Best 90% errors: peak = 45.389789, avg = 31.959749
avg err X 7.286414, Y 6.797872, Z 7.689819
avg err L* 22.816882, a* 17.109658, b* 13.253409
dispread -v -yL -P 0.5,0.5,1.5 new_i1d3-wgccfl # should
reference X-Rite's WGCCFLFamily_07Feb11.ccss
Verify results:
Total errors: peak = 57.202462, avg = 32.909617
Worst 10% errors: peak = 57.202462, avg = 48.251966
Best 90% errors: peak = 42.492927, avg = 31.204912
avg err X 6.551847, Y 6.620577, Z 7.539831
avg err L* 21.971580, a* 16.222278, b* 13.272698
dispread -v -X ~/.local/share/ArgyllCMS/i1d3_hp_lp2475w.ccss -P
0.5,0.5,1.5 new_i1d3-wgccflcustomccss
Verify results:
Total errors: peak = 56.414749, avg = 32.760133
Worst 10% errors: peak = 56.414749, avg = 48.361151
Best 90% errors: peak = 42.811007, avg = 31.026686
avg err X 6.582835, Y 6.711358, Z 7.400585
avg err L* 22.245215, a* 15.927781, b* 12.821506
dispread -v -yl -X ~/.local/share/ArgyllCMS/i1d3_hp_lp2475w.ccmx -P
0.5,0.5,1.5 new_i1d3-wgccflcustomccmx
Verify results:
Total errors: peak = 11.156676, avg = 4.562927
Worst 10% errors: peak = 11.156676, avg = 8.095053
Best 90% errors: peak = 6.792108, avg = 4.170468
avg err X 0.782303, Y 0.520996, Z 0.469771
avg err L* 1.821452, a* 3.712087, b* 1.207165
BTW, variance between measurement seems to be along the lines of (this is
between the same -yl dispread, on two different days, where the display was
turned off for the night, but in both cases has had a long time to warm
up/stabilize):
Verify results:
Total errors: peak = 6.518196, avg = 2.895045
Worst 10% errors: peak = 6.518196, avg = 4.708550
Best 90% errors: peak = 4.045933, avg = 2.693544
avg err X 0.507973, Y 0.465681, Z 0.645914
avg err L* 1.820157, a* 1.415749, b* 1.367592
What's notable is that applying the CCMX increases the i1d3's accuracy
significantly.
However, applying both X-Rite's WG CCFL CCSS and my own custom WG CCFL
CCSS, seems to deliver results that are more or less within the measurement
variance.
Given that CCSSes should be superior, this not what I expected. Is there
any chance the CCSSes are being ignored?
Possibly consider explicitly printf'ing which CCMX/CCSS is being used when
in verbose mode in a future version of ArgyllCMS.
Regards,
Pascal de Bruijn