- This topic has 9 replies, 4 voices, and was last updated 4 years, 7 months ago by
.
-
Posts
-
I would like to have accuracy as much possible. Is calibration speed : slow, patch 2681 too much? should I change? Diminishing returns?
Time to complete around 1hour 15minute.
I would like to have accuracy as much possible. Is calibration speed : slow, patch 2681 too much? should I change? Diminishing returns?
Time to complete around 1hour 15minute.
I would use the auto optimized patches for each profile, also for calibration speed you can use medium or go with slow.
I would use the auto optimized patches for each profile, also for calibration speed you can use medium or go with slow.
I also use auto optimized patches. But hot about amount of patches? For XYZ LUT 2681 seem to be ok, too much?
I would use the auto optimized patches for each profile, also for calibration speed you can use medium or go with slow.
I also use auto optimized patches. But hot about amount of patches? For XYZ LUT 2681 seem to be ok, too much?
If is it on auto optimized ok, if it is not, then don’t use it.
Calibration speed is for grey: medium vs slow, 48 vs 96 points / 256.
Profling patches is for mapping in a 3D mesh monitor behavior:
10 point x side in a 3d cube ~ 1000 patches (10x10x10 + additional grey patches for TRC)
12 point x side in a 3d cube ~ 1700 patches
17 point x cube side ~ 5000 patches
… etcI would use the auto optimized patches for each profile, also for calibration speed you can use medium or go with slow.
Calibration speed is for grey: medium vs slow, 48 vs 96 points / 256.
I see, so I use slow it is better then.
If is it on auto optimized ok, if it is not, then don’t use it.
Profling patches is for mapping in a 3D mesh monitor behavior:
10 point x side in a 3d cube ~ 1000 patches (10x10x10 + additional grey patches for TRC)
12 point x side in a 3d cube ~ 1700 patches
17 point x cube side ~ 5000 patches
… etcWell, it seem that the more the better then. But I read some there that say at some point it will be diminishing returns or waste of time. So maybe 2,000-3,000 is for best? I just don’t know/sure how to determine that point.
Cube resolution is cubic root of number of patches… so resolution does not grow as fast as you will want. Auto opt will find a good value as Simeonov said.
Depends on the display. For me doing a verification report with the extended test chart showed 175 patches producing the least errors, average and max, for one display. Another display showed least errors with 1553 patches.
Auto opt will find a good value as Simeonov said
Sorry I’m not get it. Froms what you both say it mean with setting auto opt, it doesn’t matter how much I set amount of patch 100 or 1000 or 2000 or any number, it will find good number on its own?
Depends on the display. For me doing a verification report with the extended test chart showed 175 patches producing the least errors, average and max, for one display. Another display showed least errors with 1553 patches.
Depends on the display. For me doing a verification report with the extended test chart showed 175 patches producing the least errors, average and max, for one display. Another display showed least errors with 1553 patches.
Aside from my question above. Amaze me, I think 1553 should be better, but it not. 175 give better this case how come?
Then how to know? Trial-error with less number like 175 then more number like 1553, 2000, 2700 then compare? Any rule of thumb?
Auto opt will find a good value as Simeonov said
Sorry I’m not get it. Froms what you both say it mean with setting auto opt, it doesn’t matter how much I set amount of patch 100 or 1000 or 2000 or any number, it will find good number on its own?
It will try to find best value.
Depends on the display. For me doing a verification report with the extended test chart showed 175 patches producing the least errors, average and max, for one display. Another display showed least errors with 1553 patches.
Depends on the display. For me doing a verification report with the extended test chart showed 175 patches producing the least errors, average and max, for one display. Another display showed least errors with 1553 patches.
Aside from my question above. Amaze me, I think 1553 should be better, but it not. 175 give better this case how come?
RTC errors (overshoot) and issues like that on an overall good display showing static patches.
Then how to know? Trial-error with less number like 175 then more number like 1553, 2000, 2700 then compare? Any rule of thumb?
Use auto.
Also most good displays can be mapped with a 1TRC + matrix. Using XYZLUT is for mapping irregular displays (PS, LR…) or for mapping good displays in a realistic way to get very accurate LUT3D.
If display is good behaved YOU DO WANT a simpler profile (1TRC + matrix) for Photoshop/Indesign/Illustrator because it doe snot matter how accurate profile is… image editor program is likely to suffere rounding errors in calculations (Indesign, illustrator… photoshop without 10bit OpenGL driver) sothis accuracy is ggoing to be wasted.If display is good behaved make a medium or slow single curve + matrix default 175 patches (to track TRC in greys in an accurate way). If it validates OK, use that for color managed programs (less rounding errors). If you need a LUT3D you can try a more accurate approach XYZLUT, using auto optimized.
