- This topic has 7 replies, 2 voices, and was last updated 3 years, 3 months ago by
.
-
Posts
-
Hello,
I’ve been trying to calibrate my eizo CG247x all weekend but i still see a slight yellow and cyan drift in a greyscale when appliying the lut in resolve (3d video monitor). I’m using a decklink mini monitor 4K and a i1 display plus for the calibration. I’ve been following these steps (see EIZOCG2420CCALIBTATION htlm attached), and while the colors seems to be pretty good on the report, the rgb gray balance is off (see report attached) and i don’t know what to do to fix it…
This is what i have set in displaycal DisplayCal Settings
Thanks.
Check first HW cal gray balance, so you can be sure that error is in LUT3D. Then if grey range is OK o CG (and it’s expected to be superb) check white point and profile TRCs
Hello Vincent,
I did some tests to see if the gray balance is an error of the 3DLut or not, hope this throw some light. I reset all and start from scratch. First i used color navigator to select my target (gamut native, 100cd/m2, 6500k, gamma 2.2, priority standard), profiled it with the internal probe of the monitor and save it to one of the internal CAL slots.
Once this was done, i opened Display Cal and set my correction to spectral LCD PSF WLED Adobe RGB HP DreamColor Z24x G2 and check if my whitepoint and white levels are as suposed. The whitepoint is off by almost a +1000K, so i bring it down to 6500k using color navigator and calibrating again with the internal probe. Once the whitepoint and levels are ok i do a measurement report to a simulation profile AdobeRGB (which is the native gamut of the panel). You can see in the attached report that the gray balance is pretty good.
Then i start the calibration, profiling and creation of the 3d Lut with a source colorspace set to rec709. Once it’s done i verify the lut and the report shows a worst gray balance and i am still being able to see a tint gradation when appliying the lut to a grey scale in resolve. We could say then that the 3DLut is the one that is creating this artifacts ? Is there a way to solve this ?
Thanks.
Once this was done, i opened Display Cal and set my correction to spectral LCD PSF WLED Adobe RGB HP DreamColor Z24x G2 and check if my whitepoint and white levels are as suposed. The whitepoint is off by almost a +1000K, so i bring it down to 6500k using color navigator and calibrating again with the internal probe.
Weird that is a big shift… Coloredges have internal engine based on factory calibration data for other than CAL presets. Check its white point.
Also test against CIE 2012 2 degree: dispcal -v -R -Q 2012_2 -X “path to WLED PFS HP.ccss”
Once the whitepoint and levels are ok i do a measurement report to a simulation profile AdobeRGB (which is the native gamut of the panel). You can see in the attached report that the gray balance is pretty good.
No. it is not. Native gamut is AdobeRGB + P3. Report should be done againts CN profile (it should be close to matrix + 1xTRC + BPC unless CN is configured bo something else), no simulation profiles at all. Hence the errors you see.
Then i start the calibration, profiling and creation of the 3d Lut with a source colorspace set to rec709. Once it’s done i verify the lut and the report shows a worst gray balance and i am still being able to see a tint gradation when appliying the lut to a grey scale in resolve. We could say then that the 3DLut is the one that is creating this artifacts ? Is there a way to solve this ?
Thanks.
Check TRC in DisplayCAL profile you use as LUT3D input, specially under grey RGB 100
Also test against CIE 2012 2 degree: dispcal -v -R -Q 2012_2 -X “path to WLED PFS HP.ccss”
I’ve calibrate against CIE 2012 2º, now against what i do my verification, simulation profile rec709 ?
Check TRC in DisplayCAL profile you use as LUT3D input, specially under grey RGB 100
I attached the tone response curve with and without LUT, where i can see the grey RGB 100 ?
It’s still the option “Evaluate gray balance through calibration only” available in Display Cal ? Where?
I attached a measurement report with a testchart for color accuracy and gray balance.
Thanks
Also test against CIE 2012 2 degree: dispcal -v -R -Q 2012_2 -X “path to WLED PFS HP.ccss”
I’ve calibrate against CIE 2012 2º, now against what i do my verification, simulation profile rec709 ?
I’ve said test, what you did it’s on your own. HTML report verifies with CIE 1931 2 degree. If want to test white against that observer do as instructed.
Check TRC in DisplayCAL profile you use as LUT3D input, specially under grey RGB 100
I attached the tone response curve with and without LUT, where i can see the grey RGB 100 ?
It’s still the option “Evaluate gray balance through calibration only” available in Display Cal ? Where?
I attached a measurement report with a testchart for color accuracy and gray balance.
Thanks
TRC seems fine, although that is a capture, not the profile.
To discard some ArgyllCMS issue use as LUT3D destination profile a 1 matrix + curve (BPC off?) instead of XYZLUT. If CG247 is well behaved it can pass verification. Also if you are sure that whitepoint is OK on HW, use relative colorimetric in LUT3D.If that works request in ArgyllCMS maillist or to DisplayCAL developer a “fix/hack” to embed equal TRC in XYZLUT profiles per user request so well behaved displays on greyscale won’t suffer potential rounding errors in LUT3D creation (or some other icc color managed apps that use gradients).
If want to test white against that observer do as instructed.
Attached Withepoint.jpg
profile a 1 matrix + curve (BPC off?)
I did it with BPC off and used relative colorimetric. I attached a verification chart report. Appliying this 3DLUT (single curve + matrix) i don’t see any perceptual tint in the greyscale. Shouldn’t the RGB Gray Balance be better for that kind of panel though ? Colors could be better with a xyz + matrix 3DLUT, right ?
What are your thoughts ?
Thanks.
Seems close to CIE 1931 2 degree.
Regarding your question, panel has superb greyscale. Range under 0.6 in 2nd report of this thread.
If using that trick works is because what I wrote in my previous message. Report it to Argyll & DisplayCAL developer: ask if it is possible (or just useful) in future versions to have XYZLUT single curve for well behaved displays on grayscale to have full detailed gamut correction but to avoid rounding error issues in computation with otherwise visually neutral colorless grayscale.
I would send them icc file, ti3, device information, correction… etc.
