- This topic has 3 replies, 2 voices, and was last updated 2 years, 7 months ago by
.
-
Posts
-
I’m trying to calibrate a Samsung U32J59, consumer monitor, but is 10-bit and covers above 100% sRGB. I’m using DisplayCal latest version on Windows 11.
After calibration, it looks correct, but is missing saturation compared to reference monitor. Without calibration, it does look incorrect, colors are too much saturated, but not so much, some people might like it. I do tend to like the more vibrant colors, but wheter it’s ok for me to consume films, is not good for mastering and delivering content to clients. My goal is to work with Video and Film production standards. Basic steps are to profile the display and apply it with DisplayCal Profile Loader, then generate a 3D LUT (with vcgt off) and apply on DaVinci Resolve for full color accuracy.
Calibration settings are:
Display and Instrument
- Meter: Datacolor Spyder X
- Mode: Wide LED (more accurate than Generic, at least from verification results)
- Correction: None
- Output levels: TV RGB 16-235 (this one I got wrong, my GPU and display seems to be receiving Full range signal. I thought that I should choose Limited to meet broadcast standard, is this correct? )
Calibration:
- Whitepoint: D65
- White level: 100 cd/m² (to meet standard)
- Black level: as measured (to guarantee maximum contrast)
- Tone curve: Rec 1886
- Gamma: 2.4
Profiling
- XYZ LUT + matrix
- 1545 patches
3D Lut
- Create 3D LUT after profiling box is activated
- Source colorspace: Rec 709
- Tone curve: Rec 1886
- Gamma: 2.4
- Apply calibration (vcgt) unchecked
- Input/Output encoding: Full range (got confused because doesn’t seem to have a default for this, but first I chose TV RGB on both, later I chose Full range and nothing changes visually)
DisplayCal Profile loader is working fine, browsers and Windows Photos app seems to be corrected, along with all color managed applications. For video, since I’m on Windows, ICC profile does nothing, so I have to use video player which accepts 3D LUT (Da Vinci, madVR filters or MPV working consistently).
While when viewing photos and browsing, everything looks fine, but when viewing video + 3D LUT, colors seems very desaturated. Playing same video on browser, it seems too much saturated, not ideal also, but more likeable (display seems to have “Wider” Gamut, but is still a regular sRGB). Testing with a single frame is almost always color managed, but when I see the picture without the correction, I have the same feeling, corrected picture seems to much desaturated.I have also a second monitor, regular Dell sRGB monitor, which is also calibrated the same way (except for the correction to the meter, there is one for my exact model) and on the second monitor, I don’t have the same feeling, and the changes are too subtle to even notice.
Playing with Full vs Video on the output levels settings at Display & Instrument tab is gonna be my next step, but I keep asking myself if my settings are good. It occurred to me that I should be profiling my display to be accurate to the standard used by PCs, sRGB with gamma 2.2, than on the 3D LUT tab I should generate a LUT configured for Rec 1886 with Gamma 2.4, but this might be a second unrelated issue, of matching the look of a video mastered on a Rec 1886 Gamma 2.4 calibrated OLED on a PC sRGB Gamma 2.2, not exactly cause for the calibrated result feeling less colored.
I’ve also considered that I’m too used to Youtube videos and Netflix being super colorful, then when I watch color managed videos it seems to lack saturation (my brain and my eyes are the culprit).
Another possibility would be the lack of a correction for the meter, in this case I could try to get an i1 Display Pro, as recommended, since it would be a more accurate meter than the Spyder X, then it wouldn’t need the correction for the display. Unfortunately, the i1 prices are insanely high compared to SpyderX in my country. But the correction do made some difference on the second monitor, it looked more neutral and correct for my eyes, without correction it looked slightly green tinted, but not as much as the issues I’m seeing on the main monitor, to the point that I can spot right on.
My question is actually a sanity check: am I doing something wrong?
Calibrite Display Pro HL on Amazon SpyderX Pro on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Display and Instrument
- Meter: Datacolor Spyder X
- Mode: Wide LED (more accurate than Generic, at least from verification results)
That “test” you made means nothing. It is not more accurate. If you use a wrong correction you’ll see it agaonst a reference device or visually in white.
Ypu’ll see nothing about a colorimeter correction reading a measurememnt report /verification (unless you have a true reference to compare).But if it is a QLED, Wide LED is teh closet one available on Datacolor devices.
- Correction: None
- Output levels: TV RGB 16-235 (this one I got wrong, my GPU and display seems to be receiving Full range signal. I thought that I should choose Limited to meet broadcast standard, is this correct? )
no. Source material and display are different things.
Calibration:
- Whitepoint: D65
- White level: 100 cd/m² (to meet standard)
- Black level: as measured (to guarantee maximum contrast)
- Tone curve: Rec 1886
- Gamma: 2.4
If you are using it as GUI aim for a general purpose calibration for desktop. Then aim for your desired TRC in LUT3D for each project you need (Rec709 g2.4)
Profiling
- XYZ LUT + matrix
- 1545 patches
3D Lut
- Create 3D LUT after profiling box is activated
- Source colorspace: Rec 709
- Tone curve: Rec 1886
- Gamma: 2.4
- Apply calibration (vcgt) unchecked
- Input/Output encoding: Full range (got confused because doesn’t seem to have a default for this, but first I chose TV RGB on both, later I chose Full range and nothing changes visually)
ok
DisplayCal Profile loader is working fine, browsers and Windows Photos app seems to be corrected, along with all color managed applications. For video, since I’m on Windows, ICC profile does nothing
Grey/Gamma, so better use a general prupose calibration like 2.2, then aim for wahetever you want in specific LUT3D.
, so I have to use video player which accepts 3D LUT (Da Vinci, madVR filters or MPV working consistently).
While when viewing photos and browsing, everything looks fine, but when viewing video + 3D LUT, colors seems very desaturated. Playing same video on browser, it seems too much saturated, not ideal also, but more likeable (display seems to have “Wider” Gamut, but is still a regular sRGB). Testing with a single frame is almost always color managed, but when I see the picture without the correction, I have the same feeling, corrected picture seems to much desaturated.Measure 100% saturation RGBCMY Rec709 Mp4 files in AVS forum. Likely to be a level mismatch issue. Full range.
I have also a second monitor, regular Dell sRGB monitor, which is also calibrated the same way (except for the correction to the meter, there is one for my exact model) and on the second monitor, I don’t have the same feeling, and the changes are too subtle to even notice.
Playing with Full vs Video on the output levels settings at Display & Instrument tab is gonna be my next step, but I keep asking myself if my settings are good. It occurred to me that I should be profiling my display to be accurate to the standard used by PCs, sRGB with gamma 2.2, than on the 3D LUT tab I should generate a LUT configured for Rec 1886 with Gamma 2.4, but this might be a second unrelated issue, of matching the look of a video mastered on a Rec 1886 Gamma 2.4 calibrated OLED on a PC sRGB Gamma 2.2, not exactly cause for the calibrated result feeling less colored.
Since Rec1886 and g2.4 black output offset ar enot the same, even on a 2000:1 VA, test gamma tracking.
I’ve also considered that I’m too used to Youtube videos and Netflix being super colorful, then when I watch color managed videos it seems to lack saturation (my brain and my eyes are the culprit).
Another possibility would be the lack of a correction for the meter, in this case I could try to get an i1 Display Pro, as recommended, since it would be a more accurate meter than the Spyder X, then it wouldn’t need the correction for the display. Unfortunately, the i1 prices are insanely high compared to SpyderX in my country. But the correction do made some difference on the second monitor, it looked more neutral and correct for my eyes, without correction it looked slightly green tinted, but not as much as the issues I’m seeing on the main monitor, to the point that I can spot right on.
My question is actually a sanity check: am I doing something wrong?
Levels, Rec1886 vs g2.4
Also remember that by default (AFAIK) madVR clears VCGT data so madVR LUT3D shoudl have embed VCGT, while Resolve LUT3D ****on a GUI display used for other things*** shoul not embed VCGT and let DIsplayCAL mantain system white grey calibration.
In that example Resolve LUT3D will reencode Rec709 g2.4 to display colorspace, expecting that such dusplay behaviro is mantained by something else (DisplayCAL loader).Also when in doubt… MEASURE!
Thank you for your input, it was very helpful. It’s starting to make more sense.
That “test” you made means nothing. It is not more accurate. If you use a wrong correction you’ll see it agaonst a reference device or visually in white.
Ypu’ll see nothing about a colorimeter correction reading a measurememnt report /verification (unless you have a true reference to compare).But if it is a QLED, Wide LED is teh closet one available on Datacolor devices.
Correct me if I’m wrong, the verification that takes place after profiling the display only tells me that meter was able to read correct values after calibration was applied, so you can know that your colorimeter is at least consistent with itself, is that it? So verification of ICC profile against it’s own accuracy doesn’t mean nothing, it’s always ok unless I change picture color profile in monitor directly, or if I place a filter in front of colorimeter, or anything that I should not be doing anyway?
And what about verification of 3D LUT through Resolve pattern generator? I’ve managed to set up a test to verify applied 3D LUT against standardized color space, and Delta E values are good and work as expected when using Wide LED, and higher and not ok using Standard LED.
According to Spyder X software (screenshot attached) it seems General and GB LED are definitely not my monitor, backlight it’s WLED. Standard LED would be ok, but Samsung U32J59 has a wider than sRGB gamut coverage, and it’s 95% P3, so Wide LED makes more sense to be used. Reading about Spyder X increased my confidence since a reviewer wrote that Wide LED was giving better results, regardless the screen type, link follows: https://www.chromapure.com/newgear-new.asp
I’ve also calibrated a OLED CX using Calman Home for LG and Spyder X, and results with Wide LED mode had lower Delta E values in final verification than Standard. Factory Calibration on Filmmaker Mode was closer to calibration done with Wide LED colorimeter mode, Standard mode gives a much different result, blacks and shadows are crushing. Don’t know why this happens, can I measure/verify OLED screen with Spyder X using DisplayCal and DaVinci Resolve pattern generator? OLED signal is coming from Decklink card.If you are using it as GUI aim for a general purpose calibration for desktop. Then aim for your desired TRC in LUT3D for each project you need (Rec709 g2.4)
So it should be sRGB or Gamma 2.2? Rec 1886 is basically Rec 709 with Gamma 2.4 (BT 1886 gamma curve, not power curve), so not for me, since I’m not calibrating a TV. Rec 709 is misleading, don’t know why is that for, you can’t choose gamma. With sRGB I also cannot set gamma to 2.2, but with Gamma 2.2 I can.
This Tone Curve setting is very confusing, DisplayCal is not using same names as other professional software.Measure 100% saturation RGBCMY Rec709 Mp4 files in AVS forum. Likely to be a level mismatch issue. Full range.
Can’t find those test files at AVS. Would you point me the way? In that case, how can I measure using colorimeter? Click Calibrate & Profile and only use Interactive Display Adjustment part, and take notes on R, G and B values?
test gamma tracking.
How can a make a Gamma Tracking test? Should I use Calman?
Since Rec1886 and g2.4 black output offset ar enot the same, even on a 2000:1 VA,
Didn’t understand the first part. Never understanded what black output offset means.
I’ve thought that Rec 1886 implies you are using Gamma 2.4 (not a pure power curve, but a special BT 1886 one). Calman Home for LG states that if using OLED, one should use 2.4 Power, and when calibrating a LG LED (non-OLED), one should use BT 1886 as gamma. Color Primaries always the same for SDR, Rec 709. Calling it Rec 1886 is confusing. DaVinci, Premiere, Final Cut Pro and After Effects doesn’t even mention it.Also remember that by default (AFAIK) madVR clears VCGT data so madVR LUT3D shoudl have embed VCGT, while Resolve LUT3D ****on a GUI display used for other things*** shoul not embed VCGT and let DIsplayCAL mantain system white grey calibration.
In that example Resolve LUT3D will reencode Rec709 g2.4 to display colorspace, expecting that such dusplay behaviro is mantained by something else (DisplayCAL loader).I’ve take notice of the madVR LUT 3D having then VCGT embed. DaVinci, without VCGT, matches exactly madVR with VCGT embed, and also matches exactly what MPV outputs, but using the system installed ICC color profile instead.
Final question: after getting all those stuff right, calibration is done, should the picture I see at my Samsung monitor be matching the same picture viewed on OLED CX on grading environment? Or is it a fact that OLED image is always more vibrant and has more contrast than VA, and it would only match another OLED panel?
Thank you for your input, it was very helpful. It’s starting to make more sense.
That “test” you made means nothing. It is not more accurate. If you use a wrong correction you’ll see it agaonst a reference device or visually in white.
Ypu’ll see nothing about a colorimeter correction reading a measurememnt report /verification (unless you have a true reference to compare).But if it is a QLED, Wide LED is teh closet one available on Datacolor devices.
Correct me if I’m wrong, the verification that takes place after profiling the display only tells me that meter was able to read correct values after calibration was applied, so you can know that your colorimeter is at least consistent with itself, is that it? So verification of ICC profile against it’s own accuracy doesn’t mean nothing, it’s always ok unless I change picture color profile in monitor directly, or if I place a filter in front of colorimeter, or anything that I should not be doing anyway?
Display profile could be highly idealized (matrix+1 TRC) so verification tells if profile (display description, that cpuld be idealized) matches display behavior.
Display behavior recorded in a ti3 file could be a little different from predicted bevahior of display profile.And what about verification of 3D LUT through Resolve pattern generator? I’ve managed to set up a test to verify applied 3D LUT against standardized color space, and Delta E values are good and work as expected when using Wide LED, and higher and not ok using Standard LED.
If you change measurement mode/correction across pipeline device will measure in a diferent way.
According to Spyder X software (screenshot attached) it seems General and GB LED are definitely not my monitor, backlight it’s WLED.
It is not a White LED, it could be something else.
Standard LED would be ok, but Samsung U32J59 has a wider than sRGB gamut coverage, and it’s 95% P3, so Wide LED makes more sense to be used.
As said before it is the closest you have in built in modes, “it’s all you have with spyderx built in modes”
Reading about Spyder X increased my confidence since a reviewer wrote that Wide LED was giving better results, regardless the screen type, link follows: https://www.chromapure.com/newgear-new.asp
That web has nothing to do with this subject. Spyder X built in modes are something simular to matrix corrections, only good for the device and the display measured, hence its generic modes will have some kind of error exposed to othe backlights. There are several “wide gamut led” right now in the market even if we exclude GB-LED.
“Wide LED” its all you have to measure all these different bakclights, while i1d3 CCSS could fit better newer or particular backlights.
I’ve also calibrated a OLED CX using Calman Home for LG and Spyder X, and results with Wide LED mode had lower Delta E values in final verification than Standard. Factory Calibration on Filmmaker Mode was closer to calibration done with Wide LED colorimeter mode, Standard mode gives a much different result, blacks and shadows are crushing. Don’t know why this happens, can I measure/verify OLED screen with Spyder X using DisplayCal and DaVinci Resolve pattern generator? OLED signal is coming from Decklink card.
Same as above. It seems that you fail to understand how SpyderX buit in modes work. But as I said above “Wide LED” is all that you have, so use it if you do not have access to a better probe.
It is not your displays backlight, but that built in mode is all taht you have.
If you are using it as GUI aim for a general purpose calibration for desktop. Then aim for your desired TRC in LUT3D for each project you need (Rec709 g2.4)
So it should be sRGB or Gamma 2.2? Rec 1886 is basically Rec 709 with Gamma 2.4 (BT 1886 gamma curve, not power curve), so not for me, since I’m not calibrating a TV. Rec 709 is misleading, don’t know why is that for, you can’t choose gamma. With sRGB I also cannot set gamma to 2.2, but with Gamma 2.2 I can.
This Tone Curve setting is very confusing, DisplayCal is not using same names as other professional software.Resolve has its own color management, a LUT3D.
Before you said you were using a decklink in this quoted message I thought i was a general desktop display that could be used as GUI. So with that in mind read again what I wrote.
-Hence on that situation, use as display calibration a common TRC, then make a LUT3D to whatever you want. 2 step chain: desktop general use, then from rec709 content to desktop display.
-If you want to use it with a specialized card as a reference display in resolve, this does not apply. You measure display behavior (once corrected to D65 and 100nit) then create a LUT3D. 1 step. Whatever you set as calibration TRC for profile making will be chanied to whatever you put on LUT3D creation, and that chained data computation, the final result, is the only one being applied. Hence you can put as calibration TRC whatever you want, but its’ easier to choose a gamma setting closer to display native behavior so VCGT is used anly for grey color. Then let LUT3D computation handle final gamma value.In displyaCAL profile & LUT3D creation you can see these two separate steps: display TRC + optional calibration , then LUT3D TRC change.
Measure 100% saturation RGBCMY Rec709 Mp4 files in AVS forum. Likely to be a level mismatch issue. Full range.
Can’t find those test files at AVS. Would you point me the way?
Years ago
https://www.avsforum.com/forums/display-calibration.139/
Try searching Rec709 test pattern.In that case, how can I measure using colorimeter?
If you want to measure detached to pattern rendering, to measure without an app setting a pattern and the same app measureing:
-HCFR (manual DVD mode, or something like that, not auto modo on “new file”)
-ArgyllCMS commandline spot readhttps://www.argyllcms.com/doc/spotread.html
Click Calibrate & Profile and only use Interactive Display Adjustment part, and take notes on R, G and B values?
This will give you only whitepoint.
test gamma tracking.
How can a make a Gamma Tracking test?
Measurement report (+ various options, read the forum, it has been explained tons of times)
Should I use Calman?
If you want, I won’t pay a penny for it unless I had a TV with HW calibration supported by it. In that situation is useful.
Since Rec1886 and g2.4 black output offset ar enot the same, even on a 2000:1 VA,
Didn’t understand the first part. Never understanded what black output offset means.
I’ve thought that Rec 1886 implies you are using Gamma 2.4 (not a pure power curve, but a special BT 1886 one). Calman Home for LG states that if using OLED, one should use 2.4 Power, and when calibrating a LG LED (non-OLED), one should use BT 1886 as gamma. Color Primaries always the same for SDR, Rec 709. Calling it Rec 1886 is confusing. DaVinci, Premiere, Final Cut Pro and After Effects doesn’t even mention it.Read Rec1886 definition, actual TRC values varies with display black. It’s in google.
DIsplayCAL setting, once you choose 2.2 or 2.4 is about ho to handle TRC where your finite contrast display cannot met that power law gamma, so it tries to “bend” (Y = TRC, X: input 2D graph) when closer to black.
On a 2000:1 display 2.4 and Rec1886 may differ. On a 1000:1 display 2.4 relative black output offset and rec1886 will be very different, by definition of rec1886.
Also remember that by default (AFAIK) madVR clears VCGT data so madVR LUT3D shoudl have embed VCGT, while Resolve LUT3D ****on a GUI display used for other things*** shoul not embed VCGT and let DIsplayCAL mantain system white grey calibration.
In that example Resolve LUT3D will reencode Rec709 g2.4 to display colorspace, expecting that such dusplay behaviro is mantained by something else (DisplayCAL loader).I’ve take notice of the madVR LUT 3D having then VCGT embed. DaVinci, without VCGT, matches exactly madVR with VCGT embed, and also matches exactly what MPV outputs, but using the system installed ICC color profile instead.
Read the “chain” explained above. Resolve used as GUI on a general purpose GPU has 2 chains. madVR on a general purpose GPU by default has one chain. They may be equal or not depending on your display behavior with no grey calibration appiled.
Final question: after getting all those stuff right, calibration is done, should the picture I see at my Samsung monitor be matching the same picture viewed on OLED CX on grading environment?
There may be a metameric failure on whiteopoint but you can match them visiaully (butthat implies not usng abs colorietric LUT3D, explanied in other threads)
Let’s assume that ther is no such difefrence and whirepoint looks equal or almost equal. As long as content can be show in display colorspaces , yes… untill you get near black. Your VA 2000:1 won’t go further than 100/2000 (aprox.) and near black it has to “bend” TRC or it will clip.
Or is it a fact that OLED image is always more vibrant and has more contrast than VA, and it would only match another OLED panel?
Read above.
