Incorrect color presentation with Studio Driver 595.79

[theSLY]

Incorrect color presentation with the latest Studio driver 595.79 with Asus 5070, while with the old Studio driver 581.57 works fine. Using msovideo_srgb in SDR mode with ICC profile generated with DisplayCAL + Calibrite Display Plus. Target is sRGB, Calibrate gamma to sRGB is checked. Colors are look reddish with 595.79

[guzenco]

Does the reddishness appear only after enabling clamp? Does it persist after disabling clamp? Does it affect white (255, 255, 255)?

Issue sounds similar to what people encountered with R590 drivers after using novideo_srgb. It seems that latest drivers, under unknown conditions, sometimes apply additional unknown color transformation.

From what I know, this is not likely to help, but you can try removing ColorspaceConfig and ColorSpaceLutRegistryKey from the display entries in Computer\HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\nvlddmkm\State\DisplayDatabase.
Then press Win+Ctrl+Shift+B to clear some known staff that can affect colors.

You can also try using DDU to clear unknown staff, though I’m not sure that wouldn’t return after reinstalling the driver.

If this doesn’t help, ACM should work normally. The tool is fully compatible with ACM, so for an sRGB target color space there shouldn’t be any difference, except that ACM may remove the reddishness for you.

[hadoooooouken]

I had similar behavior with novideo_srgb + 590, where some shades of gray would turn red. Btw, novideo_srgb still doesn’t work on 595.79, even though there were some promises from an nvidia representative on a forum.

The situation with 595.79 + msovideo is a bit different - many shades appear darker than they should. It’s especially noticeable with red - it now looks more like a pomegranate color.

[guzenco]

many shades appear darker than they should. It’s especially noticeable with red - it now looks more like a pomegranate color.

Your situation sounds like double desaturation that can come from a novideo_srgb clamp. Try removing ColorspaceConfig and ColorSpaceLutRegistryKey from the display entries in Computer\HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\nvlddmkm\State\DisplayDatabase. Then press "Win+Ctrl+Shift+B" to fully clear the novideo_srgb clamp that can persist through driver update.

If this doesn’t help, it will be useful to know where the problem appears with ACM (with R595 drivers there are no longer limitations for ACM availability), and whether it appears in R580 drivers when using msovideo_srgb.

[hadoooooouken]

Try removing

I ran DDU before installing the new driver, and I also checked the DisplayDatabase - it was clean.

ACM

I'm on Win10.

[guzenco]

I'm on Win10.

I wouldn’t be surprised if what you see is the result of Nvidia trying to create an ACM alternative for Windows 10. You can try using the target "Native", and maybe you will receive the sRGB color space as a result.

If distortions appear independently of the msovideo_srgb clamp, you can take measurements of them and use an ICC profile that includes those distortions to correct them. Alternatively, you can take measurements with the target set to “Native” and with "Use EDID primaries" selectet. (if that is enough to cause distortions)

[hadoooooouken]

I decided to test the 595.79 drivers again today and noticed something interesting:

If I set any resolution lower than my native one (for example, 1920×1080 or 1920×1440), the clamp (mhc2 and other icc) works perfectly and doesn’t reset. But as soon as I switch to my native 2560×1440, the switching effect becomes visible and the colors become darker.
upd. Nope. After about 20–30 seconds, some trigger activates at other resolutions as well, and the colors also become darker

The idea to try changing resolutions came from this note:
https://www.nvidia.com/en-eu/drivers/details/260362/

Display color seems faded after switching to non-native resolution [5548662]

But for some reason, in my case it worked the other way around.
I’ll try to submit a bug report to them, just in case.

Upd. Studio drivers:
595.79 - wrong colors;
591.86 (game ready) - wrong colors;
591.74 - wrong colors;
591.44 - good colors but very noticable banding is everywhere with any combination;
581.57 - good colors without noticable banding;
It's sad to be an nvidia owner these days..

[theSLY]

Does the reddishness appear only after enabling clamp? Does it persist after disabling clamp? Does it affect white (255, 255, 255)?

Sir, I mistook the oversaturated darkened colors for reddishness. In red shades this is most noticeable for me.

I am using a fresh updated installation of Windows 10 LTSC with nVidia Studio drivers 595.79. msovideo_srgb v1.2.11. The settings are below. When I turn on and then off HDR mode, the image becomes correct for about 30 seconds and then the darkened colors appears again.

It does not matter whether the monitor is connected with an HDMI or DisplayPort cable. It also doesn’t matter what color depth mode is set in the nVidia panel settings.

[guzenco]

It sounds like a problem on the driver side, most likely the result of a combination of Windows 10 and R590 or later Nvidia drivers.

Aside from the obvious potential fixes, such as upgrading to Windows 11 or using pre-R590 Nvidia drivers, you can try measuring your displays with artifacts. If they appear only with the MHC2 profile, you can set the targets to "Native" and, with "Use EDID primaries" selected. Then use the created profile in msovideo_srgb so it can try to correct those artifacts, though I’m not sure it will help given the nature of the artifacts (oversaturation in dark colors) and the way MHC2 works.

[deniscfd-cmyk]

can confirm, the darker color exist even with Windows 11. When ACM active the colors is brighter.
I can see difference even when choise Native in Target, when ACM inactive