- This topic has 22 replies, 4 voices, and was last updated 4 months, 4 weeks ago by
.
-
Posts
-
I am new to display calibration and calibrated a few monitors without problems but my Samsung GQ55Q70T QLED TV gets weird results.
I used a newly bought Calibrite Display SL for the measurements.My settings were standard with the quantum dot correction, 256 neutral patches and 2×2 spatial static dithering with novideo_srgb.
My TV settings were also standard with gamma to 2.2 and whitepoint correction.
I couldn’t disable the light sensor but the luminance test didn’t show variations so I guess it’s ok.The first measurement was with room lights off and 32 cd/m^2.
It failed because candela were low.The second measurement was with 2700K room lights on (I wanted to avoid this because I usually watch with lights off to avoid reflections and there was a little 2mm gap between TV and colorimeter)
It succeeded but the gamut coverage result was not even sRGB in any color which is not true because I tested the colors after that with paint.net.
The result itself didn’t look that different to my other monitors.I have no clue what I did wrong.
Usecase for the TV is Netflix, Prime Video etc. with Edge browser only.Calibrite Display SL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.[…] with novideo_srgb.
It succeeded but the gamut coverage result was not even sRGB in any color which is not true because I tested the colors after that with paint.net.
I have no clue what I did wrong.
Using novideo_sRGB in a wrong way. You provide no data but likely to be using TV on rec709/srgb setting while using novideosrgb based on EDID data which desaturates from EDID/native coordinates.
Sry, I didn’t clarify that but I was only ticking the dithering option in novideo_srgb and did not tick clamp.
I don’t exactly know if that activates dithering but it is mentioned in the novideo_srgb readme.
After the calibration I used the clamp option with gamma set to srgb and target to P3.
If I am wrong, measure R, G, B primaries with spotread and MS paint with and without novideosrgb. Also show us data: profile, etc.
Sure, thanks for your help.
I uploaded the TV EDID and my created ICC-Profile.
If anything is missing just tell me, but I will not be at home until Saturday to test.novideo_srgb clamp active
White
Result is XYZ: 117.069456 127.221800 136.994864, D50 Lab: 109.693215 -8.370391 -20.132100
R
Result is XYZ: 46.804511 22.779308 0.715152, D50 Lab: 54.844286 87.591951 81.057161
G
Result is XYZ: 38.304764 89.529213 9.164914, D50 Lab: 95.801143 -114.338756 96.613074
B
Result is XYZ: 22.879501 8.804290 117.867171, D50 Lab: 35.604722 87.114835 -136.290917clamp inactive
White
Result is XYZ: 116.945985 127.039050 137.163121, D50 Lab: 109.633001 -8.298442 -20.332841
R
Result is XYZ: 46.713796 22.741666 0.716176, D50 Lab: 54.805243 87.506208 80.970521
G
Result is XYZ: 38.237896 89.381273 9.162386, D50 Lab: 95.739529 -114.287186 96.515682
B
Result is XYZ: 22.072231 7.442898 115.970741, D50 Lab: 32.794628 95.542968 -139.921233Yep, its weird.
Can you disable all calibration, ensure matching levels to full and use “spotread -x -X file.ccss” to measure Red 255, then green 255 then blue 255 to spot if the primaries are OK?Here are the results:
With TV whitepoint calibration
R
Result is XYZ: 50.094563 23.344210 1.437665, Yxy: 23.344210 0.669030 0.311770
G
Result is XYZ: 39.030683 90.179815 8.214269, Yxy: 90.179815 0.284015 0.656212
B
Result is XYZ: 22.443863 7.385061 118.109000, Yxy: 7.385061 0.151711 0.049920Without TV whitepoint calibration
R
Result is XYZ: 43.392648 20.132450 1.551515, Yxy: 20.132450 0.666793 0.309365
G
Result is XYZ: 34.916561 82.471167 8.369923, Yxy: 82.471167 0.277650 0.655794
B
Result is XYZ: 33.372694 10.369264 175.814046, Yxy: 10.369264 0.152001 0.047228Can you try a single curve + matrix calibratyion + profiling? I mean; can your coverage issues be caused by an irregular volume (although the fast XYZprofile does not point to it)
Also in a*b* vs CIE xy gamut plot is not showing “typical” behavior wghen going t a*b*. If I plot 3D intersection with DIsplaYCAL profile info tool the small % coverage seems to be cause by too small blue luminance vs your reference colorspace (sRGB), hence all 3d volume from blue to C-M-W is limited vs sRGB.
Did you lower too much blue control on TV so TV is doing something weird?
Turning all other elements off (novidoe_sRGB and such) and just cal & profile to a 1curve + matrix, does it show this lower luminance of B in a 3D plot?
The small coverage seems to be caused by that, the above suggested test are for finding a way to avoid it. You may test with alternative dsylight but cooler whites, just to test if TV OSD controls when you push RGB gain too low starts to act in a weird way.
But… there is the possibility that blue led on TV is too violet so even if CIE xy seems to cover properly sRGB to a very high % the 3D volume not and will never be able to do so. IDNK how much you tweaked TV controls, hence the need of testing,I did some extensive testing and am more confused than before.
The 1Curve+Matrix profile is turning back believable results in terms of gamut coverage.
I uploaded it, if you wanna see it.
After that I did a Curves+Matrix calibration with the same test chart (small testchart for matrix profiles).
The result was the same.
After that I did again a Curves+Matrix cal but with the default testchart.
Now I got a much worse result back.
Then I did the exact same cal again.
The result was even more worse, but the difference wasn’t that huge.The one thing that all calibrations share is that the visual result is the same (also my very first XYZ LUT + Matrix one).
They look OK but with a very strange gamma shift like 1.8 instead of 2.2.
But also not really like it should be if calibrated for 1.8.The really weird thing that happens during the calibrtion itself is that the darker patches are that dark that I can’t even see them myself.
That is also the reason for a much longer (like nearly twice) cal time for the TV.Also in a*b* vs CIE xy gamut plot is not showing “typical” behavior wghen going t a*b*. If I plot 3D intersection with DIsplaYCAL profile info tool the small % coverage seems to be cause by too small blue luminance vs your reference colorspace (sRGB), hence all 3d volume from blue to C-M-W is limited vs sRGB.
Did you lower too much blue control on TV so TV is doing something weird?
Turning all other elements off (novidoe_sRGB and such) and just cal & profile to a 1curve + matrix, does it show this lower luminance of B in a 3D plot?
The small coverage seems to be caused by that, the above suggested test are for finding a way to avoid it. You may test with alternative dsylight but cooler whites, just to test if TV OSD controls when you push RGB gain too low starts to act in a weird way.
But… there is the possibility that blue led on TV is too violet so even if CIE xy seems to cover properly sRGB to a very high % the 3D volume not and will never be able to do so. IDNK how much you tweaked TV controls, hence the need of testing,this
Yeah, you’re right.
It’s only the whitepoint correction that causes this.I tried different settings in the TV menue like going for only positive values or only negative values as far as possible but nothing changes because the TV seems to equalize the settings.
For 6500K I now use the following whitepoint correction (the range is -50 – +50):
R-Gain +50
G-Gain +41
B-Gain -17R-Offset +2
G-Offset +4
B-Offset +0So the only possibility to get more accurate colors is going for a cooler white right?
Maybe you can use HCFR as a fast way to monitor B primary behavior as you tweak (reset to defaults), just to discard wild OSD behavior as you alter gain. Because of the visual aid rather than spotread -x.
if Yxy does not vary wildly, then blue primary is too violet and the edge B->CM->White of 3D Volume is going to be OOG (sRGB) always.Wow. I have a Samsung qled and there’s no way those settings are correct. Are you set to warm2?
Even if you are in the wrong tv mode and wrong color tempature in the tv on my vizio v555-j01 series normal color temp works good and the warm color temp. But adding that much red and green offset is not good. Offsets should be checked with hcfr for clipping. I cant go more than 12 positive on the -100 to +100 without red clipping. I would keep the green offset on 0 . Blue is not that much important in white brightness it just effects the color temperature like red does to but red is close to being a white brightness effector. I use normal temp since the warm color temp clipped red at 0 even. Going less than 0 will decrease contrast but that is not important when your near 1000 and not getting full scale color . Wierd your haveing problems with blue not going bright enough must be the tv wanting offsets not that high. I am guesing what the tv needs.
Maybe you can use HCFR as a fast way to monitor B primary behavior as you tweak (reset to defaults), just to discard wild OSD behavior as you alter gain. Because of the visual aid rather than spotread -x.
if Yxy does not vary wildly, then blue primary is too violet and the edge B->CM->White of 3D Volume is going to be OOG (sRGB) always.Yxy only varies a little bit, so it is like you said and the blue primary is too violet.
If I use the EDID primaries in novideo_srgb to clamp to sRGB it looks good and the result is close to my other monitors after calibration.
I will just use the EDID values and combine them with the gamma result of my ICC profile.
I found another post where you mentioned to use ArgyIICMS to do that.
