- This topic has 19 replies, 3 voices, and was last updated 3 years, 3 months ago by
.
-
Posts
-
Hi, I have an issue with DisplayCal and hope that I will get some help here 🙂
A few feeks now, I am trying to create a good 3D LUT for watching movies in SDR BT709 on a RGB Laser projector, which has a very big color gamut.
The first thing I don’t understand is, why DisplayCal shows me a small color gamut coverage of about only 82% of BT709 / sRGB after profiling (XYZ LUT + Matrix). The gamut of the projector is much bigger as you can see in the attached measurement chart of HCFR.
After creating a 3D LUT there are some pretty big issues as you can see in the 2nd HCFR measurement chart. Cyan is very undersaturated and red oversaturated.
I would be very happy if someone can lead me in the right direction and finally be able to solve the issue, as I am very happy with the projector beside the color issues.
I also attached the measurement report of DisplayCal, maybe there is a hint which shows why the issues occur.
-user messing with projector CMS => reset to defaulst and proceed again
-wild behavior “in gamut” => profile with no LUT3D won’t be able to validate. Uncheck projector settings that looks like “dynamic something” or add more nodes.The projector has all dynamic settings, which could mess up the results, disabled.
Color Management (called “Color Tuner” in the projector settings menu) is disabled as are all other dynamic settings. The projector is working in its most basic form without any “image enhancements” already.
But it is a DLP projector (which are known for low color luminance @ 100% saturation / 100% Stimulus) and I don’t know if this has something to do with the issues.
All primary and secondary colors of the DLP RGB laser projector have very low Color luminance @ 100% saturation and 100% stimulus (up to 50% too dark compared to the reference luminance values). Even if all dynamic settings are completely disabled. But then again, the primaries are much more saturated than BT709 (as far as I know, the more saturated a color becomes, the darker also it gets?).
Maybe DisplayCal is not able to correct the low color luminance and tries to compensate for this and therefore shows small gamut coverage after profiling and outputting weird looking 3D LUT?
I will also later load up the 3D LUT again before profiling and send you the measurments.
But here is also one thing I found.
If I choose a 3D LUT within DisplayCal with the rendering intend “preserve saturations” it looks like this:
In my understanding that means, that DisplayCal is able to map the colors in a better way and the low color luminances don’t affect the results?
Your screenshot only shows us the 2D gamut, not the actual color volume (including luminance).
Maybe the luminance values are very low, and this could affect the gamut coverage of a 3DLUT depending on the white luminance being asked for.
What projector is this? Formovie?
Yes, it is the Formovie Theater projector.
The luminance values are low for 75% and 100% saturations of the primaries and secondary colors (something between -25% and -50% ). Can you tell me how I can send a screenshot of the color volume via HCFR or DisplayCal?
With the screenshots of color volume I would also send the measurement report with active 3D LUT.
HCFR doesn’t have the ability to measure gamut coverage.
DisplayCAL does, it’s the result of profiling, in the folder it creates are a bunch of files. You can open your preofiling result with DisplayCAL profile information screen (the little ( i ) next to the settings dropdown).
And in that screen you can select a colorspace like xy, uv, iCtCp etc and click the 3D dropdown and export as HTML.
I have measurements of a Formovie as well and here are some 3D views of the gamut coverage.
https://nicko88.com/misc/formovie_gamut/
It claims something like 107% BT.2020. Sure, maybe at like 1/4 the full brightness. At higher brightness the coverage is a LOT lower since it can’t reach the proper color luminance there.
When you are generating a 3DLUT, I assume you are leaving the white level set to “as measured”, so it’s trying to use that as the target white level, but the gamut coverage is not able to reach as wide at those brightness levels, so the gamut coverage is smaller.
If you set a lower target white level then the gamut coverage will be able to be wider, but at a large cost of brightness output and contrast too.
Since if you are digitally lowering white point by say 50%, that would lower the native contrast by 50% as well. Because you are not actually lowering the laser power, so white gets 50% dimmer but black stays the same.
Thanks for the instructions, I will post it as soon as I am home again.
I tried almost all settings, white point as measured and also @ D65 – the small gamut coverage is identical. But I discovered when I choose the profile type “curves + matrix” instead of “XYZ LUT + matrix” the gamut coverage seems to be correct with something like 99% BT709 (due to the slightly undersaturated blue primary I guess).
But the 3D LUT nevertheless looks the same as with the XYZ LUT + matrix profile.
Your hint setting the white point at lower brightness target seems very reasonable, I have’t tried that yet and will do it too.
But as you can see at the screenshot, DisplayCal left the red primary @ 100% saturation very oversaturated instead of undersaturated like cyan etc. , even though red has also low luminance. That is something I don’t understand.
Well curves + matrix is just 1DLUT and isn’t really adjusting the colors much, just the grayscale.
It’s not about “white point” as measured (that’s the x,y coordinate color temperature), it’s “white level” as measured, vs targeting a lower cd/m^2 (nits) white level.
Ok got it regarding “as measured” and 1D LUT 🙂
But do you have an idea about the 100% saturated red point, why DC does not process it in the same way as cyan (oversaturated instead of undersaturated)? The low low luminance of red 100% is approximately the same as cyan 100%.
Unfortunately that is already getting past my understanding of 3DLUT color transformation behaviors so I don’t have an answer for you on that.
No problem 🙂
But thank you anyway, the hint regarding lower target brightness may help.
I will post the results within the next days and I hope to get forward with the 3D LUT.
For now the 3D LUT rendering intent “preserve saturation” looks the best to me, as it does not undersaturate any color but oversaturates them a bit. I managed to correct the colors even further using the color managment of the projector after applying the LUT (better brightness values and better saturation tracking).
But that can’t be the solution, as a 3D LUT should alone be enough to correct all colors.
Yeah, I think the point is the 3DLUT “can’t” correct the colors at your current “as measured” white level because the projector simply can’t do it. The 3DLUT has to make a tradeoff between color luminance and color saturation. So saturation does down but luminance goes up.
That’s why there is a “preserve saturation” rendering intent. It doesn’t make as much of a tradeoff. It preserves saturation leaving luminance to be farther off.
I did now two tests and reduced the brightness one time to 50 and another time to 30 cd/m² instead of “as measured”. The untouched brightness before was about 85 cd/m².
The BT709 gamut coverage has risen from 82 to 90% in both tests. That means even reducing it below 50 cd/m² does not make any difference.
After measuring the result with the active 3D LUT and with the rendering intent “absolute colorimetric with white point scaling” the only thing that has changed is that the 100% red saturation is now matched with the target correctly. But cyan still keeps undersaturated, even @ 30 cd/m².
But even if the result is a bit better because of the correct red saturations now, it is not a solution, due to the low contrast and brightness loss. I hoped that there would be anything I can do, even if that means low color luminance @ 75 and 100% saturation.
Just hitting the saturation and hue targets of all colors would be enough for me honestly, because it’s almost impossible for the eye to know if a 75 or 100% saturated color is a bit darker than it should be, without having a correctly calibrated display in direct comparison.
But the undersaturated cyan is easily visible for me without direct comparison.
Problem ist, that DisplayCal has no other rendering intent options beside “preserve saturation” which could provide a better result and here the oversaturation is a bit too strong for my taste. At the moment the only solution seems touching the CMS of the projector after applying the 3D LUT to do further corrections.
But maybe there is anything hidden, which I haven’t find out yet. That is the big question 🙂
The next thing I can try now is to rise all color luminances to its maximum in the CMS of the projector before profiling, maybe that will make a difference and DisplayCal will give better results.
