- This topic has 7 replies, 3 voices, and was last updated 4 years, 8 months ago by
.
-
Posts
-
Hey guys, looking for some help on creating a 3DLUT (for MadVR) for a BenQ LK970 that has an irregular native gamut coverage. My signal chain is Nvidia GTX 1080 (0-255) –> Denon 8500h –> BenQ LK970 (0-255). MadVR is also set to 0-255 so using full through the chain. Images will show the settings and results. Thanks!!
Hey Florian, the post 3DLUT results look a bit off to me, it put almost all of Cyan at the 25% saturation point. With the resulting 3DLUT I think a lot of the blue to green is lost. In testing with a few movies specifically Avengers Infinity War when Loki is holding the Tesseract in the beginning there is almost no differentiation in the cube, when disabling the 3DLUT you can see the stone and various shades of blues radiating out from it. Anyhow I was checking to see if there is something I am missing to keep the gamut as large as possible while getting the colors as close as possible rather than shrinking the gamut down so much and losing so much detail.
But that’s what you are asking for – your display cannot reproduce these tones, so they are clipped to the next in-gamut colors. It has nothing to do with “shrinking the gamut”, although the latter is probably what you want – shrink Rec. 709 to the extent that it fits within your limited display gamut, keeping a pleasing appearance (“perceptual”), and compromise on accuracy.
Seems other users have had a heck of a time with 3DLUTS with this projector also but mentioned using a concatenation method in Lightspace in order to get it right. I was reading up on it and noticed Manni talking about how DisplayCal handled it with little hassle. I am not familiar with that method nor how I would do that in DisplayCal. Is that something that makes sense? I think I understand a bit what it means but I wouldn’t know how to do it DisplayCal.
The “concatenation” method in LightSpace was/is necessary because (at least at the time) it apparently wasn’t capable of inverting the natural profile (RGB -> XYZ to XYZ -> RGB) when the display response was too nonlinear (i.e. multiple RGB triplets sent to the display produced the same or very similar XYZ readings), and also (seemingly) lacked any real sophisticated gamut mapping capabilities (apart from colorimetric).
Such problems do not exist with the Argyll profiler (although concave regions in the display gamut are always going to affect resulting LUT smoothness to some extent, not sure if that is the case for you though).
The solution in your case is clear: Since your display gamut doesn’t cover the whole of Rec. 709, you have to sacrifice some accuracy and let the gamut mapping compress the Rec. 709 space into your display gamut, by choosing one of the “perceptual” or “saturation” intents for the 3D LUT.
Sounds good, are those options under the ‘Rendering Intent’ field on the 3D LUT tab? So it sounds like ‘perceptual’ intent might be the place to start, keeping a pleasing experience and sacrificing accuracy? Thanks again Florian!
But look at the before 3DLUT picture. The Cyan secondary is out past the 75% saturation mark.
https://hub-assets.displaycal.net/wp-content/uploads/users/mystigiandoll/2019/08/09/before3dLUT.jpgSo clearly the projector is capable of that?
According to the manufacturer, the projector is supposed to be capable of 97% rec709 coverage.
After the 3DLUT, the rec709 coverage is only about 65% and the Cyan secondary is between the 25 and 50% marks.
https://hub-assets.displaycal.net/wp-content/uploads/users/mystigiandoll/2019/08/09/After3DLUT.jpgSurely can’t be correct? Now I know manufacturers lie, but surely not by that much?
I have been trying to calibrate this same projector for a friend of mine as well so I am very interested in this topic.
