correct way to cal madVR for both SDR and "native" HDR mode?

[sblive]

Hi!

First and again: many thanks for your continued efforts, much appreciated! I did not find any donate button or similar. If you accept donations, I would like to make a small contribution.

I’ve been using DisplayCAL + madVR + 3DLUTs for many years now.

Even though I have displays with native HDR modes, I’ve never used them, but “tone map HDR using pixel shaders” for now, because I assumed/read that madVR does some dynamic mapping/processing (like Dolby Vision). For that I simply created DCI-P3 and BT2020 3DLUTs out of my “normal” cal run (desktop sRGB). I assumed this was better/not worse than anything the display hardware could do internally.
Also this was for a projector which may not increase max. brightness for HDR like an LCD. Now I also have an HDR LCD (LG 27UK850) calibrated to 140nit for normal use, while in HDR mode it would have ~400nit peak.

I’ve always wondered if it would be better to use HDR mode/format (i.e. tell the display). However then I assume I should create 3DLUTs in that mode using madTPG as well, because the display will have different max. nits and EOTF/gamma curve and slightly different colors, correct?
I also assumed and am confused by the “tone map via 3DLUT” option. Is this static/worse?

For the projector I’m using the following “hack”: 3DLUT for BT709 made in normal way with “lamp low” and one for BT2020 created with “lamp high”.
This of course means that theoretical SDR video in BT2020 (or watching BT2020 HDR without turning lamp to high) would be somewhat wrong.

What is the correct way to get both SDR and HDR madVR 3DLUTs and also normal desktop sRGB in the best possible quality and with cal/color correction?

Thanks!

• This topic was modified 5 years, 6 months ago by sblive.
• This topic was modified 5 years, 6 months ago by sblive.
• This topic was modified 5 years, 6 months ago by sblive.

[Florian Höch]

I’ve always wondered if it would be better to use HDR mode/format (i.e. tell the display)

Depends. Usually when you send HDR metadata to the display, it will do its own tonemapping, which is bad for profiling because it makes it impossible to characterize the native device response. It might also be bad for madVR tonemapping because essentially you are tonemapping twice.

I also assumed and am confused by the “tone map via 3DLUT” option. Is this static/worse?

It’s static, yes.

[Author]

Posts