kramnelis
Gawd
- Joined
- Jun 30, 2022
- Messages
- 890
This algorithm I use is more competitive than Dolby Vision. 10,000nits objects are rarer than 2,000nits objects. So better saturate 2,000nits object first.Part of DolbyVision's PR:
So, based on a 10,000nit screen's curve obviously (other screens like 725-800nit and hdr1400nit have an accuracy fall-off and compress the top end so the values will be some what different compressed into the remaining range on the high end):
So,
50% of the screen displayed is at 0 to 100. (SDR ranges and down to infinite black depth in OLED). This foundation probably remains the same even on compressed after fall-off screens.
25% of the screen is at 100 to 1000nit. (bulk of which is typical mids to high mids that many HDR screens can display more or less, at least for a time if not long sustained).
The top 25%. (likely bright highlights e.g. scintillation/glints/reflections of sources, and very bright direct light sources)
The top end will be compressed on screens since practically all of them are below 10,000nit so that last 25, 25 won't be the same. For example HDR 10,000 curve on a LG OLED is around 400nit accurate then falls off to compressing the rest into the remaining ~ 400nit. Not sure where the fall-off is on the asus proarts uqx and ugc.
more from the dolby mastering page:
"I found it helpful to focus on segmenting an image and placing different portions of the shot at different nit values where they are best represented."
Again, contrast is higher lol. It's an entertainment to see what you got.