[Cin] HDR10 on radeonsi/EGL

Georgy Salnikov sge at nmr.nioch.nsc.ru
Fri Apr 25 04:56:00 CEST 2025


On Thu, 24 Apr 2025, Andrew Randrianasulu via Cin wrote:

> interesting bug report I found.
>
> https://gitlab.freedesktop.org/mesa/mesa/-/issues/12669
>
> "radeonsi: Cannot deliver HDR10 0.0-125.0 values to 16 bit float GBM
> surface, only 0.0-1.0 seems to work. "
>
> So, it sort of force-normalized at least in egl ?

See https://en.wikipedia.org/wiki/Rec._2020
-------------------------------------------
Rec. 2020 defines a bit depth of either 10 bits per sample or 12 bits per
sample.

10 bits per sample Rec. 2020 uses video levels where the black level is
defined as code 64 and the nominal peak is defined as code 940. Codes 0-3
and 1,020-1,023 are used for the timing reference. Codes 4 through 63
provide video data below the black level while codes 941 through 1,019
provide video data above the nominal peak.
-------------------------------------------

I'd expect HDR10 [0-1023] just mapped to float [0.0-1.0], from which the
nominal range be [64-940] == [0.063-0.919], and HDR range [4-1019] ==
[0.004-0.996]. Then 125.0 would be 127875, hardly presentable in 10 bit?

_______________________________________________________________________________

Georgy Salnikov
NMR Group
Novosibirsk Institute of Organic Chemistry
Lavrentjeva, 9, 630090 Novosibirsk, Russia
Phone   +7-383-3307864
Email   sge at nmr.nioch.nsc.ru
_______________________________________________________________________________



More information about the Cin mailing list