Re: [V7 31/45] drm/colorop: add BT2020/BT709 OETF and Inverse OETF

[Pekka Paalanen <pekka.paalanen@haloniitty.fi>]

[-- Attachment #1: Type: text/plain, Size: 2488 bytes --]

On Thu, 23 Jan 2025 15:16:29 -0500
Harry Wentland <harry.wentland@amd.com> wrote:

> On 2025-01-17 04:06, Pekka Paalanen wrote:
> > On Thu, 16 Jan 2025 10:56:22 +0200
> > Pekka Paalanen <pekka.paalanen@haloniitty.fi> wrote:
> >   
> >> On Thu, 19 Dec 2024 21:33:37 -0700
> >> Alex Hung <alex.hung@amd.com> wrote:
> >>  
> >>> From: Harry Wentland <harry.wentland@amd.com>
> >>>
> >>> The BT.709 and BT.2020 OETFs are the same, the only difference
> >>> being that the BT.2020 variant is defined with more precision
> >>> for 10 and 12-bit per color encodings.
> >>>
> >>> Both are used as encoding functions for video content, and are
> >>> therefore defined as OETF (opto-electronic transfer function)
> >>> instead of as EOTF (electro-optical transfer function).
> >>>
> >>> Signed-off-by: Alex Hung <alex.hung@amd.com>
> >>> Signed-off-by: Harry Wentland <harry.wentland@amd.com>    
> >>
> >> Hi,
> >>
> >> why would a display system ever use BT.2020 or BT.709 OETF or its
> >> inverse?  
> > 
> > Sorry, this is more for my own curiosity, not an argument against the
> > patch. Since hardware designers decided to incorporate these curves
> > explicitly, what use was in mind? It's likely something I have
> > overlooked.
> >   
> 
> I'm not entirely sure myself, but gamescope does use it for displaying
> game streaming content.
> 
> https://github.com/ValveSoftware/gamescope/blob/2f88849ac9dc7da5c678d5d7d3e9b58f38add1bf/src/Backends/DRMBackend.cpp#L2509

I see, apparently they are being used as a pair to make a roundtrip to
some optical space for applying a color transformation matrix. IOW, the
CTM is being applied in the assumed "camera" side of broadcasting or
grading.

ITU-R BT.2408 seems to refer to this as "scene referred mapping", whose
aim is to match cameras rather than displays (and has nothing to do
with the scene). I cannot claim to understand this case, but I suspect
it should be used only before color grading (manual or automatic
artistic adjustments, or IOW picture formation(?)) in a pipeline.

I would assume that game streaming content has been prepared for a
display at its source, which implies that grading has been done. That
means that inverse-OETF leads to an unknown color space. Applying a
matrix in an unknown color space has unintended effects. That may not
be a big problem, even less if the matrix is visually adjusted, but
it's theoretically not quite right AFAIU.


Thanks,
pq

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]