[RFC wayland-protocols 1/1] unstable: add color management protocol

[Chris Murphy]

On Thu, Feb 14, 2019 at 2:27 AM Philipp Zabel <p.zabel at pengutronix.de> wrote:
>
> In the Kernel, AdobeRGB has been replaced with opRGB. Should this be
> name="opRGB", summary="opRGB IEC 61966-2-5"?

opRGB and Adobe RGB (1998) are different color image encodings, and
they apply to different use cases. They strictly speaking are not
interchangeable even if the error resulting from substitution is
small.

http://www.color.org/chardata/rgb/adobergb.xalter
http://www.color.org/chardata/rgb/oprgb.xalter

A color image encoding includes color space, reference viewing
condition, and reference medium. The ICC profile might include tags
for the last two items, optionally, but there's no practical way for
the profile or the color management system to enforce them: it can't
make your display brighter or darken the room. The human is expected
to do that enforcement.

And just because many humans do not even try to approximate reference
viewing conditions or medium, doesn't mean it's OK for an automated
system or an operating system to substitute opRGB for Adobe RGB (1998)
as if they are the same thing. They aren't.

The opRGB use case is for camera encodings going forward. It's not
intended to automatically/transparently/silently substitute opRGB for
existing Adobe RGB (1998) images. You could say, we have nothing to do
with the reference condition or medium, so we actually can do that
substitution! I think that's a slippery slope. And you definitely
cannot get away with it in an HDR world. So my suggestion is to treat
these things as what they are: color image encodings. That's what's
meant by the name, not merely the color space.

And actually, it's probably best to use the following reference for
the name (short name) and definition (summary). And it's also sane to
include all of the listed color image encodings, in this proposal.

http://www.color.org/chardata/rgb/rgb_registry.xalter


--
Chris Murphy