Add rec2100-linear to PredefinedColorSpace

[ccameron-chromium]

What is the issue with the HTML Standard?

The rec2100-linear is an extremely useful space for canvas applications that use floating-point color values, and should be added to PredefinedColorSpace. It is wide gamut, can be used with HDR, and is a linear space (which is particularly useful for physically based rendering applications).

It lines up with kCGColorSpaceExtendedLinearITUR_2020 and with VK_COLOR_SPACE_BT2020_LINEAR_EXT.

This space (and the other predefined spaces) all share the same white. So, in particular, [1,1,1] in rec2100-linear is to match white. This is described in the converting colors section, but it's worth pointing out here.

[annevk]

Should this also be added to https://drafts.csswg.org/css-color-5/ then? Otherwise you would not be able to use the same values without conversion in fillStyle and the like, right?

cc @svgeesus @whatwg/canvas

[svgeesus]

Should this also be added to https://drafts.csswg.org/css-color-5/ then? Otherwise you would not be able to use the same values without conversion in fillStyle and the like, right?

It is in CSS Color HDR which was adopted by CSS WG resolution in 2023

It was also added to color.js in 2024 so you can use it in the color calculator eg for creating tests.

[ccameron-chromium]

BTW, this is what the PR would look like #11529.

[annevk]

Discussing this with colleagues we'd be more supportive of this if we could add srgb-linear and display-p3-linear (the latter does not seem to be in CSS at this time, not sure why) at the same time.

[svgeesus]

Discussing this with colleagues we'd be more supportive of this if we could add display-srgb-linear

Is that a copy-paste error? Should it be srgb-linear

I agree it would be useful for Canvas.

and display-p3-linear (the latter does not seem to be in CSS at this time, not sure why) at the same time.

Feel free to raise a csswg issue about that, preferably with a brief explanation of why it is needed (aligns with hardware? fits an Apple API?)

[ccameron-chromium]

I'd be fine with adding more, if there proves to be a need. In terms of linear spaces, they are all equivalent. I'd prefer smaller steps for now.

[ccameron-chromium]

I'm also fine with srgb-linear. I do want to clarify that it will be have like kCGColorSpaceExtendedLinearSRGB and not DXGI_COLOR_SPACE_RGB_FULL_G10_NONE_P709. That is, the backbuffer value [1,1,1] shall be the same color as white (modulo tone mapping), not some independent separate thing.

How about we do those two, and then consider display-p3-linear as it makes its way through CSSWG etc?

[svgeesus]

Feel free to raise a csswg issue about that, preferably with a brief explanation of why it is needed (aligns with hardware? fits an Apple API?)

I'm assuming this aligns with kCGColorSpaceLinearDisplayP3

How about we do those two, and then consider display-p3-linear as it makes its way through CSSWG etc?

Sounds like a plan. @ccameron-chromium I assume since you proposed it that there is Chromium implementer interest, could you confirm?

Who is the best contact from Mozilla to judge their implementer interest?

@annevk please raise the issue on csswg Github to motivate adding display-p3-linear to CSS Color 4.

[ccameron-chromium]

Sounds like a plan. @ccameron-chromium I assume since you proposed it that there is Chromium implementer interest, could you confirm?

Yes, confirm! (interest in rec2100-linear and srgb-linear, not-opposed to display-p3-linear and perfectly willing to add it if desired elsewhere)

[annevk]

I filed w3c/csswg-drafts#12596. If it's not too much trouble we'd much prefer adding them all at once.