Re: specifying colors outside srgb gamut with extended-linear-srgb

On 2021-06-08 08:32, Michael Smith wrote:
> Hi colorweb members.
>
> Using extended-linear-srgb seems intuitive when colors are inside sRGB 
> gamut, if I want a yellow that is twice as bright as a regular bright 
> sRGB yellow, I use (r,g,b)=(2,2,0) instead of (r,g,b)=(1,1,0).

That is outside the gamut, if you plot the 3D gamut in Lab space for 
example.

I agree its hard to see on a 2D chromaticity diagram, precisely because 
chromaticity removes the effect of luminance.

>
> What if I want my application to use a nice saturated BT.2020 yellow 
> instead, which is out-of-sRGB gamut, for example (r,g,b)=(2,2,0) in 
> BT.2020 linear would translate to (r,g,b)=(2.145699727, 2.016698845, 
> -0.237459323) in extended-linear-srgb with a negative blue value. 
It would, yes.
> Do we really expect users to specify negative color values? 

If that is the only way to get the color, then yes.

There are things like CSS Color 4,and Canvas HDR, which are color 
managed and thus can accept inputs in a variety of color spaces and do 
the conversion internally.

And then there are things like WebGPU and WebGL which are not color 
managed; you pick a single colorspace when setting up to draw and then 
everything has to be in that. Which means that other considerations 
(what is native to the hardware, which is the lower energy path) take 
precedence over, as you say, intuitiveness and forcing the user (or some 
tool) to dothe work of color space conversion.

> I'm concerned that will be non-intuitive to users without a color 
> science background. 
I have seen negative values be super confusing to people, yes (until 
adequately explained).

-- 
Chris Lilley
@svgeesus
Technical Director @ W3C
W3C Strategy Team, Core Web Design
W3C Architecture & Technology Team, Core Web & Media

Received on Tuesday, 8 June 2021 15:56:11 UTC