Re: [css-gcpm][css-color] device-cmyk() interaction with RGB-space colors.

On Tue, Sep 24, 2013 at 3:14 PM, Rik Cabanier <cabanier@gmail.com> wrote:
> On Tue, Sep 24, 2013 at 2:58 PM, Tab Atkins Jr. <jackalmage@gmail.com>
> wrote:
>> On Tue, Sep 24, 2013 at 2:31 PM, Rik Cabanier <cabanier@gmail.com> wrote:
>> > The compositing/blending spec doesn't talk about converting colors [1]
>> > All
>> > the formulas (except for 4 special ones) work on individual color
>> > channels,
>> > regardless of what they represent.
>>
>> Those channels appear to be assumed to be RGB.
>
> Can you point out where that is so I can update the spec?

Hmm, I guess it actually doesn't.  The few examples that give explicit
color numbers are all written with RGB, but there's nothing outside
the examples that actually talks about the color value.

>> If it's compatible
>> with CMYK, cool, but that needs to be specified (and again, needs to
>> be defined how you blend/composite between RGB and CMYK).
>>
>> I doubt that this "just use the channels" approach is compatible with
>> some other color spaces, like Lab, though.
>
> It is compatible. You can try it in PhotoShop :-)

How does that work, though?  I mean, HSL doesn't work in a "just use
the channels" approach, which is why the HSL blend modes are
non-separable.  Why would Lab work?

>> > I agree we need to define a way how and when individual elements are
>> > converted to the document's color space.
>>
>> Sure, something more advanced than just "always convert to RGB
>> immediately" might be worthwhile.  We talked about this at the f2f,
>> though, and some of the ideas (like converting individual pixels to
>> RGB as necessary, or converting entire elements to RGB as soon as you
>> need to convert some part of it) were all horrible and not worth the
>> benefit, so I didn't put those ideas into the spec.
>
> It seems we could define something simple. People have a tendency to
> overcomplicate color management...

Ideas welcome.

~TJ

Received on Tuesday, 24 September 2013 22:26:14 UTC