Re: [csswg-drafts] [css-color] Separation / DeviceN color support (#2023)

It's quite possible I've misunderstood your point, but I'll plunge in anyway. Feel free to correct me.

From a PDF perspective, we could work with the fallback color in almost any color-space you like. CMYK is by far the most common - these things are destined for print, after all - but I've seen Lab used. Also allowable in the PDF format is RGB, or any ICC-based colorspace - it doesn't have to be device-cmyk, it could be Fogra39  for example. Any one of these would give a colorimetric value, no?
@color-profile --orange-021c {
  src: Lab;
  components: "PANTONE Orange 021C" 60% 65.7 85.1;

@color-profile --warm-red {
  src: fogra39;
  components: "PANTONE Warm Red C" 0 0.76 0.9 0.01;

Of course interpolation on a device without the ink available will happen in the fallback space, and as we know this looks pretty rubbish with Lab. Lch would be ideal except it's not available in PDF. I can simulate a gradient between two Lch colors in Lab quite easily, by dividing the gradient up into sections until the delta-E is close enough. But when _generating_ a PDF I can't do this with spot colors defined against Lch; the decision on whether to interpolate in the named-ink or the fallback-color doesn't happen until the device _rendering_ the PDF. 

All of which means that from the perspective of someone generating a PDF, it's possible to accurately convey both the above examples, although the Lab one wouldn't be great in a gradient. But if you specified the ink in Lch (or HSL, or anything not native to PDF), we'd have to convert it to something natively supported in PDF first. That would likely be CMYK as it interpolates better. But this doesn't seem a problem to me - I think an author using a named color will be aware of this.

From the perspective of someone trying to display these on screen, I think it's even easier isn't it? Provided a space other than device-cmyk was used for the fallback color, you can get an accurate colorimetric value at any point along a gradient.

GitHub Notification of comment by faceless2
Please view or discuss this issue at using your GitHub account

Sent via github-notify-ml as configured in

Received on Friday, 13 November 2020 18:45:43 UTC