W3C home > Mailing lists > Public > public-css-archive@w3.org > May 2020

Re: [csswg-drafts] [css-color-4] Do gradients/animations using lab/lch colors interpolate in the Lab colorspace? (#4647)

From: Tab Atkins Jr. via GitHub <sysbot+gh@w3.org>
Date: Wed, 06 May 2020 17:07:51 +0000
To: public-css-archive@w3.org
Message-ID: <issue_comment.created-624773238-1588784870-sysbot+gh@w3.org>
> I thought we had decided that premultiplied alpha was a mistake and we should've just handled transparent specially? Now we want to repeat this?

Nope, premultiplied is still the way to go.  Previous threads are:

* <https://github.com/w3c/csswg-drafts/issues/2722>, where we diverted to talking about image-fading, and realized that the "correct" way to fade between an opaque image and a mostly-transparent one is *also* to do so in premultiplied space, as that's the effect you get when you do a "dissolve" transition with infinitely-small pixels
* an older pre-GH thread at <https://lists.w3.org/Archives/Public/www-style/2016Jan/0181.html> where I talk more about the decision and why it would be hard/bad to instead do a special-case for 'transparent', particularly due to the effect on animations.

> I agree that the contribution from undefined hue should merge in gradually, but over a much smaller interval.

Yeah, something that sticks to *almost linear* for most chroma differences, and only exposes a significant difference when one chroma starts getting close to zero, would be fine I think.  Basically, if you can still see a reasonable amount of the hue, it should continue to be more or less uniform.

-- 
GitHub Notification of comment by tabatkins
Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/4647#issuecomment-624773238 using your GitHub account
Received on Wednesday, 6 May 2020 17:07:54 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 06:42:06 UTC