- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Wed, 22 Jul 2015 09:28:17 -0700
- To: Chris Lilley <chris@w3.org>
- Cc: Nikita Vasilyev <me@elv1s.ru>, www-style list <www-style@w3.org>
On Wed, Jul 22, 2015 at 2:56 AM, Chris Lilley <chris@w3.org> wrote: > Monday, July 13, 2015, 10:16:51 PM, you wrote: >> Amusingly, Mike Bostock has another page describing the HCL rainbow as >> "ugly" <http://bl.ocks.org/mbostock/310c99e53880faec2434> ^_^ > > Not really seeing the relevance of that to this issue, which is about > adding LCH (and of course Lab) to css4 color. LCH(ab) is simply the > polar form of Lab: > > C = sqrt(a^2 + b^2) > H = atan2(b, a) > > Frankly the main obstacle to my putting these in CSS color 4 - which I > intend to do - is what would be a good syntactic form, how to do > fallback, etc. SVG had these already, with an sRGB fallback. It was > dropped from SVG2 because they would be better in CSS so they could be > used for HTML/CSS as well. > > Having got such a space, *one* of the things one can then do with it > is produce an even-perceptual-spacing color scale for scientific > visualisation. Note the scientific - the point there is to avoid > hiding or exaggerating small changes in the underlying visualized > values. > > "Prettiness" is a non-goal in that situation - and if desired, with > bright saturated non-evenly-spaced colors, its easy to do in RGB or > HSL. Note that CubeHelix was invented to *do* scientific visualization in a way that was attractive and had good perceptual qualities, in particular good even-and-monotonic lightness change. (I'm not against exposing Lab, whether in rectangular or polar form. But we do have to prioritize built-in color functions; we can't add everything, and as Houdini develops and exposes custom CSS functions, adding new color functions via a small, simple JS library will be pretty easy.) >> I am currently (slowly) working on turning the CubeHelix rainbow into >> something usable as a CSS color system. It's intended to be the basis >> of a new named color system, but it should be useful as an HCL-like >> system as well, that's hopefully a bit more attractive. > > You would need to demonstrate perceptual uniformity for it to be > useful as an LCH-like or Lab-like system. Other than the slight brightness movement up and down (not perfectly even like HCL, but not omg-terrible like HSL), the CubeHelix rainbow is generally similar to HCL in its perceptual properties. I'm interested in producing a rainbow because people *do* want to be able to cycle the hue and get decent results; HCL doesn't give them that, because at any given lightness, at least some of the colors are ugly. ~TJ
Received on Wednesday, 22 July 2015 16:29:10 UTC