Re: [css-fonts] selecting color or monochrome glyphs for emoji fonts

On 21/8/14 21:40, John Hudson wrote:
> On 21/08/14 1:21 PM, Behdad Esfahbod wrote:
>
>> I'd rather CSS doesn't go down that path.  CSS doesn't specify whether
>> fonts should be hinted, use embedded bitmaps or not, render as subpixel,
>> grayscale, etc.  It's out of the scope IMO.
>
> I'm in two minds on this. My first instinct was the same as Behdad's:
> that this is a font rendering issue, and not an obvious candidate for
> CSS level authoring control. However, we do expect CSS to be able to
> control colour display of many page elements, including monochrome text.

I think the choice between text-style and emoji-style rendering is a 
legitimate option in page design that should be available to authors. 
Whether the already-available (if not yet widely supported) Unicode 
variation sequences are sufficient for this, or whether a CSS property 
such as John proposed is a worthwhile addition is a further question. 
Personally, I'm inclined to favor it.

Such a property would allow authors or designers to influence the 
default rendering of these characters in the absence of variation 
selectors, and without needing to explicitly provide or specify 
particular fonts.

The defined variation sequences, when used, should (IMO) take 
precedence, so that we'd have:

   U+2615         HOT BEVERAGE (rendered in text or emoji style
                                according to font-variant-color)
   U+2615 U+FE0E  HOT BEVERAGE (text style)
   U+2615 U+FE0F  HOT BEVERAGE (emoji style)

An explicit choice of "text style" or "emoji style", whether a result of 
variation selectors or of font-variant-color, would serve as an input to 
the font selection process, so if a font specified in font-family does 
not support the requested rendering style, the UA would attempt to find 
some other font (later in the font-family list, or using system-wide 
font fallback) that can support the requested style.

OTOH, if neither of these mechanisms is used to explicitly choose a 
rendering style, the first font in font-family that can render the 
codepoint would be used.

> While I can see the point of John's proposed font-variant-color
> property, I'm actually more interested in possible CSS interaction with
> colour font technologies that enable user-defined colour values for
> polychromatic glyphs. For example, the Microsoft CPAL table spec allows
> for user defined palettes, and CSS seems to me an obvious place to
> provide for author definition of colours.

Yes; but that's an entirely separate topic.

JK

Received on Thursday, 21 August 2014 21:12:05 UTC