W3C home > Mailing lists > Public > www-style@w3.org > September 2010

Re: CIE color definitions in CSS3 color module

From: Tab Atkins Jr. <jackalmage@gmail.com>
Date: Fri, 17 Sep 2010 16:25:09 -0700
Message-ID: <AANLkTin+A3o=5p=vAvxVLCom2cLN+pDwCQUVLex7ggpJ@mail.gmail.com>
To: Alan Gresley <alan@css-class.com>
Cc: Alexis Shaw <alexis.shaw@gmail.com>, www-style@w3.org
On Fri, Sep 17, 2010 at 4:17 PM, Alan Gresley <alan@css-class.com> wrote:
> Tab Atkins Jr. wrote:
> [snip]
>> Returning to the core of the issue; what's the use-case for these new
>> ways of defining a color?  I know that various other color definitions
>> *exist*, but I don't understand what they're used for, or if they're
>> useful for the web.  Could you elaborate on this?
> To display more of the colors that people can see on a future display
> device.
> I previously said this in this thread.
>  | If such a limited gamut of sRGB color-space can produce
>  | 16,777,216 values or 16,777,215 colors (minus black),
>  | this far exceeds the the average 10 million that a person
>  | can observe in theorized xyz color space.
> What I didn't say was that sRGB only contains about half or even a third of
> the colors that people can perceive. About 5 or 7 million colors can not be
> seen on current display devices.
> I can't create colors which I used with oil paint. These are Cadmium Red,
> Cobalt Blue, Magenta, Brilliant Green. The greens are the most prominent
> colors that are missing from a RGB color space pallet.
> Already we have cameras that have a larger gamut than RGB color space. Some
> of these images can be manipulated with various programs. The only problem
> is that the OS (ie. Window 7), lest any UA tired to such system can not
> shows these missing colors.
> Also, why not allow a display device to show ultra-violet.

I would much rather wait for a hypothetical future display device to
become real before trying to address its needs.  Engineering to meet
predicted future needs has a generally poor track record on the web.

>> Note, as well, that CSS generally assumes that colors exist in the
>> sRGB colorspace, and defines interpolation for things like transitions
>> and gradients in terms of that space.
>> ~TJ
> Yes, that is the challenge, it is not simple algorithms. For CIE xy
> chromaticity space and CIE xyY color space [1] we see this.
>  | If one chooses any two points of color on the chromaticity
>  | diagram, then all the colors that lie in a straight line
>  | between the two points can be formed by mixing these two
>  | colors. It follows that the gamut of colors must be convex
>  | in shape. All colors that can be formed by mixing three
>  | sources are found inside the triangle formed by the source
>  | points on the chromaticity diagram (and so on for multiple
>  | sources).
>  | The diagram represents all of the chromaticities visible to
>  | the average person. These are shown in color and this region
>  | is called the gamut  of human vision. The gamut of all
>  | visible chromaticities on the CIE plot is the tongue-shaped
>  | or horseshoe-shaped figure shown in color. The curved edge
>  | of the gamut is called the spectral locus and corresponds to
>  | monochromatic light, with wavelengths listed in nanometers.
>  | The straight edge on the lower part of the gamut is called
>  | the line of purples. These colors, although they are on the
>  | border of the gamut, have no counterpart in monochromatic
>  | light. Less saturated colors appear in the interior of the
>  | figure with white at the center.
> 1.
> <http://en.wikipedia.org/wiki/CIE_1931_color_space#The_CIE_xy_chromaticity_diagram_and_the_CIE_xyY_color_space>

If there's no simple algorithm to convert from CIE spaces to sRGB,
then this is dead in the water, isn't it?  Browsers work with colors
in sRGB.  You can sometimes specify colors that officially lie outside
the sRGB gamut, but it's clamped appropriately when actually fed to
the device.

If there's no good algorithm to convert from CIE colorspaces to the
gamut used by current devices, and no actual devices that display
colors in the CIE colorspace gamuts, then what's the use of specifying
colors in a CIE colorspace today?

Received on Friday, 17 September 2010 23:26:02 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:49:47 UTC