Re: CSS3 color and RGB color spaces?

On Thursday, May 9, 2002, 1:29:14 AM, Peter wrote:

PS> Chris,

PS> Now that many output devices (LCD monitors and inkjet printers)
PS> are shipping that are outside of the sRGB gamut, I'm not convinced
PS> that the current wording of the color module is acceptable any
PS> longer.

I agree the wording is not good. I don't agree that the specification
is insufficient, however.

PS> Shouldn't CSS3 consider using scRGB instead, or saying something
PS> different about how to deal with RGB values that are out of the
PS> sRGB range (the current wording only states that they should be
PS> clipped if outside the device's gamut, but says nothing about
PS> dealing with devices outside of the sRGB gamut).

I agree that the wording should be improved, but in fact it *does* say
something. It says that values should be clipped to the *device*
gamut, not the sRGB gamut. So for example

rgb(-12%, 32.456768%, 108%)

is a valid CSS color Its outside the sRGB gamut - the red component is
negative and the blue component is hypersaturated. In a particular
device gamut, that might well be a displayable color. If so, it will
be displayed. If not, it will be clipped.

I think the current spec tries to say too much in too little space, so
this is too subtle a conclusion and should be explained step by step
with several examples.

Step one, using % gives arbitrary resolution not the 256 steps that
the other notations give

Step two, negative values and values greater than 100% are not only
allowed, but are also useful They represent colors outside the sRGB
gamut.

Step three, to display a color acurately the specified sRGB value is
converted in a CMS to the color space of the target device.

Step four, some colors outside the gamut of sRGB may thus fall inside
the device gamut (and equally, colors in the sRGB gamut may be outside
the device gamut).

Step five, colors outside the device gamut will be clipped.

Currently this is all done in one example and furthermore, one that
uses an identity transform (device colorspace == sRGB) and its not
clear all that is happening.

-- 
 Chris                            mailto:chris@w3.org

Received on Wednesday, 8 May 2002 20:10:47 UTC