[cairo] Spot colors (and CMYK)

[Bill Spitzak]

James Cloos wrote:
>>>>>> "Francois" == Francois Robert <frobert at atex.com> writes:
> 
> Francois> Just a side note: In PDF, the blending space is _explicitely
> Francois> specified_ (see section 7 on transparency), and it can be any
> Francois> flavor or gray, RGB or CMYK (but not L*a*b*). Not sure if it
> Francois> is something that Cairo must care about ?
> 
> Probably just to the extent that cairo documents why spaces like L*a*b*,
> L*u*v*, HSV, Yuv, and similar do not make good spaces for blending:
> 
> ,----< quoted from PDF Reference 1.7, §7.2, p519 >
> | the compositing computations in spaces [which represent lightness and
> | chromacity separately] do not give meaningful results when applied
> | separately to each component
> `----
> 
> -JimC

Compositing only really makes sense in a linear space where 0,0,0 means 
zero light energy. It works identically in XYZ and in *linear* RGB (no 
matter what 3 primaries or what whitepoint is chosen). The result is 
very very close to actually combining all the photons together when you 
add, and muliplying by a constant is attenuating the light by that value.

Non-zero black (such as in Yuv) makes compositing completely useless. 
Non-linear (such as Lab, HSV) makes compositing rather useless, though 
extremely small color shifts work. Compositing is sRGB does work 
reasonably well despite the non-linear nature.