On Mon, 6 Dec 1999, Tim Little wrote:
> RGB space also adds to the problem. The linear interpolation causes
> different hues and saturation to appear across the triangle as well. I
> have implemented a software rasterizer that works in HLS space and the mach
> banding is greatly reduced there while still doing gouraud shading.
I'm rather suprised at that. The mach band comes about because the
eye is looking for intensity changes as a hint to shape. Essentially
dropping the colour out of the equation shouldn't make a difference.
> With a
> simple lightness difference, you only vary the L component and the H and S
> stay fixed. Mach banding still occurs but it can be greatly reduced by
> using a color space that is more natural to humans than RGB, which is
> native to the computer.
This is somewhat true - but I think HLS is only more natural to
the higher brain functions. We *think* and *talk* in terms of brightness,
and colours. Nobody calls Yellow "Greenish Red" - and we describe
colours in terms like "baby blue" and "dark red" which are clearly
"desaturated blue" and "deintensified red".
However, our eyes have colour sensors that operate in essentially
RGB (although the Green sensors are more of a Yellow [because that's
the colour of sunlight] - and our blue sensors are closer to cyan
[because it gets you better contrast against the sky] - our Red
sensors are pretty close to CRT Phosphor red [presumably because
it helps for spotting berries in bushes]).
(Did you know that the humble domestic goldfish has the best
colour perception of any known creature? I think it has
something like seven different sensor colours.)
But in any case, HLS lighting breaks down badly when you have
coloured light sources.
HLS lighting math looks very attractive until you have one
red light and one green light!
Even if you accumulate the light colours after HLS->RGB conversion,
you have to worry about things like green lights shining on red
polygons (result==black) which is hard to do in HLS space.
Steve Baker (817)619-2657 (Vox/Vox-Mail)
Raytheon Systems Inc. (817)619-2466 (Fax)
Work: [EMAIL PROTECTED] http://www.hti.com
Home: [EMAIL PROTECTED] http://web2.airmail.net/sjbaker1
_______________________________________________
Mesa-dev maillist - [EMAIL PROTECTED]
http://lists.mesa3d.org/mailman/listinfo/mesa-dev
