> From [EMAIL PROTECTED] Tue Dec 7 13:33:19 1999
> FYI, I'm the originator of this mail thread. I've decided to
> implement true Phong into Mesa using the flag GL_PHONG for
> glShadeModel(). [...]
> This addition to Mesa should be relatively trivial, so I expect to have
> something up and going soon.
I still think this is a bad idea, as it makes the lighting model
internally inconsistent. Per-vertex lighting calculates a color. The
rasterization stage decides what to do with those colors. "GL_SMOOTH"
means to interpolate the vertex colors across the polygon. What does
GL_PHONG mean if lighting is disabled and someone provides per-vertex
colors? Or if they have GL_LIGHT_MODEL_COLOR_CONTROL set to
GL_SEPARATE_SPECULAR_COLOR? You'll basically have to short-circuit
the entire vertex lighting calculation in order to implement this,
which means that some operations (like turning off lighting) won't
work as expected.
You think this work will be trivial, and it will be be easy to
implement the *specific* functionality you want, but weaving it into
the rest of the OpenGL state machine will take a lot more work than
you're expecting. In order to make this useful to other people, it
would be nice if your changes fit the OpenGL system well and
orthogonally, if not intuitively. That's why I suggested implementing
"EXT_fragment_lighting", and that I would help. What did SGI say
about that?
I don't want to discourage you from doing cool things with Mesa, of
course. And if these changes will never see the light of day, then I
don't care what you do.
-Brad
--
Brad Grantham, [EMAIL PROTECTED], http://alt.net/~grantham
_______________________________________________
Mesa-dev maillist - [EMAIL PROTECTED]
http://lists.mesa3d.org/mailman/listinfo/mesa-dev
