While working on the texture image code I had to fix some
things related to color index (paletted) texture images. That
got me thinking about other texture formats like depth textures.
So, I implemented the GL_SGIX_depth_texture extension and then
GL_SGIX_shadow to exercise it. GL_SGIX_shadow_ambient was a
simple addition on top of that.
Together, these extensions implement shadow maps (invented by
Williams, I think). Basically, the scene is first rendered from
the point of view of the light source in order to generate a depth
map. Then, the scene is rendered from the regular camera's point
of view. Texgen and a special texture matrix compute the projection
of each vertex (as a texture coordinate) into the light's coordinate
system. During texturing, we compare the texture R component to
the depth map value at (S,T). The result of the comparison determines
whether or not the fragment is shadowed.
The technique isn't perfect but can work pretty well in the right
situation.
There's a demo in Mesa/demos/shadowtex.c
My implementation isn't 100% complete either. Right now, the
depth texture is only sampled with nearest/nearest filtering
and the texels are always stored as floats. I'll try to fix
those things in the not-too distant future.
-Brian
|
