First of all, the technique you're presenting here
doesn't use stencil-shadowing (that's the thing
with geometry of shadow casters beeing extruded in
light-direction to the frustrum's bounds, then even/odd
testing the number of shadow-surfaces actually drawn
behind the scene).
Also it looks like you're using the non-hw sccelerated
version of shadow maps (FP-Textures carrying depth,
instead of real hardware depth-buffers, moving the
depth-comparison and weighting step the the hardware).
So if you disable Z-testing, the outcome should
look like always when you disabkle z-testing, at least
without any respect to shadows.
Maybe you should first take a look at wich format
your shadow maps are really using (does it depend
on the cards used?).
Gruss
Jan Bruns
"Noley Edge"
gmail.com> schrieb im Newsbeitrag news:bcceedb9-0277-4d50-935f-3f425c6fb4d6@i20g2000prf.googlegroups.com...
>I am rendering shadows using depth stencil buffers and a shadow map...