Here are two new images of an interactive demo of our GPU-based
collision detection (what, I'm spamming flipcode again ? ). The
difference with the previous iotd is that lighting is now computed in hardware with vertex and
fragment programs coded with Cg, and is thus much faster. The Cg
programs perform low dynamic range image-based lighting through the use
of diffuse, specular, and environment cubemaps. Fresnel effect (more
reflections at grazing angles) is included.
Thanks to Paul Debevec for permission to use his light probe