understanding rendering

you''re asking the difference between realtime graphics and slow time (ray traced) graphics.

to answer the first question, during rendering, opengl may be used to accelerate some portions of it, but it''s mostly the cpu that handles it. (traditionally)

i suppose the fundamental difference would be that in realtime, we''re concerned with speed, making use of the hardware, and generaly we work at a vertex level and interpolate across the pixels of the triangles they form. we''re interested in a fast approximation of the real thing.

in slow time graphics, we''re interestd in the quality, so a lot of effort is put into accurately modelling the interaction of objects in the scene, and everything happens at the pixel level. a ray is cast from the veiwer through each pixel, and then the ray tracer determines how it bounces around in the scene to get the final color.

some of this may change as graphics hardware becomes more and more powerful. already, the cgshaders.org website has some shaders that claim to do fragment raytracing in hw for the upcoming nv30. i haven''t looked at these in depth yet, though.

i haven''t done either process justice with this post, but hopefully this gives you the gist.

dedrick@whiteknucklegames.com