On 13/11/06, Mark Heuymans <[EMAIL PROTECTED]> wrote:

>>I like the AO effect, except for the render times. Why not go for
>>full GI if it's only a little bit longer to render?
>This will be fixed (atleast to some extent) in V6: the RayTracer VSL
>object can evaluate only the ray distance, which should make AO quite
>a bit faster.


That's nice to hear. I want V6!
Is it possible to implement some kind of adaptive raycount, i.e.
before starting the real tracing, the Raytracer object looks around a
bit for close objects or close surface curvatures and determines
raycount from there. For this first step, no materials would have to
be evaluated, just geometry. Basically, what I mean is: use high
raycount/low blur where it's needed near details and corners, low
raycount/high blur in the middle of large areas. The Blur effect
would have to be adaptive too - difficult...

Some sort of adaptive sampling would be nice, but if I remember
correctly, the RayTracer object already does this with higher
RayCounts. This doesn't affect the blur kernel size though. What
you're describing is pretty close to how photon mapping works, which
would be nice to have in RS.

tnx,
Mark H


Reply via email to