Perhaps you could very cheaply (scanline/zbuffer/hardware?) render a
low-resolution "view" from the primary intersection point, and then use the
brightness/contrast information to affect the distribution of the rays cast
from that point?



> > Sorry Christoph, I guess I've been reading too many global illumination
> > papers, and they've been numbing my brain :) From my understanding so
far,
> > importance sampling attempts to reduce the rays needed, and/or put the
ones
> > cast to better use by not simply shooting rays/particles in a random
manner,
> > but by attempting to adaptively "target" them to the areas that will
make
> > the most difference.
>
> Ah, i think i understand.
>
> AFAIK POV-Ray does not do this currently, but i'm not sure how you would
> accomplish that anyway (how to know in which directions sampling is
> needed)
>
> Christoph

Post a reply to this message