>   In fact, if we give the exact same scene made of triangle meshes to
POV-Ray,
> most people will probably be quite suprised how fast POV-Ray renders it.
> POV-Ray can be incredibly fast when rendering triangle meshes. I have gone
> up to more than 30 millions of triangles in a scene, and POV-Ray renders
it
> in about 1 minute (800x600, antialiasing, three light sources). A scanline
> renderer couldn't do much better.


I guess this probably going to be quite a dumb thing to say as I know
nothing about the mathematics of raytracing, but if this is the case
wouldn't it be possible to write a pre-processing algorithm that converted
all the CSG objects into triangle meshes?

Is it the parsing that takes the time on complicated scenes? I found that
the two things that slow my renders down to walking pace were many object
(300,000 blobs in one scene) and media... I'm guessing the later is due to
excessive computational times for particle interactions and the former is
simply a bounding consideration? Would these be quicker as meshes?

Regards,

Luke

Post a reply to this message