In article <chr### [at] povrayorg> , Chris Huff 
<chr### [at] maccom>  wrote:

> It sounds like it's a completely different kind of renderer, sharing
> most of the SDL, but using other rendering and lighting methods(like
> "real" radiosity, not Monte-Carlo raytracing like POV uses).

Interesting enough, their description implies it will support "radiosity"
and POV-Ray uses a different not "real" radiosity.  Both are "radiosity"
regarding the effect they simulate, and their approach will force
tessellation for most objects.  As far as I know others have tried to
combine the classic radiosity with raytracing and the results were hardly
ever fast enough for complex scenes.

> They also don't seem to know that POV itself will be rewritten soon...

Their description shows some misunderstanding.  The fundamental raytracing
algorithm has been developed over 30 years ago, so whether it was
implemented ten years ago or ten days ago will not make any difference in
output quality or speed.

Besides, features like radiosity don't just make a difference of 40 instead
of 25 seconds to render (as said in their description).  A good
approximation of a complex scenes will take much longer, and since one
usually ends up solving a big matrix, memory and/or processing requirements
are extreme, it only makes sense for animations (because it is viewpoint
independent) but for still images the method currently used by POV-Ray will
likely give a better results on non-supercomputers.

And, right now there is no word on support of future official features like
photons, which probably add more realism to a scene than a more precise
radiosity algorithm.

In conclusion, the most useful feature they offer is the reuse of POV-Ray
scenes, but I would expect them to look just different, not necessarily
"more realistic"...


    Thorsten

Post a reply to this message