Quiet an extensive set of questions you've got there...

> 1. What is a common trace time for complex images (lots of reflection /
> refraction, many objects)

This truly depends on the scene, how much the objects interreflect each
other etc.

> 2. How exactly to use adc_bailout?

I never fiddle with that one unless using Focal blur.

> 3. Aborting?

AFAIK, POV-Ray automatically saves to disk, but in bunches. I believe that
can't be properly controlled (in terms of every 25 pixels or such), but
rather in Bytes. But someone should correct me on this.

> 4. Pointers / hints?

I've got none.

> 5. Reasons for scanline renderers being faster - is scanline better?

Scanline has some advantages that a raytracer doesn't have. For one, it
makes use of the GPU's it may find on a Graphic card. Most of the
calculations don't require heavy floating point calculations, which is
pretty slow on some processors and, AFAIK, not a part of the CPU which is
needed that often, hence it isn't as accessible and quick like other
operations. The main performance part lie in those two points: making use of
the graphics hardware and the slow performance when comparing that to the
floating point calculations. Just look at typical PC Games, they're scanline
all over the place, and they've sometimes got tons of triangles to display.
But then again, scanline simulates a lot and doesn't always have the precise
calculations raytracing offers.

Regards,
Tim

-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>

Post a reply to this message