Matti Karnaattu wrote:
> 
> [...]
> 
> If I understand this right, we can calculate pretrace block size to match
> with error_bound using formula:
> 
>   pretrace (_end) = error_bound * minimum_reuse

No, pretrace_end is the pretrace block size as fraction of the render
size.  error_bound on the other hand refers to scene geometry.  Therefore
you can't give a precise best pretrace_end for every render.  It strongly
depends on the view of the scene and the camera.  

> We have to assume that low_error_factor is one. This pretrace value is
> smallest useful pretrace_end value and lower values are waste of time. When
> always_sample is set to off this value should be used to get equal detail
> when using radiosity always_sample on. Am I correct?

As already said you are making wrong assumptions from the beginning but
concerning planning the pretrace another radiosity parameter is important
too: nearest_count.  It strongly affects whether new samples are taken or
not, both during pretrace and during final render.  And always_sample does
not really turn taking of samples completely off during final trace, IIRC
it just sets nearest_count to 1.

> Radiosity samples are distributed evenly if there is one pretrace step. Is
> there any use to set pretrace_start to different value from pretrace_end
> value?

Yes, there is - quality of results tends to get better with larger
pretrace_start because the choice of positions where radiosity samples are
taken is better balanced.  The difference in quality of course might be
low compared to the slowdown.

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 28 Feb. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message