> if a window is not bright
> enough, you cannot plce it behind the camera and expect to get radiosity
> from it.

This statement clearly indicates that you're not aware of how radiosity
actually works. It has nothing to do with the brightness of an object. It's
the size.

Samples are taken during the rendering to ensure that the *image you see* is
properly lit. With some precalculated distribution function, rays are sent
into the scene to check for lighting conditions. The amount of rays traced
this way depends on "count". The level of rebounces, quite obviously, is set
with recursion_limit.
Nontheless, only the visible area is sampled. Additionally, the sampling may
(and most probably will) miss small objects on some occasions, while hitting
them on others. Simple scenes with lots of large objects thus usually
require a smaller count than complicated scenes with small objects scattered
around in a large volume.
So, if you look at one wall, that wall is properly lit using radiosity. But
there are no samples on the other walls. If you'd simply turn the camera
around without taking additional samples, you'd see lots of dark spots, as
there are no samples whatsoever (depending to a degree upon
recursion_limit).

To ensure a more consistent lighting, you can simply render several views of
the scene and save the individual radiosity data using "save_file". AFAIK,
you can then copy-paste the content of these files into one big file and,
when rendering a final pass for the image or even animation, use load_file
to load that file (remembering to set "always_sample" to off). Unless you
have too few samples, POV-Ray won't add new ones, and the lighting should
stay pretty consistent. I haven't tried and tested this however, but it's
what some said in regard to this on earlier occasions.

As for the original post: the above described technique is whats changing
the brightness. Depending on the placing of the samples, they may yield
different results, depending on if they hit bright objects, or not. I
imagine the relation between viewing angle and sampling in regard to surface
normal is also a factor, but I'm not sure about that part. Nontheless, a
slightly different setting of the scene returns different radiosity sampling
and thus, different lighting conditions.

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

Post a reply to this message