[Radiance-general] RE: Ambient file Q's (Fitzsimmons, Rob)

[Bruce Sounes]

Rob, 

Oh so familiar! Rob's comment on using mkillum is right, I think. It
looks like you have no direct sources - primary or secondary -
contributing to the illumination of surfaces visible in the scene. You
should be able to achieve far better results with lower ad's ab's etc.

Please someone correct me, I'm speculating much of the following: 

The ambient resolution (ar) corresponds to the typical spacing of
calculated ambient values (octree size/ar for their approximate spacing,
and the size of resulting splotches). 

In corners, sploches arise where two adjacent values do not adequately
represent the soft gradation into the corner - hence the too dark/too
light splotches. This suggests upping the ar value. The aa works as a
kind of refinement toggle, forcing additional samples when the
difference between adjacent ambient is greater than a factor of aa. 

Splotches occurring arbitrarily, for instance the one in the foreground
just to the right of the wall, suggest inaccuracies in the ambient
calculation, most likely because of insufficient samples - therefore
more ad's and as's may be necessary. However I think random sampling
falls down without a direct source at least one, maybe two bounces away
from visible surfaces. 

Every project seems to generate a new problem. I'm constantly trying to
figure Radiance out. 
  

Bruce Sounes