Mark James Lewin wrote:
>
> This light was square and used 25 spotlights are per Kari's suggestions.

For this one area_light D*x, D*z, 5, 5 jitter adaptive 0 would be
enough. I would say that if the nearest object is within 3*fade_distance
and more accurate lighting is needed then divide the source. Try it first
with one source and see what it looks like. cornell.pov in the beta
distribution can be used to see how dividing the source affects the lighting.
The N parameter divides the source to N^2 sub sources. N=3 was mainly
used to get better lighting on the front of the tall box, the rest
of the scene is ok with 1 source. Notice how the light intensity remains
constant even when the source is divided without altering source brightness.
It's because the light source color brightness is actually brightness/area
so when a source is divided to smaller sources they still have the same
area brightness.

> I used radiostity with a recursion limit of 5 (probably a little excessive)
> and here are the results.

Looks good and a fortunate coincidence is that the scene contrast
ratio matches nicely that of a typical monitor so no post processing
is needed :) 


_____________
Kari Kivisalo

Post a reply to this message