stbenge <^@hotmail.com> wrote:
> This problem is caused by something else. I ran into it again last
> night. The radiosity count was at 300 when it happened. As it turns out
> the problem has to do with my setting for adc_bailout. I had it set to a
> value of 0.1. I changed it to 1/256 and the black splotches are gone! I
> found no mention in the docs about such behavior, but it makes sense in
> a way.

Hum... why should it make sense to get black splotches if your adc bailout is
too high? ;)


> I don't know why raising the count would make black splotches
> suddenly appear though, but since I have a fix I don't really care at
> this point. I thought I'd let you know about it, just in case it's not
> supposed to work that way.

It is indeed not supposed to. But I expected something like this from the code -
I just didn't bother to try and come up with a test scene - instead I just fixed
that thing... ;)

It's quite a similar problem as with the max trace level: In both cases, if some
ray caused a sample to be taken, the relevant values (current trace level and
weight) were "inherited" by the sampling rays. If the origonal ray happened to
be reflected, these values would leave comparatively few trace iterations
(either due to the trace level, or due to the ADC bailout). But for the
surrounding samples the original rays might have come directly from the camera,
equipping the sampling rays with both a higher number of trace levels to go and
a heavier weight, so they might see a mirror or window where that other sample
just saw a pitch black square.

Post a reply to this message