"triple_r" <nomail@nomail> wrote:
> Did you ever figure out the problem with discrepancies in the intensity?

You mean the issues I mentioned recently, with that crazy factor I found
necessary? Those are solved. I found out I had a major misconception about how
to integrate the incoming light intensity; instead of adding up the sample
values (scaled reciprocal to sample density), I basically just computed a
weighted average. With the far-off regions having much lower density and
therefore higher weight, but contributing virtually nothing but blackness, this
of course was *far* too dark >_<.

The results are still too dark for some yet unexplained reason, but since a
constant factor of 2.00 seems a perfect match regardless of choice of sample
points, scene scale, material parameters and what-have-you, I'm confident
enough that the output is right, so I'll just postpone that to later.

Instead, I'm now a bit worried about the single-scattering term; I'd have
expected it to contribute more to the overall brightness, but I don't have any
reference. I don't trust Tariq & Ibarria's results (and besides it seems that
they just vanished from the 'net these days), and the original paper by Jensen
et al. does not specify the values used for their "marble" bust shots showing
the individual components, which they custom-tailored to get an approximately
1:1 contribution of both terms.

Post a reply to this message