[Bf-committers] Does Cycles waste half it's possible performance or am i wrong?

[Brecht Van Lommel]

On Sun, Oct 28, 2012 at 1:49 AM, Tobias Oelgarte
<tobias.oelgarte at googlemail.com> wrote:
> But if you have samples with errors, then this error will only
> contribute to one pixel. Especially caustics will take a long time to
> get clean if they are a seldom occurrence, because a neighbouring pixel
> might just miss nearly any time. If you would send in rays which "start
> at subpixels" of the film, then it is much more likely to get a better
> average (lower error), because most samples contribute to multiple
> pixels. A firefly found at the film-grid-intersection would evenly
> affect at least four pixels, giving that ray four times the probability
> to contribute the same result to all those pixels. Rays close to the
> center of a pixel will loose that effect. That way it would be about 2
> times less noise in average.

In fact this method can reduce noise in some situations, since uneven
sampling of the pixel filter can introduce noise. It also avoids
rendering padding pixels on tiles which cost render time too.
http://lgdv.cs.fau.de/publications/publication/Pub.2006.tech.IMMD.IMMD9.filter/

But regarding the 2x less noise, there's two wrong assumptions. First
is that you can't just average 1 and 4, it would be closer to
something like 1.333x I think (the fireflies at the center of the
pixel will stile take long to get rid of even if some others get
easier).

Also, the shape of the gaussian function makes it so that for typical
filter widths most of the samples contribute little to other pixels,
which lowers the noise reduction further. At the corners they
contribute evenly yes, but there's no linear transition to the center.

Brecht.