[Bf-committers] Possible Render Noise Filter?

Brecht Van Lommel brechtvanlommel at pandora.be
Sat Nov 17 22:57:59 CET 2012


I think such a method would be useful for preview rendering, to get a quick
idea of the lighting of noisy renders.

However for final rendering it doesn't look very useful in its current
form, it would need a feedback loop with the raytracer to guide sample
placement, and a reliable error metric to decide where to use more samples.
Such smoothing of samples is not used much in production rendering besides
irradiance caching, and that's only for indirect diffuse light, where you
then have to very carefully pick an error metric to decide when to
interpolate and when to take more samples (using shortest hit distance,
irradiance gradients, neighbor clamping, ..).

This method generalizes that sort of thing to glossy, depth of field, etc,
.. which gives a good improvement for few samples but there will be
diminishing returns for more samples.

Brecht.


On Sat, Nov 17, 2012 at 10:24 PM, Jason Wilkins
<jason.a.wilkins at gmail.com>wrote:

> Ack, well, if it isn't available for use in open source projects I
> guess that would be that.
>
> As for quality issues, the techniques makes rendering certain things
> *possible* because a scene that would take a year to render can
> instead take a day or two.
>
> I was also thinking it would be useful for improving the appearance of
> Cycles rendering in the viewport, since you could generate fewer
> samples but still generate fairly noise free and representative image.
>
> If I have time I'll investigate that further, since there seems to be
> room to innovate further with their technique.
>
> On Sat, Nov 17, 2012 at 12:20 PM, Harley Acheson
> <harley.acheson at gmail.com> wrote:
> > I had assumed that some stumbling blocks with this particular one is that
> > it is "protected by copyright", shown there "for reasonable academic fair
> > use" and that "the ideas presented in this paper are available for
> > commercial licensing through the UNM technology transfer office"...
> >
> > Harley
> > _______________________________________________
> > Bf-committers mailing list
> > Bf-committers at blender.org
> > http://lists.blender.org/mailman/listinfo/bf-committers
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>


More information about the Bf-committers mailing list