[Bf-cycles] Micropolygon implementation

Brecht Van Lommel brechtvanlommel at pandora.be
Mon Dec 30 20:43:23 CET 2013

I'm well aware of micropolygon rendering and it is great for
displacement, but it's problematic for global illumination and path
tracing (note the paper only shows direct light). The industry trend
seems to be to move away from it and towards path tracing, due to more
interactive previews and simpler lighting workflow. As I understand it
even Pixar didn't rely on it that much anymore for Monsters University
in the sense that they had all geometry in memory for raytracing
rather than generating it on the fly and discarding as in REYES.

However, we can make our displacement workflow much better, I just
don't think that REYES/micropolygons is what we should focus on.
There's ideas that we can use and distinctions get fuzzy because
modern production renderers aren't purely one or the other, but it's
definitely been a conscious decision on my part to approach this from
the path tracing side.

You can read e.g. these articles to get an idea about the trends in rendering:

On Mon, Dec 30, 2013 at 7:07 PM, x3108 at chello.at <x3108 at chello.at> wrote:
> Hello,
> maybe you know, but what if this slipped trough your fingers…
> i stumbled some years ago over the renderants papers and its reyes micropolygon GPU implementation.
> http://www.gaps-zju.org/project/renderants.html
> Now as i understood the papers and some code is official ?
> Would be the actual cycles code be feasible for such an implementation or needs it a too complex overhaul ?
> Look at the GREAT displacement images (ohhh.. :-D):
> http://www.gaps-zju.org/project/renderants.html
> Imagine the displacement speed. I mean micropolygons would be definitely a way to go for the cycles roadmap ? ;-)
> This would make Cycles definitely THE displacement solution on GPU, and in terms of speed would beat Reyes stuff like PRman and 3Delight hands down!
> enilnacs
> _______________________________________________
> Bf-cycles mailing list
> Bf-cycles at blender.org
> http://lists.blender.org/mailman/listinfo/bf-cycles

More information about the Bf-cycles mailing list