OctaneRender™ Standalone 3.00 beta 3

[gah5118]

I also want to throw out there..... making the resolution node disconnect so we have to switch everything to film settings, has cost us a lot of time. there must be a way that a if then statement could be thrown in to update the nodes when porting over.

[calus]

The adaptive subdivision doesn't appear to work in the real world (at least for meshes) so we will probably take it out for now.

--
Roeland

Hey Roeland,

Adaptive openSubdiv works well in Maya or Max viewport,
but yes in Octane we are not there yet, for example here is a Torus in OpenSudiv GLviewer :

you can see even at lvl 1 the torus is smooth (and screen space tesselation is off),
in Octane the same torus is never tesselated whatever the lvl is.
Aren't we supposed to use an OSD tesselator with the weighted bi-cubic patches instead of using the coarse vertex ?

Also OpenSubdiv in Octane is a CPU preprocess and is a bit slow (V2 compared to some other renderers OSD),
so I guess it only use a primvarRefiner Far API layer and no OSD layer?
Why don't use stencil table and a OSD api layer evaluator with Cuda backend ? sounds to be way more faster.

Sorry if what I say doesn't make any sens, I tried to read carefully the Opensubdiv API documentation, and I understand its primary purpose is for interactive display so may be hard to use for a renderer.





If someone is interested by the openSubiv GLviewer, here is an opensudiv 2.3.4 build for win 64:
https://github.com/takahito-tejima/Open ... ster/win64