rendering issues: maxman & bmrt

[Mauritius]

The solution is very simple: triangulate your mesh. The reason is more difficult to explain. I quote a text I wrote for the Wings RIBBit manual which can be downloaded for free from here (http://io.plastickitten.net/Wings%20RIBBit%20Manual%20w-i-p.pdf ):

"The RMan interface requires polygons to be planar. Though most renderers don’t care whether the input matches this criteria, some — in particular those supporting ray-tracing — take advantage of it and produce wrong images if polygons are non-planar. The artifact you’ll encounter most likely in such a case will be triangular cracks in your model which are aligned somehow with the directions of your original polygons.What happened is that the first three vertices of your polygon where used to derive a plane for ray-intersection by the renderer. Additional vertices are only used to bound this plane and are essentially projected onto it, if they didn’t lie on the same plane, which is the case for non planar polygons."

In general, throwing polygons at RMan complaint renderers is not at good idea. At least not if you have a geometry that is ought to look perfectly curved (and which should include your case which will look best if rendered as a sudivison surface).
I urge you to read the chapter "Primitives and some Philosophy" in the PDF linked above.

Cheers,

Moritz

[thomaspecht]

thanks for the input. i've tried to apply this in 3ds max by different ways: triangulating the mesh with max's builtin functions and using the "MM model" modifier that comes with maxman - both to no avail.

a non-displaced model will render just fine but a displaced one always comes out with huge cracks and noise regardless what i set up. maybe a bmrt specific issue? there is another, older post regarding this that can be found via forum search. i might as well contact the thread starter if he found a solution.

by the way: bmrt does not seem to like subdivision surfaces -either it takes ages to render those or the renderer simply hangs when the mesh is being set up as SDS with maxman's modifier.
i didn't have the time to find out... ;)

right now i cannot even test how other renderman-compliant renderers behave since maxman is rather limited to only a few commercial ones and apparently does not work out of the box with freeware like aqsis.

[Mauritius]

The solution to your problem is here (http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&threadm=0tY67.108071%24EF2.14777248%40typhoon.nyroc.rr.com&rnum=1&prev=/groups%3Fhl%3Den%26lr%3D%26ie%3DUTF-8%26selm%3D0tY67.108071%2524EF2.14777248%2540typhoon.nyroc.rr.com).

As I said, RMan renderers are not very optimized for polygons and hence these may require special treatment under certain circumstances (e.g. displacement shaders).
What's more, BMRT is known to have trouble with subdivison surfaces as they where introduced into the renderer shortly before Exluna got founded and Larry Gritz, BMRT's author, focused on its commercial descendant, Entropy. From then on, the renderer has practically neither seen any updates nor bugfixes, heck it even saw downgrades (motion blur and dof where removed due to a patent Pixar holds on stochastic sampling).
BMRT hence is no longer the free RMan renderer of choice now for well over two years.
3Delight is free and this renderer is both very mature and fast as well as capable of ray tracing and limited GI. Pixie, which is even open source supports monte carlo irradiance and photon mapping.
All the other basic reasons to use a RMan compliant renderer, like e.g. perfect surface quality, fast subpixel displacement, programmeble shading true 3d motion blur etc. come for free anyway with any of the free RMan renderers which particularly includes Aqsis (and the commercial ones anyway).
I urge you to switch to 3Delight or Pixie. If rendertime is no concern, use Aqsis, they need beta testers who create nice images (not just images) badly.
Ultimately MaxMan will output RIBs and they can be sent to any renderer via command line -- not only BMRT.


Cheers,

Moritz

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.