Jan. 6, 2003 01:51:01
I am curious. With Select, can non listed PRMAN compliant renderer be used?
In Mantra ROP, when either PRMAN or Entropy are selected, it places the renderer call. Same call that is placed in BASE_PATH variable under envrionment variable. So, for example, if I wish to use AQSIS, I would select PRMAN or NT Renderman in MantraROP, then simply change the renderer call to ‘aqsis’. Or is there more to to it?
Is Mantra, itself, a PRMAN compliant renderer with raytracing abilities? Can SLC and SLX shaders be used in Mantra centric SHOPs?
Jan. 6, 2003 14:02:33
By PRMAN complient do you mean a render that complies the The Renderman Spec(https//renderman.pixar.com/products/rispec/rispec_pdf/RISpec3_2.pdf
prman is the name of pixars renderman render(PhotorealisticRenderMAN).
I know select can output geometry as a rib, but have know idea if can output secens as rib.
Jan. 7, 2003 12:00:25
I may be wrong, but I thought Select didn't have RIB capability…possibly a piece of geometry snippet but that's a far call from actually being able to use with a RIB renderer per se. There's *lots* of other stuff that's needed that Houdini Master can output with Rman ROPs.
As far as what renderers work - Master works well with all versions of prman, RenderDotC, and Entropy - plus you can hack it to work with BMRT fairly readily.
I wouldn't go down the road of trying to do RIB renders with only geo snippets(if you can even make those in Select). Way too much hassle…I'd either use mantra or pay for Houdini Master(or whatever else can generate RIB - I've lost track
Jan. 10, 2003 03:57:39
OK so lets assume I have Master.
In Mantra ROP I select Renderman. Houdini places renderr renderer call in the line. So to get it to render with Aqsis, I just change the call line to aqsis right. Obviously the shaders I would be using would be SLX compiled not slc. Or renderdc or wrenderdc for RenderDotC.
Could you explain that ‘with a little hacking Houdini can be made to work wioth any PRMAN renderer’. What do you mean by hacking. As long as RIB file is legal, and the renderer suppourts functions, any PRMAN compliant renderer should parse the RIB and spit out an image.
Jan. 10, 2003 16:02:15
Well, it wouldn't necessarily work - it might, but bits might be broken. That's why the ROP has options for prman, renderdotc, entropy…despite the fact that RIB is a standard - well, there's always variations that happen with every implementation. Even prman strays significantly from the prman spec - they've added extra things that aren't in the spec, but they try to adhere to the spec at a minimum so that proper RIB won't break, per se. However, you'll be giving up certain things. With these other RIBs, there will be certain things broken. Just to give you an example - when Entropy came out - using prman RIB broke it. The reason? The Renderman spec says you *must* declare the types of params when you call a shader. For years, Pixar relaxed this and their interpretor would let you say:
Surface “SuperPlas” “Ka”
and it wouldn't break - it would pass the value to the renderer and let it decide at shader evaluation time whether the value fit the type. Houdini, understandbly, did this because it made for smaller RIBs. Feed this to Entropy, and it puked everywhere. They followed the spec more rigidly, and required:
Surface “SuperPlas” “uniform float Ka”
To Entropy, the prman RIB was nothing less than a messy syntax error.
This is why there's not just one “RIB” output. There's also lots of kewl GI/raytracing stuff associated with the latest prman release that you can't take advantage of unless you toggle on the V11 version, or hand edit the RIB…