Inside Sabertooth
Learn how Sabertooth uses 3ds Max to create 3D interactive projects, including HBO Go’s Game of Thrones interactive experience
  • 1/3
You are here: Forum Home / Autodesk® Maya® / Dynamics / Dynamics not evaluated during network rendering
  RSS 2.0 ATOM  

Dynamics not evaluated during network rendering
Rate this thread
 
61076
 
Permlink of this thread  
avatar
  • JuliuszK
  • Posted: 16 October 2011 11:28 PM
  • Total Posts: 1
  • Joined: 16 October 2011 10:59 PM

Hi All!
We have some problem here which started to occur only recently - basically when we render stuff on the renderfarm, we get a large percentage of frames with dynamics not evaluated. (we use it for things like wobbling ears, pony tails and so on).
Has anyone had similar issues? We’re trying to establish first where the cause is - is it renderfarm (plugin not loading, network licensing issues?) or is it a problem with a scene or software itself.
The things we noticed:
- it happens randomly - sometimes it’s 10%, sometimes 50% of frames without dynamics calculated
- when sending a job with a multiframe task, all frames in a batch will either be correct or wrong. So this must be something at the initialization process, as it happens once for entire batch.
- the same job re-submitted from Deadline (our network rendering client) gives diffrent result, so we keep re-sending faulty frames and eventually we get everything done. So the maya scene itself seems to be OK.
- everything renders fine locally, using batch render.

If you need more info, let me know.

Any ideas on what might be causing this would be highly appreciated.

Thank you,
Juliusz



Replies: 0
avatar

Did you cache all of your dynamics? nCloth, and any dynamics, won’t necessarily work correctly unless cached. Also, different network rendering solutions have varying levels of support for cached data. With some network rendering software the cache is sent to each render node as needed to render a frame. Other software requires that you copy the cache to each node manually. Some software does not support caching at all. Check your documentation.



Replies: 0
avatar

Aaron’s got it right - the reason is that when you are using dynamics, the position of any dynamic object or particle in any given frame is evaluated based on its position in the previous frame - but if you are distributing frames randomly across a renderfarm, the evaluations won’t be available consistently from one node to another.

When you render dynamics, the evaluations take place during the render unless you cache your particles and/or bake your simulations.  When you do that, the positions are recorded in a file and no longer evaluated.



Replies: 0