Rendering crash...

[Diurno]

Hi all, i've a scene with a crowd simulation (a very large crowd) and when i try to render it, it crashes randomly after some frames. It takes a long time to preprocess the render, then the render itself it's not bad at speed. I can't understand if i've done something wrong to make it crash or if my machine is not enough powerful to handle that simulation...

[xsisupport]

If it can render some frames, then use xsibatch to render the scene is chunks eg frames 1-5, 6-10, 11-15, and so on...

[Diurno]

mmmm i never use the batch rendering, but if it should help i'm ready to try.
The thing is that i'm reading now the documentation, but i can't understand what i've to write in the command prompt.
the syntax should be:

xsibatch -render C:\Users\User 17\Desktop\Softimage Projects\Crowd\TestCrowd1.scn
[-frames <1-10,11-20,21-30>]

my scene is 550 frames so obviously i will go on after 30 from 10 to 10 until 550.

[Diurno]

nvm, i figured it out, the correct syntax is:

xsibatch -render "C:\Users\User 17\Desktop\Softimage Projects\Crowd\Scenes\TestCrowd.scn" -pass Default_pass -frames 1-5,10-15 (and so on).

Thanks, i hope tonight i can get some sleep without put my alarm every 30 min to check for crashes

[xsisupport]

You need a separate xsibatch job for each frame set.

Something like this:
http://xsisupport.com/2011/07/19/creating-a-batch-file-for-rendering-frame-sets-with-xsibatch/

but I can't remember if that continues on when one crashes; I know I answered that question before but I have to find the answer again

[Diurno]

well, before i tried with the Command Prompt of Softimage, if i have to create a script, i really don't know where to start...

[csaez]

Are you caching the sim before render?

[Diurno]

Are you caching the sim before render?

I tried with caching by the option in the CrowdFX property, but it doesn't help at all, when i scrub the timeline i got every time the "simulation time", and i'm sure i've done it right cause i've done 10 times to be sure...

[Diurno]

i tried to batch render the scene but after 1 frame it stopped, no error message or whatever , it simple give me the new line to write....

EDIT: Right now i'm trying to cache the simulation with the FBX export... hope it solve my problems

[Diurno]

Update: with the Fbx export the crowd seems much lighter to render, but there is a problem, i followed XsiSupport tutorial to randomize the color of the shirt and now in the fbx is gone and i can't apply on it cause the particle cloud is gone in the fbx....

[Diurno]

Update: i've done again the simulation, this time cached by Fbx, and the rendering is much faster, but still crashes randomly.... can't see why... There is a after crash log somewhere to read the error?

[Hirazi Blue]

Not sure about a real crash log, but something you could try is turning on the Mental Ray Messages to full and log them to a file as described here. That way you can study the log file you've created after the crash, which MIGHT help (at least to figure out what Mental Ray was doing just before the crash occurred).
;)

[Diurno]

Thanks, i'll give a try, but i'm starting to think that a 10 milion triangles crowd simulation is more than my machine can handle

[Diurno]

i tried the batch render with all the log information active, the last 10 lines before he crashed were those:

# INFO : JOB 0.8 52 MB debug: memory increased by 64 bytes to 17292956 byte
s
# INFO : GAPM 0.10 52 MB vdebg: JOB_EXEC split object for item bba
# INFO : MSG 0.16 52 MB debug: WIN_NT: setting thread to non-overlapped sock
et operation
# INFO : MSG 0.16 52 MB debug: thread 16 running
# INFO : MSG 0.16 52 MB debug: thread padding address 00000000A4F0DC50 size
7168
# INFO : MSG 0.16 52 MB debug: starting thread
# INFO : JOB 0.16 52 MB debug: starting new thread

C:\Program Files\Autodesk\Softimage 2013 SP1\Application\bin>

[Hirazi Blue]

Oops, I must admit that to have been less helpful than I hoped...
Were there any warnings and/or errors (earlier on) in your message log file?

[Diurno]

Oops, I must admit that to have been less helpful than I hoped...
Were there any warnings and/or errors (earlier on) in your message log file?

nope...it simply....stop....