05-08-2014, 06:53 PM
This is an oldy-for me. I used to work in studio's with Max (windows) and i first saw this there with the scanline renderer. (I've done a lot of smoke in my time. 
)
But I had the same issue last week in maya/mr/linux/render batch and the circumstances are the same. I believe it suggests a memory issue with multiscattering.
Common cirsumstances.
1) multiscattering is 'on'
2) there is a certain quantity of smoke in the fumefx grid-its disspating and at the point were its nearly completely gone-but *not quite*- on that given frame - the software will crash the render with a memory error.
The frame cannot be completed (128 bytes it should be 7.0MB)
3) uncheck multiscattering and that same frame renders normally. The entire sequence is fine. Thats my workaround.
Here is a test scene - repro:
1) different resolutions mean the 'bad frame' shows up at different frames.
At this resolution (2) if you cache it locally the bad frame should be on -or very close to-frame 62.
2) Do a render batch of just frame 62-if it renders correctly (other than 128 bytes) try the entire 68 frame sequence. A higher resolution (0.5) had the circumstances happening at frame 50.
3) If you can repro then simply uncheck the multiscattering and you should be ok.
If you find a cure-I bet it accounts for a few reported render crash issues.
) But I had the same issue last week in maya/mr/linux/render batch and the circumstances are the same. I believe it suggests a memory issue with multiscattering.
Common cirsumstances.
1) multiscattering is 'on'
2) there is a certain quantity of smoke in the fumefx grid-its disspating and at the point were its nearly completely gone-but *not quite*- on that given frame - the software will crash the render with a memory error.
The frame cannot be completed (128 bytes it should be 7.0MB)
3) uncheck multiscattering and that same frame renders normally. The entire sequence is fine. Thats my workaround.
Here is a test scene - repro:
1) different resolutions mean the 'bad frame' shows up at different frames.
At this resolution (2) if you cache it locally the bad frame should be on -or very close to-frame 62.
2) Do a render batch of just frame 62-if it renders correctly (other than 128 bytes) try the entire 68 frame sequence. A higher resolution (0.5) had the circumstances happening at frame 50.
3) If you can repro then simply uncheck the multiscattering and you should be ok.
If you find a cure-I bet it accounts for a few reported render crash issues.