I'm running a packed RBD sim on a prefractured object with about 17,000 pieces, some large and some very tiny.
The trouble I'm having is that when things get going, Bullet is generating so much impact data that I am running out of memory on a machine with 64GB of RAM.
I was able to get a previous version cached where there were about 10,000 pieces and fewer tiny ones.
Is there any way to trim the kind of impact data that is generated? I'm using it for secondary effects but I probably don't need *every* little collision that occurs.
-Jon
H14 Bullet generating too much impact data
4236 4 2-
- jparker
- Member
- 321 posts
- Joined:
- Offline
-
- cwhite
- Staff
- 794 posts
- Joined: Oct. 2012
- Offline
That memory usage sounds suspiciously high - I'd suggest sending a bug report with that file to support.
Turning off “Add Impact Data” will prevent the Bullet solver from recording anything to the “Impacts” subdata on the DOP object - that can reduce the memory usage somewhat if there are many impacts and you don't have any need for that data.
Turning off “Add Impact Data” will prevent the Bullet solver from recording anything to the “Impacts” subdata on the DOP object - that can reduce the memory usage somewhat if there are many impacts and you don't have any need for that data.
-
- jparker
- Member
- 321 posts
- Joined:
- Offline
Okay, I was kind of suspicious about it, too. I'll have to untangle the sim from all the shot-specific assets and settings and send it in.
Maybe if it is the impact data causing the memory problems I'll just come up with a post-sim solution to imitate the regions where the impacts take place. They don't need to be too specific, anyways.
Maybe if it is the impact data causing the memory problems I'll just come up with a post-sim solution to imitate the regions where the impacts take place. They don't need to be too specific, anyways.
-
- jparker
- Member
- 321 posts
- Joined:
- Offline
Sorry to not follow this up with a real bug report (still haven't had time to clean up the file).
The memory problem went away when I stopped writing out the impact data with a SOP. My memory usage is a lot more sane now. So maybe there's a bug in there somewhere (sim data -> sop file output).
Unfortunately… I lost the impact data, so I'm working around it. There are a few ways to approximate where the impacts happen so it's not a huge problem this time.
The memory problem went away when I stopped writing out the impact data with a SOP. My memory usage is a lot more sane now. So maybe there's a bug in there somewhere (sim data -> sop file output).
Unfortunately… I lost the impact data, so I'm working around it. There are a few ways to approximate where the impacts happen so it's not a huge problem this time.
-
- jason_iversen
- Member
- 12994 posts
- Joined: July 2005
- Offline
jparker
Unfortunately… I lost the impact data, so I'm working around it. There are a few ways to approximate where the impacts happen so it's not a huge problem this time.
Perhaps kook into the “Impact Analysis DOP” - this filteres down Impact information nicely. I'm not certain, though, that it will reduce the Impact data; it might only add new data which is less intense.
Jason Iversen, Technology Supervisor & FX Pipeline/R+D Lead @ Weta FX
also, http://www.odforce.net [www.odforce.net]
also, http://www.odforce.net [www.odforce.net]
-
- Quick Links


