RBD Collisions calculation method...

   4446   4   0
User Avatar
Member
50 posts
Joined: March 2011
Offline
Hmm. Another issue to fix, as I solved the interpenetration problem, now i'm facing calculation time issues.

Here's the thing, for my caps I ran a test. Well you'll figure out my problem from yourselves….

Timeline: 240 frames (houdini default)

Caps collision mode test:

Raycast with no proxy volume and 30 dubsivisions: 1m06s

Raycast with simple geometry and 10 subdivisions: 1m06s

Implicit Box with 5 subdivisions… Can you guess???? Yup.. 1m06s

–' How the hell… Well I know Houdini is quite different from other softwares but… Really? Am i missing something again? Cause I really can't figure it out. This way bullet will become just pointless to do what i want…..

Do you have good news for me? Hope so…
Thanks in advance….
User Avatar
Member
8551 posts
Joined: July 2007
Online
As you may know Houdini has 3 RBD engines you can choose from
so on RBD object there are 3 Tabs with corresponding settings to those solvers
the tabs are called

RBD Solver | Bullet Data | ODE Primitive

so anything in Bullet data will be used by Bullet Solver which you are using currently
and everything in other 2 tabs will be ignored by Bullet solver

so because Volume settings you are changing are part of RBD Solver tab, they are ignored by Bullet therefore you are getting the same sim time
Tomas Slancik
FX Supervisor
Method Studios, NY
User Avatar
Member
50 posts
Joined: March 2011
Offline
Hmmm, as i was guessing, i missed that point… Well now i figure out that even with a dense mesh bullet is still faster than the implicit box method of rbd solver… Then how can i provide a proxy mesh for collision calculations within the Bullet Solver? Cause i'm trying to fill a bottle with small bottle caps and it's painfully slow after the 10th bottle cap is generated….. at frame 1000 the bottle was still not empty and my computer totally freezed. Well i guess i could use the auto-freeze to solve this issue but you may have a better solution for me….
User Avatar
Member
2624 posts
Joined: Aug. 2006
Offline
Satamanster
Hmmm, as i was guessing, i missed that point… Well now i figure out that even with a dense mesh bullet is still faster than the implicit box method of rbd solver… Then how can i provide a proxy mesh for collision calculations within the Bullet Solver? Cause i'm trying to fill a bottle with small bottle caps and it's painfully slow after the 10th bottle cap is generated….. at frame 1000 the bottle was still not empty and my computer totally freezed. Well i guess i could use the auto-freeze to solve this issue but you may have a better solution for me….

If you take a look at the RBD example files your question is covered.

Rob
Gone fishing
User Avatar
Member
8551 posts
Joined: July 2007
Online
Satamanster
… Then how can i provide a proxy mesh for collision calculations within the Bullet Solver? Cause i'm trying to fill a bottle with small bottle caps and it's painfully slow after the 10th bottle cap is generated…

simply simulate using low res caps, then switch to hi res after sim
here is an example

Attachments:
cap_simulation_proxy.hipnc (288.2 KB)

Tomas Slancik
FX Supervisor
Method Studios, NY
  • Quick Links