Particle collisions go crazy on collider
#1
I have particles split with a cluster with gravity applied and particle bind hold each together, when they hit the ground collider they go really crazy. Cuda collision on partcle bind stops the problem but it's very slow so would rather use without.

Thanks


Attached Files
.mp4   collision-faulty.mp4 (Size: 5.68 MB / Downloads: 46)
  Reply
#2
Can you post the file?
  Reply
#3
(02-29-2024, 02:16 PM)tyFlow Wrote: Can you post the file?
Sure

https://www.dropbox.com/scl/fi/03u07dky0...blmr6&dl=0

Thanks!
  Reply
#4
In this case I would recommend decreasing your bind solver steps and collision/bind stiffness...the problem is occurring because particles are essentially getting tangled during the collision step, and the strong forces trying to untangle them are introducing all the extra motion...and because the binds are unbreakable, the clumps of particles sort of fight against themselves and spin around.

I'm not sure what kind of material you're trying to simulate, but here's a scene where things don't explode...and I enabled bind breaking for small stretch/compress values so that if any particles end up in a state where they cannot untangle themselves...they just break. The result is much more stable.


Attached Files
.max   particle-test_02.max (Size: 692 KB / Downloads: 90)
  Reply
#5
(02-29-2024, 05:33 PM)tyFlow Wrote: In this case I would recommend decreasing your bind solver steps and collision/bind stiffness...the problem is occurring because particles are essentially getting tangled during the collision step, and the strong forces trying to untangle them are introducing all the extra motion...and because the binds are unbreakable, the clumps of particles sort of fight against themselves and spin around.

I'm not sure what kind of material you're trying to simulate, but here's a scene where things don't explode...and I enabled bind breaking for small stretch/compress values so that if any particles end up in a state where they cannot untangle themselves...they just break. The result is much more stable.

Ok thank you for the file and info!
  Reply


Forum Jump: