Struggling to understand predictability of particle movement

Struggling to understand predictability of particle movement with region interaction

I'm having a hard time understanding particle movement in what I thought should be a relatively straightforward setup. Basically, a ship flies thru a meteor shower, some bounce off the ship and spawn more particles. Yes, it's the Jupiter 2.

It seems Fusion Studio 9.0.1 only uses the particle point position to calculate region interactions even if the particles are something larger than a point. Most particles behave as expected, but a visible percentage either fly partially or totally thru my collision mesh region even with the sub-frame calculation accuracy cranked all the way up to 10. Also, some particles seem to get stuck on the collision region border and slide along it, sometimes along their velocity, but sometimes in complete or partial circles around the surface before detaching at some random angle.

I've tried high poly meshes, low poly meshes, smoothed, non-smoothed, I've tried adding additional directional forces or turbulence to give the sticky particles an extra shove to get them off the mesh surface, moving the mesh as well as the particles, just the mesh, just the particles, etc. The results just seem inconsistent no matter what I do. Changing nothing but reseeding a node will only change the point in the simulation where the wackiness happens, and it's not at all predictable when or why. Is Fusion's particle system just not accurate enough for this kind of thing?

Any pointers would be greatly appreciated.

Erich

Edit: Not sure if the mesh carries thru with the code. Get it here: https://www.dropbox.com/s/fg3p0rkr3kh5v5u/collision_Jupiter2_tripled.obj?dl=0


Sample of wha't going on.

Comp for that video is too big to post the code directly, so download it here:

https://www.dropbox.com/s/6l865vsacibppgn/J2_meteor_regions_test02c.comp?dl=0

I suspect that Fusion is not capable of calculating the intersection between the ship mesh and all the replicated asteroids meshes. It would take forever i guess so it's limited to the particle pivot only.
To address the issue, assuming that the asteroid mesh is enough regular, id' use an invisible mesh as a bounce region, built from the original ship mesh with an external thickness that will make the asteroids "look like" they're bouncing on the original spaceship hull.

Edit: By the way, trying to render your comp hard crashes my Fusion. Don't know if it's due to the particle system or the missing of the ship data. Anyway, i guess this should be considered a Fusion bug...

Umberto,

Thanks for the idea, but having a larger collision mesh unseen by the camera is how the comp is already set up. I added the replicated Shape3Ds just to better see what is going on. The same mesh penetration and strange behavior also happens when the particles are only points or blobs.

Here's a link to a Dropbox folder with both the comp and the obj mesh. Could you try it again with the obj mesh loaded please? If it still crashes I'll send it to tech support.

https://www.dropbox.com/sh/e5nwh4pupgjivt0/AACUwen3b6ZTmzCBTLmtWGQfa?dl=0

Thanks,
Erich