ok, so I'm trying to do a simulation where there are maybe 20 balls, then 10 balls within each of those balls, and they will just roll around and collide with one anther … think lottery balls within lottery balls.

Anyway, I'm messing with the invert sign, and the different collision models, and so far the only thing that seems to work is if for the outer spheres I model two spheres, then make a hole, then weld the stuff together so that it's a continuous surface and turn off laser scan.

This method seems a bit slow, and I don't like the little hole in there so I was wondering if there was a better method for “inner” collisions. The invert sign works great and quick until the outer balls bang against each-other and then they pass though one another.

Any ideas?

Have you tried not using the Volume representation at all? You can deactivate the Volume stuff with a toggle and it'll use the Thin Plate collider to solve for polygon collisions. Try it and tell us if you get better results?

ah ok, I'll try that next … right now I'm doing two separate DOP passes which are dependent in order .. in and out of SOPS … one is just the outer balls, and the other (separate sim) is the inner with reverse sign … seems to work but haven't tried the thin plate

ok yeah thin wall works much better for some things. So is thin wall the “actual geometry” in the sim, similar to how particles collide?

It does seem to get “stuck” or go through the geometry more often but a very strong force from a forcefeild can be used with a very short limit distance to push the stuff back (by normal or inverted normal) far enough to get the next collision when using gravity.