concave hulls, kinematic rb's, and collision proxies

[nFrame]

Hey guys, I'm working on a shot right now with a vehicle busting through a living room wall.

I'm using proxy collision geo for the vehicle for obvious reasons, but am having trouble getting the collisions down. please read on before suggesting untick the "polygons as convex hulls" box

I'm keying the vehicle blowing half way through the wall in sops, then letting dops handle the rest (with use deforming geo on). I'm using an active value dop to switch between kinematics and simulation on the vehicle. The vehicle is a proxy cage, modeled in softimage. It's perfectly manifold, sealed, and has nice quads.

now here's the wrench in my head, when the active value is 0, it seems to not respect the polygons, and my so called "concave hull" does not collide at all. It only starts to collide when the active value is 1. If I use "polygons as convex hull" it works, but obviously there's issues with the collision geo.

even when I replace the vehicle proxy with a simple box (about the same size) it has the same issue.

If I turn off "use deforming geo" and just use initial velocity to drive the vehicle through the wall, it works just fine, and seems to respect the concave hull of my proxy geo.

can anyone else replicate this? or know of any workarounds? I cannot share my scene file, but I'm sure I could throw something together if need be.

Best!

[anim]

it'd be easier if you posted some example to begin with to see exactly your problem

here is an example of concave object breaking the wall while switching from keyframed to simulated at frame 44, without using deforming geometry (as you don't need it if it's only reason in your case is to get animation in dops while inactive)

rbd_activate_concave.hip

[nFrame]

hi Tomas, thanks for the reply & your file

I've managed to figure out the concave hull issue, and will end up using only keyframes for the final vehicle animation. I'm fairly certain what I thought was a concave hull problem was actually related to some strange relationship(bug?) between the glue cluster sop strength, and the glue network constraint dop strength. I can't really make sense of the two strength settings, and am going in circles trying to figure them out. I don't seem to remember having this problem in 12.1. actually, I'm fairly certain you could just leave the glue network dop at -1 strength, and the glue cluster sop strength would take care of things. Has anyone else noticed anything strange?