Weird rigid body activation behavior

Support Particles and Physics Simulations
1

Hello! I am trying to create a rigid body sim of roof tiles falling off a roof, but I am getting some strange behaviour. Basically, I am starting all the roof tiles as deactivated, and I’d like to activate only some of them here and there. I am testing out the system: the roof tiles all stays in place at the beginning, so they behave as expected. However when I activate a few of them, suddenly ALL the rooftiles get activated and explodes… on one side this might have to do with the size of the collision bound, maybe there’s some overlapping. However I cannot figure out why, if at the beginning they stay deactivated, it’s enough to trigger one to trigger the whole lot of them, even if they’re really far and there’s no apparent physical contact between them.

In one screen you can see the rooftiles staying in place before the big ball falls on them (sim already started). However, in the other screenshot you can see the moment the big ball is touching some roolf time, they all get activated…

image
image838×553 45.9 KB

image
image1384×912 113 KB

I tried to make the pieces really small, but the issue persist. Same with trying different collision shape option.

I should mention that the tiles are a GN generated mesh. I applied the modifier and then separated the single pieces. I also moved the pivot to the geometry for each tile. I thought it might be an issue of shared data, but also making the pieces single user didn’t help

Any idea what I might be doing wrong please?

2

If there is no problem with the center axis and scale of the object, there will be no problem.
refer to the video for simulation information

1 Like
3

Thanks. al the transforms are applied, so that’s not the problem I think. I am failing to understand what’s going on with the collisions. Although there shouldn’t be any overlap, I think there’s still something that is driving the collision system crazy.

Just tried the same setup in houdini and it doesn’t have any problem, so I wonder if the bullet implementation in blender just cannot handle this…

4

In blender, simulations are problematic if the scene is too small.

I don’t use simulations much, but the scale of the scene is often a problem.

Modeling is done on a 1:1 scale with actual dimensions, but simulation is done by increasing the scale by three to five times.

1 Like
5

I suspect there’s something that blender doesn’t like about the mesh being separated from the same one… or anyway, it’s a weird behaviour that doesn’t happen by generating the mesh 'by hand" which is however impossible in this case, too much time.

The sad thing is that I am trying to do the same in houdini, and although I am less proficient with it and so I am struggling to find how to do a couple of things, the sim works perfectly out of the box, is faster and it doesn’t blow up…

Thanks for your help so far, if I find something I’ll post it here!

1 Like
6

Could you share your file here? I’d be interested to check it out

2 Likes