Weird unexpected bounce behavior

Hello,

I stumbled across a really strange rigidbody-bounce behavior that can't possibly be intended like that.
I've got the following setup:

I have spheres with a rigidbody, no drag, angular drag or gravity and collision detection set to continuous dynamic as well as a physic material with static friction, dynamic friction and bounciness set to 0.5 (everything above 0 generates similar results) and both combines set to maximum.


Those spheres hit a box at a 30° angle without colliding with each other. Each sphere's velocity is 0.01 higher than the previous one. The result is this:


I also used animation curves to graph the sphere's speed and its angular velocity after the impact in relation to its initial speed:

Speed:

Angular Velocity:

So the behavior is very strange in general but at a certain speed, the spheres periodically don't get any angular velocity at all from the impact, which makes it even weirder.

I also tried to lower the fixed timestep from 0.02 to 0.01 but that way the only thing that changes is that the "no angular velocity behavior" just starts at a lower initial speed.
And I tested using continuous speculative collision detection, but that leads to a whole different, but also very strange behavior.
I've put graphs to those tests into a zip file because I can only have 5 files in a post.


So my question is: What the hell is happening here?

 

Just curious, if you leave a little more time between spawning spheres does the behavior change? (Are they hitting each other?)

 

Maybe there is some code in Update? I haven't specifically tested this, but I think that even the instancing code should run in FixedUpdate.

 

Thanks for the replies!

They shouldn't collide with each other. I put the spheres on a layer and disabled collision with the same layer in the layer collision matrix. But I made another test with more time in between the spawns, just to be sure

The interesting thing is that it doesn't seem to matter, whhere/when the spheres are spawned, because the same speed always results in the same behavior. But I made a new test where I spawn the spheres in FixedUpdate. Just with eight different speeds this time (looping): 3.5, 3.7, 3.9, 4.1, 4.3, 4.5, 4.7 and 4.9. This also nicely shows, how inconsistent and unintuitive the bounces are, depending on the collision speed.

 

This is definitely interesting in relation to this thread:
https://forum.unity.com/threads/experimental-contacts-modification-api.924809/

 

I've noticed another strange issue but it could be something I'm doing wrong as I'm new to Unity.
What do you think of this behavior ?
Video below :

 

That's related with the parameter "Bounce Threshold" in Project Settings > Physics. When the relative velocity of a contact is below that threshold then the object won't bounce.

 

Thanks, glad there's a way to fix this!

 

Did you find a solution for this? I don't know the size of those balls but maybe they are small balls. If this is the case, increase the ball size to 1 might help