[ODE] Bouncy sphere/trimesh collisions

[Geoff Carlton]

As Jon stated, the problem is the penetration.  A large penetration 
causes a large correcting velocity, and that makes the object bounce 
up.  I believe the result is something like:
    final_velocity = max(correcting_velocity, surface_bounce)
So even with the surface parameters set to 0 bounce, the object will 
still bounce after penetrations.

Apart from ERP, look at these parameters:
    dWorldSetContactMaxCorrectingVel
    dWorldSetContactSurfaceLayer

Playing with these parameters can eliminate almost all visible bounce 
from the simulation.

Geoff


Jon Watte (ODE) wrote:

>
> ODE has to apply some force to make the geoms separate, because 
> they're already interpenetrating at the point of collision. The 
> momentum imparted by that force is retained once the objects have 
> separated. I believe you can attempt to make it better by reducing 
> ERP. Another approach would be to detect the interpenetration, and do 
> your own position constraint resolution instead of a velocity/force 
> constraint solution -- however, if you take that to the extreme, you 
> get a totally separate solver from the ODE solver...
>
> Cheers,
>
>             / h+
>
>
> Brad wrote:
>
>> I've got a sphere that collides with a trimesh at a fairly high 
>> velocity (though not high enough to make the sphere go through the 
>> trimesh or anything). I've got bounce set to 0. Sometimes when the 
>> sphere collides with the trimesh (which is a fixed, bodyless 
>> geometry) the sphere bounces up pretty high, and other times it works 
>> right and pretty much doesn't bounce at all. Is there something I'm 
>> doing wrong?
>
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
>