Unexpected extreme bounciness

i've been playing around with creating dynamic and static physicsnodes, and am trying figure out a bit more about how the geometry works. i've managed to create functional scenes by attaching the object model as a child to the physicsnode and then calling generatePhysicsGeometry(), but run into some odd results when calling generatePhysicsGeometry(true)  (btw: i discovered that if no argument is given it defaults to false).

If i try to generate geometry that is triangle accurate(true parameter) on my static node (or any other for that matter), then when an object collides with it, it bounces much more than it should. bouncing appears normal when geometry is computed with the false parameter, which just simply puts a box around the object. this approach works if the object is a simple floor rectangle, but not if it is a more complex shape.

in order to try to circumvent this issue, i took a look at the TestTriMesh example, and implemented physics meshes as outlined. the geometry works properly for the object, but the bouciness issue still remains.

has anyone else run into a similar situation and discovered the reason for this?  :?

Well, actually…

I've been testing the physics in my game lately, and my character box was very bouncy, even when I turned bounciness almost completely off… I've been using triangle accuracy. I don't know the cause for this problem, although.

I never noticed that. Maybe you should post it on the ODE mailing list to ask if it is an ODE issue? Can't imagine a reason why jME Physics is causing this.

Thanks for the replies!

I'll see if I can find anything out on the ODE mailing list. I've run another test, and have discovered that this also occurs between two dynamic objects that use meshes for collision bounding.

Does that trigger any other thoughts?


It turns out that the trimesh/trimesh collisions don't work right in ode, and its a known bug. this being the case, i've decided to not use trimeshes except for critical purposes, but i've noticed a similar issue with two box-bounded dynamic physics objects trying to stack.

In my scene, i've created a floor with a box on it, and a box on top of that. when the simulation starts, the boxes should drop a bit and stack properly. however, the boxes jitter around a bunch after starting the simulation.

i'm going to start playing around with the collisions, but that doesn't really seem like the right way to rectify the situation (by coding to not react to the colliding object or so). thoughts?


scratch that, the boxes just needed some mass