Bullet 2.82 Sphere/TriangleMesh restitution issues.

[programmerjeef]

Hello, this is my first post on the forums.

Currently I'm using the Torque3D game engine with Bullet physics version 2.82. I have a sphere and a triangle mesh, with restitution on the sphere as 0.5 and on the triangle mesh it is 1.0. I go up about 100 meters or so, and when I fall, some places on the mesh it bounces correctly, but on other places it does not (bounces very little/barely).

I have looked everywhere, probably about 10 hours of research into this by browsing a lot of the old topics about sphere restitution issues, and have tried things such as having more iterations per step, enabling splitImpulse, setting the restRestitution to 1e30, and change the btSphereShape to use btMultiSphereShape with a local transformation at the origin.

My current setup im using in Torque3D regarding bullet:

- The timestep torque simulates physics is on a fixed update with a delta of 32 milliseconds per tick off of a process list.
- using a dif file that was transformed into dae then dts (dts and dif [obsolete] are proprietary file extensions for models specific to Torque)
- Sphere radius is 0.2 units
- I have CCD enabled
- I have also had to modify the triangle mesh to make it stop bouncing the sphere when it rolls over edges that is listed in these pages:
https://code.google.com/p/bullet/issues/detail?id=676
http://bulletphysics.org/Bullet/phpBB3/ ... f=9&t=9977

****Note on the second link, I only did the bottom part, not the top part as that didn't fix the issues.****

I am trying to make a marble platformer game.

Any help would be REALLY appreciated. I have been messing with physics for over a year, bouncing between physics engines such as the default physics in Torque3D, to physX and even ODE. I really do not want to change again, and would like to stick with Bullet.

Thanks