Real-Time Physics Simulation Forum

Congratulations! good to see you and Bullet getting some recognition.

I'll get the heightfield demo working and let you know.

oops. sorry .looks like i added my original project version rather than the instrumented one. try this.

If you're using eclipse, you might find this project setup helpful :

http://code.google.com/p/bullet-xna/dow ... akechanges

i've included a patched version of the jbullet jar below

it should be in lib/jstackalloc/stack-alloc.jar as part of the normal jbullet distribution. Have you built JBullet from source before? if not it's a little tricky making sure that the instrumentor runs correctly on it. If you're using eclipse, you might find this project setup helpful : http://code....

Give this a try... (you'll need to rename the .txt to .java) , it's based off my c# version and hopefully I've found all the stack.allocs needed for JBullet. As it stands it expects the heightmap data as a float[] but you can change that.

Yep exactly. In which case btHeightfieldTerrainShape should do the trick. Just had a quick look and seems as though it doesn't exist in JBullet as it stands, which is a bit annoying. It's a pretty simple shape collision shape though so shouldn't be too difficult to add in. I'll see if I can drag tog...

Depends on the form of your terrain really. Theres a grid/heightmap based shape, or failing that bvhTriangleMesh for more complex versions.

I was looking at the : "Why all Spheres does not falling down? Instead of it, they are balancing on eachother. And only when I move the lower Sphere, they are doing something like fall. But actually they does not moves in Z direction, only X and Y. So why it is?" part and checking the beha...

hmm. running out of ideas I'm afraid. Can you try leaving the default angularDamping and restitution values alone and see if that helps?

Did it change your simulation at all? and just checking you made the change in the addBall function as well?

You most likely need to use the GImpact shapes instead. Have a look at that demo it should have what you need.

Yep, though should be :

RigidBodyConstructionInfo ballBodyConstructionInfo = new RigidBodyConstructionInfo(1,ballMotionState,ballShape,localInertia);

I think.

Been a while since I looked at JBullet, but I think your code for CalculateLocalInertia on your shapes is wrong.
You need to pass in a Vector3f to that call so that the values can be set, that should then be passed to the RigidBodyConstructionInfo.

You could modify the libgdx code to give you more access to the callback info, something like the below, though it's not particularly thread safe. That would give you the hit point and contact normal. But you'd still need to use that to decide which face had been hit. public static ClosestRayResultC...