Real-Time Physics Simulation Forum

Then I suggest setting it to sleep before the first frame after you instantiate it.

Its unfortunate that it seems unresolved in that thread. Though Issue 27, which Erwin started and marked as fixed might have a small patch. http://code.google.com/p/bullet/issues/detail?id=27 I think it would be best if you recreated your scene using a bullet demo template so that those working on t...

Well that's obnoxious.

Can you implement and use a bullet debug drawer in wireframe?
It might be a precision error(which may be solved by changing bullet to double precision).

Acceleration is a measure of distance per second per second, so you should take 3 sample points and determine the acceleration from that. Look up basic physics videos on how to do that, it's like the inverse of finding the position.

Do you think you could do a recording to show it?

Well, one body would be visible, with some mass. The other would not be visible, with some mass. The current problem is that the inertia is being calculated wrong as it assumes that the mass is evenly distributed, which it is not. That's why you are currently doing a wheelie. So, if you actually cal...

Does it jitter across triangles that are sufficiently large, or between triangles?

You could have a base frame which the wheels are attached to and then have another (non-rendered) body which is attached to it using several nonflexible constraints that has a lot of mass, so you're simulating how it would work with an engine inside.

The problem is if you want to be dynamic and not have the car bounce a bit when the player first steps on it. Such constraints have to simulate to find their equilibrium, if you hardcode such, then every single car you have better be 100% the same in every way. Its not too hard to do this pre-simula...

Ah, here's an idea, if you're loading physics and graphics stuff, you could load all of the physics data first, then in another thread, let the world simulate (and let things fall asleep) while you load graphics resources(but don't instantiate new entities while you simulate, that's stupid). That wa...

have you considered splitting your terrain mesh into paged sections?

That doesn't seem right, Have you checked to ensure that the collision flags are the same between debug and release?

HPL 1 / 2 use Newton Dynamics, not ODE.

You may be able to glean how they did it from their sources.
https://github.com/FrictionalGames/HPL1Engine

@gokoon: Why did you post that here?

This was an old topic that need not be revived...

Great find Ganaboy