Real-Time Physics Simulation Forum

Those two code snippets are clearly not doing the same thing. If you initialized the transform in the first snippet like this, then they would be: btTransform world_transform(m_player_collider_objects[ i ]->getWorldTransform()); The first snippet is clearing out the rotation. The second is preservin...

IMHO changing the way the physical world is stepped is completely unnecessary for networking the simulation. When you asked about this, I thought you were going to be trying to do something way crazier than just networking. I think you will only make more work for yourself, but that's certainly your...

Ah OK, I assumed the problem for you was getting it built for the platform. How it works is identical from platform to platform ... the only difference being under-the-hood math optimization based on available instructions for your CPU. I have the same physics code in Win/Lin/Mac/Android. That's why...

Why would that exist? Do you need convincing that it actually runs on android? Because it does. EDIT: I guess this wasn't a very helpful post so I will add that I have seen people just dump all the files into their android project and build it this way. I did this at first, but in the end I used CMa...

You are doing 500 iterations in the solver and you are surprised at the FPS drop

Just flip the one axis that's getting inverted

I doubt anything is different about the kinematic character controller. Bullet3 is about the GPU based physics pipeline.

Interesting, I didn't know about this. Look in the Benchmark and ConcaveRaycast demos for #ifdef BATCH_RAYCASTER I guess you'll have to define that before you can see how it works. Note that this isn't going to make raycasting threadsafe. It looks like it uses a thread barrier to perform raycasts in...

Sanity check, you are passing params in the order (w, x, y, z) and not (x, y, z, w) ?

Even reading non-atomic values can cause tearing, and anything worth reading is not going to be atomic (eg. floats and vectors and etc). I don't see how you would assume raycasting into the physical world from another thread would be safe. But anyway, it isn't.

Yeah the results are definitely not sorted by distance

Sure. First of all it's really helpful to be able to debug draw direction vectors, so if you spend a bit of time on that you will be glad you did. I tore this out of a section of steering code and quickly rewrote it to make sense and be readable for your use case (again, I'm using OgreMath but it's ...

OK well assuming you are familiar with Ogre, or willing to look into the two methods I mentioned in OgreMath then here is some simple code. Note it is rotating the Ogre head ... which faces unit Z. Also Ogre is Y up. Lastly I assume the turret only aims left and right, not up and down: // current or...

OgreMath has both a Vector3::getRotationTo (retrieve the shortest arc quat between two vectors) and a Quaternion::Slerp (spherical linear interpolation between two quats)

The major problem with this is that a collision is not guaranteed to be detected. If two objects collide but are then completely restored all in a single stepSimulation call above, then the if-block might never be called. Isn't that the exact case for which CCD (continuous collision detection) is d...