Real-Time Physics Simulation Forum

Thanks. I was using the wrong master, which doesn't have the btSequentialImpulseConstraintSolverMt source files. I'll build it now and let you know what I think after a couple of days of testing.

I've been avoiding the variable update due to the assumption that it can be buggy, so I decided to go with the deterministic approach. I'll see how it works later, though.

It requires some understanding of kinematic objects and how they interact (or don't interact) with the physics world. It is pretty straight forward, but sadly I have found Bullet's character controller classes overly complex and messy, which makes following them rather tedious. I wrote my own CC fro...

Can you call stepSimulation(customFixedValue) only when "total >= doStepValue" instead? I could, but it would mean duplicating what Bullet is doing in the stepSimulation method, and trying to have both synchronised perfectly could be problematic. In fact, I did try but a lot of the update...

I've tried getting this working, but VS is throwing an assert: btAssert( gBtTaskScheduler != NULL ); // call btSetTaskScheduler() with a valid task scheduler first! - line 444 of btThreads.cpp I chose PPL in cmake and I'm creating the world like so: broadPhase = new btDbvtBroadphase(); collisionConf...

Hi, I have the physics engine set to a fixed 60hz and my engine is free running. If free running at 120hz then the physics step will skip one update and add the deltatime to some sort of accumulator. Is there anyway, without altering Bullet's source code, to determine if Bullet will skip a particula...

Are you doing it like this? Update() <--update to desired position Physics() Apply() <-- apply to actual position/build matrix etc Draw() You could take advantage of the userdata() pointer Bullet offers and set up an automatic callback, or just create pointers to the rigid's world position/quat/etc....

Thanks, but the callstack wasn't giving me much information that I could interpret as something useful, other than to help me figure out when the crash happened. However, some of the sourcecode I am using I wrote a while ago for 32 bit builds, and after changing the following this assert has stopped...

Somewhere during the step simulation call I am getting a triggered breakpoint in the btAlignedObjectArray.h file. Here: SIMD_FORCE_INLINE T& operator[](int n) { btAssert(n>=0); btAssert(n<size()); return m_data[n]; } n = -1; This only happens in DEBUG build on vs2015 compiling x64. It works OK i...

Bullet is pretty much straight forward for setting up. However, you will probably need to use CMake to create the build for VS2015. Just run CMake, set the directories to the root directory of Bullet, configure and create. After that you just open the project CMake created and build your debug and r...

Hi, I am only using a sweep test for the collision detection and I'm handling the response myself. The code is different from the character controller because I'm attaching to ground with a ray (stays parallel with ground - ground normal is 'up') and other things. The problem I am having is with cur...

Nevermind, I got it working using a spherical convex sweep test. Out of curiosity, how expensive is this function compared to the multiple ray trace calls I was using before?

Hi, I'm using raytrace for an obstacle avoidance function in 3D, but I'm having to use around 4-5 feelers to try and cover the boundary of the 3D object they're attached to. Does Bullet have a 3D method so I could project a 3D primitive (cube or cylinder with non-uniform scaling) to simplify the tas...

I found using a cylinder instead of a capsule gave better results, but you will have to setup stepping yourself. I have yet to update Bullet to the newer version so I don't know what state the CC is in, but I got some of the source for the stepping from one of those Bullet patches offered by the com...

I'm away from my home PC at the moment so it will have to wait a few hours. However, from the top of my head I am using indexed triangles. So, the line is created using the two points:

glm::vec3 points[2]

And rendered using 3 indices:
short indices[3] = { 0, 1, 0 }