Real-Time Physics Simulation Forum

Bullet is deterministic when using the same machine and and executable, if you reset the simulation properly, disable solver randomization and use a fixed timestep (check the Bullet demos restart function). On different machines, compilers etc, the simulation will obviously be different. Why? If yo...

I don't know how 'Bullet' solve this problem. But I think that for proper kinematic anlaysis, the function which shows some warnings for redundant constraints can be helpful. With a PGS iterative solver, redundant constraints do not cause any issues. If I remember correctly (and someone please corr...

You're welcome. Glad you got it working!

The problem may be due to the fact that there's no agreed standard definition of the order of rotations for Euler angles. The page you got that code from assumes NASA's standards, while I couldn't quickly figure out what order that Dark SDK uses. I can get you as far as a quaternion to matrix conver...

OK, I think I understand the confusion. Yes, PGS solves LCPs. But "Gauss-Seidel" (GS) basically means to solve the equations iteratively. This may not be obvious, but as Erwin mentioned, this is mathematically equivalent to SI (Sequential Impulses). So it sounds like what you are using is ...

jam wrote:Wow, I arrived at a nice stable looking solution. Sorry to say it wasn't PGS

...

Anyway, I basically went through all the impulses in the scene and applied them separately, and did this for all the contacts 10 times

Sorry, I'm a bit confused, how is that not PGS/SI?

Martin- The simple answer to your initial question is: Your "equations of motion" are only correct for a constant force during the entire timestep. Any force that depends on time, position, or velocity will NOT be constant. And that's just about every force that I can think of, with the po...

If by special situation, you mean a light object with constraints to two heavy objects, you are right. The problem is that by working on velocities, the iterations transfer momentum around the system. And a light object can't hold that much momentum, so it takes a lot of iterations. So if you're cra...

My opinion is that large mass ratios are a real problem with PGS/SI methods. I know one solution is to 'not have large mass ratios'. That's not always an option, especially in non-game applications. One possible solution is to detect such situations and solve a smaller subsystem using a more exact (...

The force I apply to both mass points is -k t d (each would get half of course), where k is the spring "value", Sorry I'm not answering the question you want answered, but I am not sure about that anyway. Regarding your "each would get half of course" clause above, I disagree. I...

Yeah, I think you've pretty much discovered the one massive problem with position-based physics. Restitution, and any energy preservation for that matter, cannot actually be controlled in any straightforward way, basically because energy and velocity are related (and velocity is auxiliary at best in...

The equation you gave failed to also divide by the time difference (0.0166667 for 1/60 of a second).

I haven't implemented such a solution, but you might instead store of the existing collisions at the time you go to sleep (this would be per-object storage). So while you are sleeping, you don't do any extra work or reporting of intersections. Then when you wake an object up, you "activate"...

Yeah, I'd probably agree with that approach. With the range of your world, I'm pretty sure you'll need to store double-precision positions (at least for the base object position, not necessarily vertex or other local-space data). However, after broad phase, if you can transform objects to a local sp...

Actually I only scanned the article, but I wonder if we can expect physics to go the same way like graphics and soon we need to implement OpenPhysics or DirectPhysics interfaces instead of running our own engines what would be only interesting for hardware vendors? Graphics has almost come full cir...