Re: CCD or not?

2006-10-24T16:48:32+00:00

Hi Pierre,

Recently Bullet got some huge refactoring, where a lot of abstractions got removed. This included the CCD stepping. The CCD time-of-impact calculation query is still available, but there is only a btDiscreteDynamicsWorld which doesn't use it (it is just penetration recovery based). I plan on adding the btContinuousDynamicsWorld this week, which uses both time of impact and penetration recovery (hybrid).

It will be both for dynamics versus dynamics, and dynamic versus static. I want to implement 2 different solutions, one based on timewarp (Brian Mirtich) which subdivides the timestep. But first I want to add a solution that adds additional contact points, based on the time of impact, and these 'temporal contact points' will be fed into the penetration depth.

Probably a picture/draft paper would explain this idea better.
Thanks,
Erwin

Hi,

I downloaded the Bullet-2.1-demo-win32 zip file and ran the demo. If I set the shooting speed to 600, enable CCD, and shoot a box... well, it goes directly through the brick wall as if CCD wasn't enabled.

Is that normal? (I just started actually playing with Bullet so maybe I'm missing something)

Or is the CCD only used for dynamic-vs-static, maybe?

Thanks,
- Pierre

Statistics: Posted by Erwin Coumans — Tue Oct 24, 2006 4:48 pm

CCD or not?

2006-10-24T11:47:19+00:00

Hi,

I downloaded the Bullet-2.1-demo-win32 zip file and ran the demo. If I set the shooting speed to 600, enable CCD, and shoot a box... well, it goes directly through the brick wall as if CCD wasn't enabled.

Is that normal? (I just started actually playing with Bullet so maybe I'm missing something)

Or is the CCD only used for dynamic-vs-static, maybe?

Thanks,
- Pierre

Statistics: Posted by Pierre — Tue Oct 24, 2006 11:47 am

Real-Time Physics Simulation Forum

Re: CCD or not?

CCD or not?