[ODE] non static deltaTime in the step (worldStep(world,deltaTime) instead of worldStep(world,0.05)) => problem?

[Erwin de Vries]

> I had a similar problem for a while. I traced the problem down to
> the fact that I was running ODE at 500Hz and my collision detection
> at the frame rate (roughly 100Hz).  When I ran the collision
> detection at the same rate as the physics, the problem went away.
>
> Since the subject has come up - is there a way to run the collision
> detection at a lower frequency than the physics? Should my split
> frequency example have worked?

You dont want to. In all my profiling tests the time spent in collision
detection is significantly lower than the time spent doing dynamics. I think
it would be entirely possible to generate contactdata once, and dont destroy
the jointgroup until you're done doing dynamics, dut it creates incorrect
behavior. AFAIK there are no advantages when doing this.

Erwin