[ODE] Fixed stepping

Jon Watte (ODE) hplus-ode at mindcontrol.org
Fri Aug 18 10:58:17 MST 2006



Megan Fox wrote:
> I would suggest not replacing the stepper function, as there remains a
> great deal of work that must be done per-step rather than
> per-batch-of-steps.


If that's the case, you simple call dWorldStep() with a value of 1.

To some people it seems crazy to run collision detection at a rate 
different from simulation, but if you know that you can get away with 
collision detecting 50 times a second, and then step 200 times a second 
to get a harder "feel" in the integrator, there's no reason not to do 
that. Your collision joints will still prevent penetration as appropriate.

In fact, I bet you can get a pretty decent simulation by pre-detecting 
simulations using swept primitives, and pre-creating contact joints 
before penetration. I must try that some time.

All this talk of changing the step function, alternate step functions, 
etc, sounds no good to me. There is no reasonable use for variable step 
sizes in ODE today. I'd prefer we set the step size on the world, and 
tell the stepper functions how many steps to run, and leave it at that. 
If you need variable step size, just change the world parameter before 
each step.

The bonus would be that the impulse calculation functions would know 
what to do without a manual step size argument.

Cheers,

			/ h+



More information about the ODE mailing list