[ODE] Auto Disable Improvement patch #1586738

John Miles jmiles at pop.net
Wed Nov 1 11:57:09 MST 2006

That's a very good point, too: those damping options should be part of ODE
as well, with non-zero defaults.  Most people want them, whether they know
it or not.  Those who don't, will know how to turn them off.

Either way, damping or no damping, instantaneous velocity is NOT the correct
parameter to use for any type of auto-disabling system.

-- john

> John Miles wrote:
> > The usual example is something twitching and jittering on the
> ground when it
> > should be at rest.  The object's average velocity is near zero, since it
> >
> The problem there is that something is twitching and jittering. That's a
> problem in your simulation that you need to fix. In my experience, the
> biggest source of jitter is not using a fixed time step. The second
> biggest source is not using any angular momentum dampening. The third
> biggest source is not using any linear momentum dampening.
> Using a fixed timestep, adding a small counter-torque to each object,
> and a small counter-velocity to each object, each step, will typically
> make simulations settle down and auto-disable just fine.
> Cheers,
>           / h+

More information about the ODE mailing list