[ODE] Another Tri-Collider Question

Vadim Macagon vadim_mcagon at hotmail.com
Tue May 13 20:02:02 2003


> Well since I really didn't want to switch to double precision I tried to
> come up with
> an alternative, and I just did a few minutes ago :). The penetration depth
> is zero when two things are just "touching", that's the goal of collision
> resolution, that zero depth was never achieved in sphere/mesh collision,
it
> would go very close to it like 1.48e-05 and then jump back to 0.002 or
> something, I guess that's a precision issue somewhere down the line.
Anyhoo
> what I did was - if penetration depth goes under 1.5e-05 I say it's close
> enough to "touching" and set penetration depth to zero, and voila it's now
> rock stable. I don't yet know if it solves the "fear the wireframe"
problem,
> but hey I'm still on a high :)

Hehe, well that was a load of carp :). It only worked because I didn't set
surface
parameters properly, or when mu is zero, however setting soft CFM or global
CFM to a small value (in my case 0.0005) seems to eliminate the jittering.

Vadim.