[ODE] Dynamic mesh collision

[Amund Børsand]

Amund Børsand <amund@c2i.net> skreiv:

> whitt013 <whitt013@bama.ua.edu> skreiv:
> 
> > ODE would still provide the "illusion" of deformation if your ERP is less than 
> > 1 (like 0.9) and your CFM is slightly greater than 0 (like 1e-5).  In that 
> > case, it takes more than 1 timestep for two intersecting bodies to be pushed 
> > apart, allowing the tires to sink through the ground just a tad.  From the 
> > top, it looks like the tire just deformed to fit the surface.  And it's a lot 
> > cheaper to use one non-deformable cylinder than a whole mesh with pressure 
> > calculations....

> Yes, I know, this is a fair "shortcut" for on-road and high-speed car
> simlations, but it behaves quite unrealistic when crawling over sharp
> rocks with near-flat tires. Would be fun to try to implement it, though,
> even if it would be slow (we'd just make it optional). I've never seen
> that being done before.

By the way, the mesh with pressure is not such a big deal, it can be
done quite fast (only problem is, it can't rotate and move too much per
timestep, but that's no problem in my case), but it's the collision
detection I suddenly started worrying about. In this case, however, I'd
guess that a vertex-triangle test could be sufficient. Is it
possible/difficult to do that with OPCODE/ODE?