[ODE] Should trimeshes react to physics?

Keith Wiley kwiley at cs.unm.edu
Fri Dec 9 16:19:49 MST 2005

I've been having similar problems.  I'm using a trimesh as a terrain  
(not really hi res mind you).  I have rectangular blocks sitting on  
the terrain.  For the most part it works really well, but  
occasionally one of the blocks suddenly explodes upward with great  
force.  I'm pretty certain the cause of this is that the block is  
allowed to fall through the trimesh a little bit and then suddenly  
the collision gets detected and the block is eject forcefully from  
the collision.  I think decreasing the step size might help, but it's  
still confusing sphere behave just fine with trimeshes, and more to  
the point, the error results from the block falling into the trimesh  
for a few consecutive frames, so over the span of multiple calls to  
step the simulation forward the collisions is missed, allowing the  
block to fall partly into the trimesh.  I don't see how a smaller  
step size would help alleviate a problem that spans multiple  
consecutive calls to the world stepper.

On Dec 8, 2005, at 3:31 PM, Marco Grubert wrote:

> Jon Watte wrote:
>> Trimeshes have been around for years, and are stable, as long as you
>> don't try colliding them against other trimeshes.
> Even when limited to box collisions, trimeshes are not all that  
> stable. It
> mostly depends on the spacing of triangles. If there are few  
> triangles the
> results are quite good, but with higher density meshes (e.g.  
> terrain) the
> collider creates too many and often times conflicting contacts that  
> break
> the solver.
> - Marco Grubert
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode

Keith Wiley         kwiley at cs.unm.edu         http://www.unm.edu/~keithw

"I do not feel obliged to believe that the same God who has endowed us
with sense, reason, and intellect has intended us to forgo their use."
                                            --  Galileo Galilei

More information about the ODE mailing list