[ODE] Trimesh problems

Geoff Carlton gcarlton at iinet.net.au
Mon Jan 9 18:58:11 MST 2006


I've done many changes, but the main two were that collision normals are 
set to the triangle normals, and also I merge duplicate contact points 
(and interpolate any normal to an average value) before I step the 
world.  Until I did that, I found ODE generally worked, but spheres and 
other objects had wierd snags and effects, depending on the underlying 
triangles representing a surface.

However, that doesn't sound like what you describe, so you may be better 
off trying to tweak the parameters first.  In an earlier email you 
mentioned these:

dWorldSetContactMaxCorrectingVel and dWorldSetContactSurfaceLayer

They are extremely useful for eliminating bounce.

Cheers,
Geoff

Anders Olofsson wrote:

>That doesn't sound right. Why would I want to push it towards some edge?. 
>Me hacking code in ODE is like monkey do brainsurgery. Funny but none will 
>survive. Well, I'll try.. Did you allways use the triangles normal instead 
>of normal pointing towards some edge?. I don't immediately see how this 
>applies to jumpy spheres on large triangles though. Did you modify 
>something more in the trimesh/sphere code?.
>
>Downloading source.
>
>Thanks!,
>
>--Anders Olofsson
>
>At 00:51 2006-01-10, you wrote:
>  
>
>>The standard Trimesh code has some problems - for example, colliding 
>>objects with a completely flat trimesh will end up with many non-vertical 
>>contact normals.  This occurs because the flat triangles try to push the 
>>sphere "away" towards an edge, sometimes with almost horizontal 
>>normals.  Its unworkable in its current form, and I can't see how anybody 
>>uses it without hacking away into the code in ODE (as I did).
>>
>>Thanks,
>>Geoff
>>    
>>
>
>_______________________________________________
>ODE mailing list
>ODE at q12.org
>http://q12.org/mailman/listinfo/ode
>
>  
>


More information about the ODE mailing list