[ODE] Re: Roller-coaster

Geoff Carlton gcarlton at iinet.net.au
Mon Oct 17 09:05:18 MST 2005

It may be easier to stomp the normals you get from ODE with your own 
ones (either flat face normals, or an interpolated normal), before you 
step the world.  I have found this solution is far better than allowing 
ODE to try to deduce its normal direction based on the proximity to the 
triangle edges.


Aras Pranckevicius wrote:

>>That is a good idea, but realize its not just the normals that are used for
>>friction. When sliding an object over a triangle-mesh you might encounter
>>collisions with the 'internal edges', even when you discard friction. One
>>way to solve this is to handle this is to preprocess the mesh and 'remove'
>>the internal triangle-edges from collision detection. This requires some
>>non-trivial work.
>AFAIK, there was a patch to unstable branch (done by Cyan guys?) that
>preprocesses trimeshes and marks "unnecessary" edges (ala GPG4
>article). I think it wasn't implemented in all colliders though.
>Aras 'NeARAZ' Pranckevicius
>http://nesnausk.org/nearaz | http://nearaz.blogspot.com
>ODE mailing list
>ODE at q12.org

More information about the ODE mailing list