[ODE] ODE Unstable Commit: Trimesh Preprocessing

[Geoff Carlton]

This looks good - I've immedately got from unstable and merged it fine 
with my code.  Thanks, a nice improvement like this doesn't come along 
every day!

However, I've noticed that it seems to do two unnecessarily 
normalisations per edge check.  To work out the concavity, it performs a 
dot product on two normalised vectors.  As this is only done to check 
the sign, I believe the result would be the same even if the two vectors 
weren't normalised.

Geoff


Colin Bonstead wrote:

>Ok, it took me about a half hour, but I finally did a commit to the ODE 
>unstable branch.  For anyone else trying to do this, make sure you're 
>logged in via SSH, otherwise it won't let you commit.
>
>Anyway, what I committed is a patch I did last year to add trimesh 
>preprocessing.  I based it on the chapter "Fast Contact Reduction for 
>Dynamics Simulation" in Game Programming Gems 4.  Basically it's just 
>marking all the interior and concave edges (and their vertices) as 
>ignorable.  It requires an extra byte of memory for each triangle if you 
>use it.  (If you don't call the preprocess function on your trimesh it 
>should function the same way as before, and not allocate any extra 
>memory.)
>
>To use it, just call the function dGeomTriMeshDataPreprocess on your 
>trimesh.  Once you've done that the data is cached and ready for use by 
>the colliders.  Currently, only the ccylinder-trimesh collider uses the 
>data, since that's the only one I use.  However, it should be fairly 
>easy to add it to any of the other colliders, if someone wants to do 
>that.
>
>If you want to cache the preprocessed data so you don't have to compute 
>it every time you start your simulation, you can use the functions 
>dGeomTriMeshDataGetBuffer and dGeomTriMeshDataSetBuffer.
>
>Colin Bonstead
>Cyan Worlds, Inc
>
>
>
>_______________________________________________
>ODE mailing list
>ODE at q12.org
>http://q12.org/mailman/listinfo/ode
>
>
>
>  
>