(trimesh-trimesh) Re: [ODE] Difference Between Bodies and Geoms

[Adam D. Moss]

Bill wrote:
> which is exactly what I'm going to be doing, heh. So why can't it handle
> 
> trimesh-trimesh when both trimeshes are the same? I could try and fix 
> it, I just need to know what to fix.

As discussed earlier, the new dGeomTriMeshDataSet() function
(see section 10.7.6 in a recent ODE manual in CVS) associates the
geom's homogenous transform matrix with the TriMeshData rather than
the geom, which is the problem.  So you can't share the TriMeshData
without also sharing trimesh-vs-trimesh's idea of the previous
transform, which will make it unhappy.

Anyone diving into that code might go for extra credit by
adding a flag on trimesh geoms to indicate automatic generation
and caching of this matrix just-in-time from its associated
body's position/rotation/velocities.  I don't know how easy
that would be (depends how enforced he separation is between
ODE's collision and simulation) but it'd remove the need to
do dGeomTriMeshDataSet() for the users who aren't often warping
their geoms around by hand.

--Adam
-- 
Adam D. Moss   . ,,^^   adam at gimp.org   http://www.foxbox.org/   co:3