[Bf-codereview] MDef BVH Support (issue 4529048)

[lukas.toenne at googlemail.com]

Looks good to me, only the use of epsilon in bvhtree made me wonder, but
i may be wrong about that one.

Other than that i'd say: ready for trunk!


http://codereview.appspot.com/4529048/diff/1/source/blender/editors/armature/meshlaplacian.c
File source/blender/editors/armature/meshlaplacian.c (right):

http://codereview.appspot.com/4529048/diff/1/source/blender/editors/armature/meshlaplacian.c#newcode1740
source/blender/editors/armature/meshlaplacian.c:1740: mdb->bvhtree =
bvhtree_from_mesh_faces(&mdb->bvhdata, mdb->cagedm, FLT_EPSILON*100, 4,
6);
Is the epsilon necessary here? Afaik the epsilon value simply "inflates"
the bounding volumes, which should not have any effect on the
intersection tests, except slightly lower performance? To my
understanding, for nearest point search it defines the search radius
(see Shrinkwrap), but other callers all have epsilon=0.

http://codereview.appspot.com/4529048/