[ODE] Objects fall through trimesh.

[Jon Watte]

Do you have actual contacts at the point where you're sinking through?
If so, what is the normal, depth, and force of those contacts?
If not, can you step through to figure out why you don't have the contacts?
Another classic issue with trimeshes is when the winding is different than you expect. Are you drawing with backface culling turned on? If not, turn it on, and make sure everything still looks right.
Last, sometimes you're bitten by things like whether you're supposed to count indices or triangles -- just do a double-check to make sure.

Btw: Tesselating triangles out of a BSP is not so hard if you can deal with re-generated triangle lists. Build a structure of all edges, with pointers to the triangles sharing each edge. Then, sort by length of edge. Start with the longest edge, split it by putting a vertex in the middle, remove the two previous triangles and add the four new you just created. Insert the new triangles/edges into the data structure at the appropriate place. Repeat until the longest edge is shorter than your tolerance.

Cheers,

			/ h+


-----Original Message-----
From: ode-bounces at q12.org [mailto:ode-bounces at q12.org]On Behalf Of Chuck
M.
Sent: Sunday, August 22, 2004 8:25 AM
To: ode at q12.org
Subject: [ODE] Objects fall through trimesh.


Unfortunately in some but not all cases, boxes, spheres, and capped
cylinders all seem to "fall through" triangle meshes in my simulation.
I tried decreasing the step size tenfold but to no avail. Also, I
recompiled ODE without the MERGECONTACTS option in the trimesh sphere
source file.  In addition, I tested various orders of indexing the
vertices in each triangle. Finally I noticed that tesselating my
triangle mesh for my landscape solves this problem, however I also use
BSPs in my simulation which cannot be tesselated very easily.

I have read three months of the mailing list archives but no valid
solutions could be found. Please help me to resolve this problem as
proper collision detection is quite necessary for my project.

-- 
thechuckster.beigetower.org
_______________________________________________
ODE mailing list
ODE at q12.org
http://q12.org/mailman/listinfo/ode