Ball Model Scaled = Inflatable Punching Bag

I have a soccer ball model that is extremely large for my scene so I set the scale to be 0.0035f to make it look correct.  When I create a DynamicPhysicsObject for it and add it to the scene it acts very strange like either the center of gravity is above it quite a bit, or it's just really bottom heavy.  I'll post some code this evening if this isn't enough to go on, but I believe I recall some previous discussion about scaling a model causing problems with physics and I'm not sure how to solve it.



darkfrog

I don't think that the scaling is the problem. Most probably your model is simply not correctly centered. The center of gravity is forced to (0,0,0) in object coordinates, where object is the node or trimesh you are passing to DynamicPhysicsObject -> make sure your spheres center is at (0,0,0).

Thanks, I'll give that a try.  My bounding sphere looks correct though…?



darkfrog

Ah, but boundings surround your object regardless of whether they are centered in their local coordinate space or not.



As a sanity check, you might try altering one of the vertices to be at (0,0,0) and see if it get's shifted to somewhere in the middle of the ball, or somewhere off to the side or off centered.

Well, I was able to scale the object down in 3d studio and then export it, but that makes the ball appear (even when it's set to 0, 0, 0 on local translation) to be way off the screen and not look right at all.  What is strange is that at the normal size it looks fine, and if I scale it down and export it as an obj file it also works fine, just not a scaled down 3ds file.  It would seem as it's a bug in the 3ds to Jme converter, but I need to look into a little further.  I've been able to reproduce it twice now but it's not really a big issue to me as I'm able to use obj instead.  The only big problem I'm having now is figuring out how to export a texture image from 3d studio that I can use with my exported object.



darkfrog

I take it you didn't try altering the vertices?  Just making sure it is 0,0,0 centered in the modeller is not enough as it may be incorrectly shifted somewhere else down the line… (like in the conversion process)

Do you mean in the modeling program or in jME?



I modified the sphere to be perfectly 0, 0, 0 which solved the strange center of gravity problem, but the problem stills exists in reference to a scaled down 3ds model.



darkfrog