I have been playing around with bullet soft bodies for a while. I am mostly trying to understand the the soft body demos shipped with bullet esp. the Collide2 demo with 3 bunnies (demo 9). Now in my own code I keep all of the initialization and soft body/collision objects the same.
In the soft body initialization, i scale the soft body by 6 units similar to the demo and then at rendering time, I down scale the points by 1/6 and the output is exactly as in the bullet demo. If however, i dont scale the soft body, the output is incorrect and the bunny simply collapses into a sphagetti as it hits the ground plane. In addition, the collision margin is also a bit high. My question is why does scaling matter so much in soft body demos? I'd be glad if somebody gives me some pointers on this in particular on the soft body internals in bullet. I have had a hard time finding this information.