I have found some bugs in the built in collision detection functions. The collision detection works fine when using a primitive such as a cube or sphere against a 3d object. However collision detection between two 3d objects returns some unexpected and wild results.
I have found one major part of this issue to be scaling. A scaled object retains it's original data for collision detection. So once scaled it is treated as if it where it's default size.
Thanks for looking into this.
" The best slaves are the ones who think they are free. "