Can someone tell me if in wrong? Do i create boxes or pheres same size and position as model in med, then create these models in my actions, update there positions same as main model and use these for collision detection? Or is there a more reliable way?