separate collision mesh generation from visual, still works assigning the same mesh to both - lets us have consistent collision between censoredsigns being enabled/disabled