How are the depth sensor patterns projected onto the scene?