It's a cool idea, but only being able to see one face at a time makes it very hard to figure out the effects of the rotations. Maybe you could allow camera movement or animate the in-between states of the rotations to make the mechanic more understandable