So how is facial animation done? Do you make duplicates of the whole model for blendshapes? Seems a bit inefficient when you'll only be moving a couple of verts on the face.