You may be making things harder than they need to be.

Unless you are going to intentionally turn the figure during the animation so the viewer can see how thin it is, there is really no need create a super thin model. It's easier to adjust lighting and textures to visually imitate a flat object, especially if you'll be doing any type of head or body turns if it's a life form.

If you really want to have flat objects, I'd suggest modeling the 3D object and then reduce the depth in one axis prior to rigging. (x=100%, y=5%, z=100%) This technique is something I did on the attached image for a Carrara project a loooong time ago.

If you can show a style sample, I can help you brainstorm.