If you associate a normal with each point you can add a rotational spring constant by looking at the dot products of neighbouring normals. It's not going to do cape too well since they're papery (which is different physics again) but it would give more accurate stiffness.[1]
The hard part is not so much doing the physics of the cloth, but dealing with the constraints. Making sure the cloth doesn't pass through itself is challenging enough, making sure it doesn't pass through anything else is even harder.
I think I could deal with the physics for the cape part but I simply don't have any time for the programming. If anyone is desperate to do some coding I'd be happy enough to help.
Tim
[1] Actually I've got an idea how you could do paper too by considering lines as potential creases but it all gets a bit nasty.
The hard part is not so much doing the physics of the cloth, but dealing with the constraints. Making sure the cloth doesn't pass through itself is challenging enough, making sure it doesn't pass through anything else is even harder.
I think I could deal with the physics for the cape part but I simply don't have any time for the programming. If anyone is desperate to do some coding I'd be happy enough to help.
Tim
[1] Actually I've got an idea how you could do paper too by considering lines as potential creases but it all gets a bit nasty.