The cube, the epicycles and the human face
9 points by nomemory
9 points by nomemory
We only use a limited number of circles, so the generated point does not always lie exactly on the face. Before drawing the final line, we move it back onto the surface.
Bit of a cop-out, unfortunately
I would have accepted the poorer reconstruction. It would also have been neat to show how adding more terms improves the reconstruction error.
I have nothing to say other than that this was neat.
Finding a good way to sample the surface was probably the most difficult part.
Please, do tell :)