Closed alecjacobson closed 8 years ago
The limitation of lscm
to 2D meshes was entirely artificial. Just now I've pushed a new version that is based on another paper (doing the same thing) which is easier to implement for 3D surfaces.
Regarding choosing the fixed points, this is a drawback of these type of parameterization papers that are so sensitive to the choice of fixed points and their arrangement in the plane. More advanced/recent methods will also try to optimize the boundary mapping or rather allow the boundary to move freely in the plane. I particularly like the recent work of Smith and Schaefer called "Bijective Parameterization with Free Boundaries". Though I don't know if there's an implementation around.
Yeah, thanks for your answer!! the problem of calculating boundary conditions is drawing me crazy.
I've seen another way of calculating boundary, but that is an incomplete way. http://nbviewer.jupyter.org/github/gpeyre/numerical-tours/blob/master/matlab/meshdeform_1_parameterization.ipynb
https://github.com/gpeyre/numerical-tours/blob/master/matlab/toolbox_graph/compute_boundary.m
From personal email...