CGAL users discussion list

["Laurent Rineau (GeometryFactory)" <>]

On dimanche 04 septembre 2011 10:21:10 Zohar wrote:
> Hi,
> 
> First congratulation on the forum interface, it's more convenient.
> 
> Second I have a problem with make_surface_mesh when reconstructing from an
> implicit function. When I want higher resolution, i.e. I decrease the
> radius or distance bounds in the criteria, it fails with:
> 
> CGAL::Polyhedron_incremental_builder_3<HDS>::
> lookup_halfedge(): input error: facet 2960 shares a halfedge from vertex
> 3581 to vertex 351 with facet 1784.

I guess your iso-surface is not manifold. What is the implicit function?

> So I thought instead of using Exact_predicates_inexact_constructions_kernel
> to use Exact_predicates_exact_constructions_kernel, but I had compilation
> problems. I fixed them with changing in:
[...]
> , but when I ran it with the changes, it just ran forever.

make_surface_mesh is really not designed to run with the "exact-exact 
kernel". 
The meshing algorithm is iterative, and the coordinates a point at a step is 
computed from points inserted before in the triangulation. With the exact-
exact kernel, that means that the construction graph of new points will be 
bigger and bigger, and that both the memory consumtion and the running time 
are O(n^m) with a big m.

-- 
Laurent Rineau, PhD
R&D Engineer at GeometryFactory           http://www.geometryfactory.com/
Release Manager of the CGAL Project       http://www.cgal.org/