CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

If you want to remove all geometric checks, you need to go down to the
combinatorial part and use the Halfedge_data_structure
http://www.cgal.org/Manual/latest/doc_html/cgal_manual/HalfedgeDS/Chapter_main.html

but you will use all utility functions.

Maybe using the combinatorial map is better in that case:
http://www.cgal.org/Manual/latest/doc_html/cgal_manual/Combinatorial_map/Chapter_main.html

For your flip problem I would naively say that a non border edge is
flippable if the other possible edge intersects it (do_intersect which
works correctly with an inexact construction kernel).

Sebastien.


On 03/02/2012 04:26 PM, Zohar wrote:
> Hi,
>
> I need some tips concerning the polyhedron.
>
> 1. I convert my mesh to the polyhedron structure using the incremental
> builder. Specifically I'm deriving from Modifier_base, add vertices and then
> facets. There are meshes where the incremental process fails, and it
> complains about stuff. Is there a way to construct a polyhedron and bypass
> these tests? I mean tell it to construct the DCEL based on the supplied
> connectivity (facets), with no relation to the geometry.
>
> 2. My predicate that tests if an edge is flippable, by checking for
> intersecting triangles, small dihedral angles, and slivers
>
> http://cgal-discuss.949826.n4.nabble.com/Intersecting-with-Cartesian-kerenl-td4425392.html
>
> is still buggy. For example it fails when I use iostream to convert the
> resulting polyhedron to another type of polyhedron (based on another
> kernel). The failure is the one mentioned in my previous question. Is there
> a more robust way to test is an edge is flippable? Maybe there's a way to
> exploit the inner mechanism of the mesh simplification that tests if an edge
> is contractible?
>
> 3. I wrote a basic scheme that collapses edges in a poly that is purely
> triangular (I'm aware of the simplification package):
> If the vertex to be merged has a valence of 3, then I simply
> erase_center_vertex.
> Else I delete the two neighboring edges using join_facet, and then I
> contract the edge using join_vertex (that must have at least quad facets).
>
> It works fine, until at some point it complains and crashes. I didn't debug
> it, but my only explanation is that it contracted edges that led to
> intersection. And again I'm asking, how do I tell the DCEL to perform
> connectivity operations regardless of the geometry?
>
> Thanks
>
>
>
>
>
>
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Editing-a-Polyhedron-3-object-tp4438763p4438763.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.