CGAL users discussion list

["Laurent Rineau (GeometryFactory)" <>]

On mardi 14 juin 2011 21:30:44 Michel Audette wrote:
> Hi again,
> 
> I managed to get a result by supplying an angle parameter to
> detect_features...
> 
>   domain.detect_features(0.1);
> 
> Does such a low parameter in fact detect the whole surface input as domain?

When you use the CGAL::Polyhedral_mesh_domain (or 
CGAL::Polyhedral_mesh_domain_with_features), then the surface of the domain 
is 
the set of triangles of the polyhedron (the polyhedron must be triangulated), 
and the interior of the polyhedron is the interior of the domain.

As for 1D-features of the domain, you specify what you want with 
Domain::add_features (but polylines of the set of polylines must intersect 
only at their endpoints).

With domain.detect_features(0.1), all edges of the polyhedron will be used as 
feature curves, but those where the dihedral angle at the edge is smaller 
than 
0.1°.

> On the other hand, I'm seeing a lot of slivers. Perhaps it's my choice
> of parameters, but can you indicate to me the relationship between
> edge_size, facet_size and cell_size? I thought it would be something
> like r, root(3)/4 r^2, root(2)/12 r^3, but the example code that I'm
> seeing
> 
> [examples/Mesh_3/mesh_polyhedral_domain_with_features.cpp , with
>  Mesh_criteria criteria(edge_size = 0.025,
>                          facet_angle = 25, facet_size = 0.05,
> facet_distance = 0.005,
>                          cell_radius_edge_ratio = 3, cell_size = 0.05); ]
> 
> does not corroborate this.

I suggest you send me your code and your input polyhedron, so that I can see 
what happens. The 3D mesh generator of CGAL does not prevent slivers. That is 
the job of the mesh optimizers (named perturb, exude, lloyd and odt).

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