CGAL users discussion list

[Mariette Yvinec <>]

When you use a c3t3 as the data structure for the mesh generator of CGAL
the triangulation embedded in it has to be a Delaunay triangulation
more precisely weighted Delaunay.
However, c3t3 is only a data structure on top of a triangulation
that allows to mark cells of the triangulation that are  part
of a  given subcomplex the triangulation which is the mesh.

I quote  here from the documentation of c3t3 :

The concept MeshComplex_3InTriangulation_3 provides a data structure to 
represent and maintain a 3D complex embedded in a 3D triangulation. The 
data structure includes a 3D triangulation and marks for each cell of 
the triangulation whether or not such cell belongs to the 3D complex. In 
Cgal 3D mesh generator, this data structure is used to maintain, during 
the refinement process, the current approximation of the domain to be 
discretized and to represent the final mesh. Each subdomain is 
approximated as the union of some cells of the embedding triangulation. 
The 3D complex is always a pure 3D complex, i.e., it includes a subset 
of the cells of the triangulations as well as their subfaces.

Thus the c3t3 offers member functions to handle the subsets of marked faces.
Therefore  the class has nothing to do with the fact that the 
triangulation is
Delaunay and I don't see why it could  not be used with any 
triangulation of CGAL.


Noura wrote:
> Hello, 
>
> I generate a multi-domain 3D tetrahedral mesh from a 3D labeled image and I
> would like to use CGAL mesh data structure to maintain and process it.
> Ideally I would like to have a MeshComplex_3InTriangulation_3 like class.
>
> From what I have understood from the forum and user manual, not every
> triangulation of a domain can be loaded in a c3t3 object, it should be a
> delaunay one and mine is not.
>
> Is the solution reimplementing a MeshComplex_3InTriangulation_3 that has a
> Triangulation_3 type of triangulation and a MeshDomain_3 to go along with
> it? 
>
> Or using Triangulation_cell_base_with_info_3 object with the info being the
> subdomain index? In that case I would not have the iterators to go over
> every subdomain independently.
>  
> Is there a straight forward way that I missed? 
>
> Thank you for your attention, 
>
> Noura FARAJ 
> PhD student 
> Telecom ParisTech, CG Group
>
>   

--
Mariette Yvinec
Geometrica project team
INRIA  Sophia-Antipolis