CGAL users discussion list

[Stephane Tayeb <>]

Ianic wrote:
> Hi again!
>
> Now that I have my mesh connectivity output, I tried to mesh a CT voxelized
> image of a sandstone porous sample. It has only one subdomain - the pores.
> So, theoretically, I expect the resulting surface to be manifold and all
> facets to have 3 neighbors. When I looked at the output, I saw that some
> facets had more than 3 neighbors, and looking at the resulting mesh itself,
> I saw that there are some regions where 4 facets meet at one edge. I wonder
> whether there is any way to guarantee the manifoldness of the boundary
> surface? Or should I then try to construct the surface mesh and feed it to
> the 3d mesher afterward?
>
> Yan

Hi,

There is currently no way to guarantee the manifoldness of the output 
surface in CGAL 3D Mesh Generation package. Decreasing the approximation 
error might help to get the result you need.

As you say, you can try to use CGAL Surface Mesh generation algorithm 
(which has such a feature) to get a manifold surface. Then, you can 
insert the vertices of this surface mesh in a c3t3 and refine it with 
the 3D mesher (note that oracles with the same behaviour should be used).

Stéphane.


--
Stephane Tayeb
Software engineer - INRIA Sophia Antipolis
Geometrica Project-Team