CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

Thanks for the report, it was actually a bug.
It is fixed in https://github.com/CGAL/cgal/pull/2558

Thanks again,

Sebastien.

On 10/23/2017 03:40 PM, Adrien Lefieux wrote:
> Dear CGAL staff,
>
> Lately, I have been looking a bit at the surface mesh skeleton system.
>
> I tried to get the correspondence between skeleton vertices to surface 
> vertices correspondence, as displayed on the figure 66.2.c, which looks 
> like what I expect: surface vertices are close to skeleton vertex as 
> they are linked during the "shrinking" of the surface.
>
> However, trying on the example simple_mcfskel_example.cpp, I obtain the 
> following result (I could not reproduce the octopus, not having the 
> mesh, so I took the teddy bear). This is not what I would be expecting 
> as the correspondence appears to a bit everywhere on the upper part of 
> the teddy.
>
> Is this normal, to be expect or something is going wrong?
>
> Inline image 1
> Using CGAL 4.11
>
> This is just the CGAL::extract_mean_curvature_flow_skeleton 
> <https://doc.cgal.org/latest/Surface_mesh_skeletonization/group__PkgMeanCurvatureSkeleton3.html#gaf4b49adaa500340ce11630f91b320aec>(tmesh, 
> skeleton) and then the correspondence is build for one skeleton vertice 
> v and then the correspondence is accessed via: skeleton[v].vertices 
> (just as in the example). This examples uses Polyhedron_3 with the 
> simple cartesian kernel.
>
> CGAL version 4.11 and 4.10 gave me similar results. However, with CGAL 
> 4.9.1 the results look much better to me. Another way to see the issue 
> is to load the correspondence-poly.cgal and, for version 4.10 and 4.11, 
> the correspondence polylines are not all within the boundary of the 
> domain. on the contrary, for 4.9.1 it seems to be (I can provide other 
> images if you wish).
>
> Inline image 2
> Using CGAL 4.9.1
>
> I use Eigen 3.3 on an OS X machine with clang (Apple version 9) with 
> release flags and boost 1.65.1 .
>
> Thank you very much for the hard work,
> Adrien Lefieux