CGAL users discussion list

[Simon Giraudot <>]

Le 03/11/2017 à 16:59, Thomas Holzmann a écrit :

Hi,

the function poisson_surface_reconstruction_delaunay very frequently crashes on my datasets.

I use standard parameters (I also tried varying the parameters) and I compute the average spacing with CGAL::compute_average_spacing. I also remove outliers with CGAL::remove_outliers.

On two of my datasets I regularly (at least at 2 out of 3 runs) get this error:

CGAL ERROR: assertion violation! File: CGAL-4.11/include/CGAL/Surface_mesher/Surface_mesher.h Line: 253 Explanation: (56.0006 41.2038 -0.0783118) is already inserted on surface.
After asking google, I found out that setting the Non_manifold_tag() might help. But this didn't help in my case. Then I usually get other errors:
CGAL::Polyhedron_incremental_builder_3<HDS>::
add_vertex_to_facet(): input error: disconnected facet complexes at vertex 225:
           involved facets are: 137 134 135 (closed cycle) and.

or this one:

CGAL::Polyhedron_incremental_builder_3<HDS>::
lookup_halfedge(): input error: facet 1803 shares a halfedge from vertex 3005 to vertex 3004 with facet 1788.

Can someone give me a hint what the problem could be? My input point cloud has several big holes. Might that be a problem?
Thanks!
Hello,

It seems that the surface mesher does not manage to mesh the implicit function at the required precision. Have you tried to increase the spacing parameter or the sm_distance parameter?

The holes should not be a problem in theory as long as the inferred surface is closed. Poisson cannot work properly if the inferred surface is not closed (if it is not surrounding a volume with a well-defined interior and exterior). Badly defined or badly oriented normal vectors can also lead to problems.

-- 
Simon Giraudot, PhD
R&D Engineer
GeometryFactory - http://geometryfactory.com/