CGAL users discussion list

[Jane Tournois <>]

Hi Alexandre,

facet_distance is, for a "restricted Delaunay" facet, the desired upper 
bound on
the absolute distance between the circumcenter of the facet, and surface.
This distance is computed as the distance between the circumcenter and 
the intersection of the facet dual (ray or segment) with the surface.

So as facet_distance gets smaller, facets indeed become more 
horizontal/vertical, but I'm afraid you will still get approximations in 
the corners of the voxels.

You can also provide the mesher with polylines to be "protected", i.e. 
present in the output mesh. If you provide as "sharp features" the edges 
of the cube [0,0] x [L,L] you can make sure they will be in the output 
mesh (maybe refined, but still rectilinear).

Hope it helps!
Best,
Jane.

Le 06/10/2015 17:13, Alexandre Lapene a écrit :
> Hi Jane,
>
> Thanks for your quick reply.
>
> There are two reasons why I'm trying to use small values for facet_distance.
> The first one is I'm not sure of the definition of facet_distance. I had
> assumed it was a relative value, likely based on the domain size, diagonal.
> Also while reducing this value below the theoretical minimum value you
> mention, I noticed improvement in the quality of the surface mesh. It is an
> absolute distance?
> Second reason is I also want to have very flat surfaces for my bounding box
> surfaces (squares) as that would make my life easier for simulation. To tell
> you the entire story, I ultimately need to give separate attributes for the
> boundaries and one way is by detecting, after meshing, x=0,x=L,y=0,y=L
> etc... faces. So if you have an alternative I'm more than interested.
> Thanks!
>
> Alexandre
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Isosurfaces-from-Gray-level-image-tp4661158p4661200.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.

--
Jane Tournois, PhD
R&D Engineer at GeometryFactory
http://www.geometryfactory.com/