CGAL users discussion list

[Mariette Yvinec <>]

It is not clear from the pictures that the problem is
limited to edge recovery.
It seems that the cylinder surfaces  thenselves are not accurately 
represented
in the mesh you have generated.
Could you be  more precise about the input you gave to the mesher,
in particular about the MeshDomain argument ?

As for edges, we are currently working on
a version of the mesh generator able to generate meshes
in which input curves (or sharp edges)  can be accurately represented.
This version will be included in the release 3.8 of CGAL
which is to appear in march 2011.
If you can't wait until this release appear, a simple work around
consists in sampling densely enough the curves you want to be accurately 
approximated
in the mesh and insert the sample points  into the  c3t3.triangulation()
prior to launching refine_mesh_3.

toddoe wrote:
> Hi All,
> First thanks for all the help in the past. I now have a well-working
> compiled solution (cgal 3.6.x).
>
> Here is a major issue that I have attempted to 'work-around' but cannot
> succeed.
>
> I am trying to mesh a simple 'embedded' cylinder where the inner cylinder is
> in 'contact' with the outer 'ring'. and the whole thing is capped at the
> ends. (see links to pics later).
>
> unfortunately, there are serious problems. at the intersection of the 'caps'
> the inner/outer cylinder tetrahedron are very jagged. :(
>
> see these images (generated using matlab):
> cylinder 'side on': http://biomeq.com/meshpicstemp/sideon.png
> cylinder 'oblique': http://biomeq.com/meshpicstemp/oblique.png
> without end caps: http://biomeq.com/meshpicstemp/withoutcapoblique.png
> top view annotated/problems:
> http://biomeq.com/meshpicstemp/topviewannotated.jpg
>
> the 'frownyface' shows where the mesh is 'spiking' into the adjacent
> domain... this disables ANY contact algorithm, since some elements are
> 'spiking' or protruding into adjacent mesh domain... a very big problem.
>
> i have tried a wide range of parameter values and even with extreme
> refinement always have the same problem.
>
> i realize that this artifact is likely due to the lack of 'edge preserving'
> in the current 3D mesh implementation.
>
> BUT: does anyone have any ideas or workarounds for this issue?? I cannot
> generate any FEA contact problem given this unfortunate spiking. I have
> tried all kinds of mesh-parameter variations and to no avail. It just
> produces smaller spikes.
>
> Also, if the "edge-preserving" issue is the main problem, is there any
> time-line for solving this? I've noticed that this was proposed for CGAL 3.7
> but would like to inquire if this is really on the books.
>
> I ask with great interest because we have a large potential for grants, etc.
> here at Drexel University but with this artifact we cannot create contact
> meshes... thus, we cannot create the meshes for preliminary data. it's
> crippling. 
>
> We can write grants with funds to support CGAL, but we need to have some
> idea of how to solve this problem and a timeline.
>
> thanks very much in advance for any info.
> best,
> todd
>
>
>
>
>   

--
Mariette Yvinec
Geometrica project team
INRIA  Sophia-Antipolis