CGAL users discussion list

[Kim <>]

Dear Simon,

It is supposed to be a structure with holes : ). The signed distance
function is in principle a level-set. With the signedDistance I am taking
the intersection between the implicit poisson reconstructed function that
represents the outer shape and an implicit defined infill that describes the
inner topology. This inner topology is described by a triply periodic
function, in this case a gyroid. These are the holes I guess you are
observing when you set the parameter to 0.01. The inner structure is similar
to what you guys generate in the "3D Periodic Mesh Generation" package if I
am not mistaken. Only there you get uniform repetitions and, although I
removed it from the minimal example to keep things as simple as possible, in
general I don't have uniform repetitions but alter the geometry of the
infill locally thoughtout the volume.

Anyway, when setting the parameter to 0.008 it takes a bit longer than with
0.01, but still under a minute when I run it. At the end it outputs:


Bellow I have added a link with the dump-bug file it generates when it
fails. The structure you will see in the .mesh file is basically what I
expect to get out of it (so it seems to fails towards the end), and I guess
its similar to what you get when you set the parameter to 0.01.

https://we.tl/t-gMM1MJmQdg <https://we.tl/t-gMM1MJmQdg>  

I also don't think anymore its related to the offset, before there where
some problems in the offsets I was creating but I have since solved them
(and thought for a while it solved the problem). 

Although I suspect it to be coincidence, the strange thing is I have not
encountered the error yet when not offsetting my surfaces. I have also
noticed its more likely to happen for lower facetDistance values. Finally,
in the past I have come across some shapes that it sometimes succeeds in
meshing and sometimes not without changing any parameter.

In case it helps to reproduce the problem I have compiled using GCC 8.3 (on
linux) with GMP, MPFR, eigen 3.3.7 and boost 1.67 available and found.

Thanks for taking the time to have a look at the problem!

Best,
K





--
Sent from: http://cgal-discuss.949826.n4.nabble.com/