CGAL users discussion list

[Olivier Devillers <>]

Le 08/10/15 22:32, Robert W. Johnstone a écrit :
> Hello Everyone,
>
> I understand that there will likely be an impact on robustness, but I 
> would like to know if there is a way to snap the coordinates of a 
> Nef_3 to integer coordinates.
>
> The program will create the union of many polyhedra, which I'm 
> planning to do with an exact kernel to avoid issues, and then 
> post-process the resulting polyhedron to snap the vertices. Are there 
> any suggestions to how this could be done efficiently?
The only work I know on the topic is by Fortune (a bit unpractical)

You can round n triangles (possibly intersecting)
to O(n^4) non-intersecting triangles with endpoints on a grid of step 
1/n and preserving relative positions
with displacement less than 1.