CGAL users discussion list

[Olivier Devillers <>]

Jens K. Becker a écrit :
> Yes, they are for tetrahedrons. Thats where the infinite points come 
> in. If you
> give the triangulation a set of points in a plane, the triangulation 
> algorithm
> will automatically insert points so that it can make tetrahedrons.
> Theoretically, they are "just somewhere in space", meaning very far 
> away from
> the points you give it. It then uses your points and these infinite 
> points to
> make tetrahedrons. This leaves you with a set of cells (each cell is a
> tetrahedron) and one point of each cell is an infinite point. Get rid 
> of these
> and you have triangulated your plane. Thats how I do it.
>
> Jens

There is always exactly one point at infinity. If your points are coplanar
t->dimension() will be two and only the vertices 0,1 and 2 are 
meaningful (vertex 3 can be ignored).

If you want to triangulate coplanar 3D points, the best is to use 2d 
triangulation on 3d points as suggested
by Andreas, either with one of the suggested traits class (that 
basically ignore one coordinate) or
with your own traits class.