Marco Musy

You are right... for some reason vtkSelectPolyData does not pass the data associated to points and faces (it should)... What about creating a point id array and then interpolating it:...

Hi the first thing i notice is that one face you create is reversed. but that doesn't seem to be the origin of the problem, in fact your code above...

I also guess it is the `subdivide()`.. checking intersections in a nested loop might become slow (?), you may consider writing a custom `intersectWith()` function if needed..

``` vedo --info ``` ``` vedo version : 2022.3.0.dev0 https://vedo.embl.es vtk version : 9.0.3 python version : 3.9.7 (default, Sep 16 2021, 08:50:36) [Clang 10.0.0 ] python interpreter: /Users/mmusy/Software/miniconda3/bin/python vedo...

I just tested it on vtk9.2.0 and it crashed python without the `subdivide()`, and it works fine with it. if your planes are just 4 coplanar points you may consider...

> Yeah, basically anything that you do to alter those specific points (subdivide, a shift, rounding the verts, etc) make it work. yes... it really looks like a numerical instability...

OK, thanks for investigating... the tolerance should be to define the how close the algorithm should check i guess, I submitted an issue to the vtk developers as it's definitely...

yes, sure, this is the thread: https://discourse.vtk.org/t/vtkintersectionpolydatafilter-crashing/9428

I tested on `vtk9.2.0rc2` but my guess is that the bug is there in all versions.

Hello :) I guess it's possible with a bit of work from the Delauney triangulation which is the dual of Voronoi: https://en.wikipedia.org/wiki/Delaunay_triangulation you can use `delauney2d(...)`, then use `cell_centers()` to...