Geodesic Path Issues: Computing length of 1D mesh and finding shortest path doesn't always work

[banesullivan]

The length calculation for the geodesic path is not correctly implemented:

pyvista/pyvista/pointset.py

Line 1413 in 7909cbe

length = self.geodesic(start_vertex, end_vertex).GetLength()

The GetLength() method of a VTK mesh actually returns the length of the diagonal of that mesh's bounding box:

V.GetLength() -> float
C++: double GetLength()

Return the length of the diagonal of the bounding box. THIS
METHOD IS THREAD SAFE IF FIRST CALLED FROM A SINGLE THREAD AND
THE DATASET IS NOT MODIFIED

Note that this is relevant to #252

Pinging @supersubscript - are you using this feature? If so you may be getting incorrect results...

[banesullivan]

Also note that the shortest path is not always found:

import pyvista as pv
from pyvista import examples

sphere = examples.load_globe()

geodesic = sphere.geodesic(10, sphere.n_points-40)

p = pv.Plotter(notebook=0)
p.add_mesh(geodesic, line_width=10, color='red', label='Geodesic Path')
p.add_mesh(sphere, show_edges=True, )
p.camera_position = [-1,-1,1]
p.add_legend()
p.show()

[banesullivan]

The length issue should be resolved by #261 but I'm still concerned that the path is not always the shortest path as shown in #260 (comment)

[supersubscript]

Huh, that's confusing. Thanks for raising the issue, @banesullivan. It does indeed affect my computations.

Also very strange that path isn't always the shortest one.

[banesullivan]

I suppose the shortest path issue is on the VTK side with the vtkDijkstraGraphGeodesicPath algorithm and not something we can fix in PyVista? I’ll post this issue on VTKs discourse to see if anyone there has insight.

[banesullivan]

Posted on https://discourse.vtk.org/t/shortest-path-with-vtkdijkstragraphgeodesicpath/1150

[banesullivan]

Closing issue as I don't think there is anything we can do on the PyVista side

[banesullivan]

So this is actually just an artifact of how I created this Globe mesh - it’s split on the 0 degree meridian so the path has to go to the poles to get around it