Laplacian edges seem to jump?

[FranzEricSchneider]

I was looking at the edges for the skeleton_graph (thanks for #3 by the way) and it looks like the edges are jumping from spot to spot and skipping a lot of the skeleton. Is that intended?

Brown(ish) is the original cloud, blue is the skeleton (contracted points) and green is the skeleton_graph (open3d LineSet)

[FranzEricSchneider]

Another example:

[meyerls]

Hi @FranzEricSchneider,

Sorry for the late response. The green line set is the topological representation of the skeleton. In the next days I will review the documentation and the code to make things more clear. Stay tuned!

[FranzEricSchneider]

Is a non-topological set of skeletal lines available that follow the shape better? For the use I was trying this for, for example, the topological lines weren't of use since they don't follow the object shape.

[meyerls]

No, currently it is not implemented. I will implement it in the next update!

[FranzEricSchneider]

No worries, for reference what I was trying to do what go from (point cloud -> skeletal line segments). I think the default Laplacian Contraction on Points method from the paper doesn't really do that, it just calculates the compressed points. I think there are a number of ways that line segments could be generated on compressed points, depends on which approach you find the most simple and compelling.