How to setup radius or nbr of neighbor pt for curvature calcuation ( mesh.Comput_Curvature)

[DRLing2021]

In our project, curvature, gaussan and mean, are used to identify concave pts from convex pts. We tried several uitilities and found vedo's performance is satisfying. In fact, the curvation results relate to neighbour pts which is controled by search radius or nbr of neighbour pts. Unfortunately we could not find the setting of radius for Compute_Curvature() function.

Can anyone help to tell what is default value setting for the radius or nbr neighbour pts in vedo curvature calcuation? How to change them when needed? The function we used is: mesh.compute_curvature().

[marcomusy]

Hi, soory for the late reply ! unfortunately I don't think there is an option to specify the radius in vtk/vedo.
For relatively small meshes you may consider averaging over connected vertices, eg:

from vedo import *

# s = Sphere(res=150).wireframe().c("yellow")
s = Mesh(dataurl+"teddy.vtk").subdivide()
s.compute_curvature()
s.cmap("coolwarm", "Gauss_Curvature", vmin=-10, vmax=10).add_scalarbar()

# select one point on the sphere using its index
index = 150

# recursively find connected vertices
ids = {index}
for _ in range(5):
    l = []
    # todo: subtract the already found vertices
    print(len(ids), "vertices found")
    for i in ids:
        l += s.connected_vertices(i)
    ids = set(l)  # remove duplicates

ids = list(ids)
vtxs = s.vertices[ids]

# create a red point at the selected point's location
apt = Point(s.vertices[index]).c("red5").ps(15)
cpts = Points(vtxs).c("blue5").ps(10)

# show the sphere, the selected point, and the connected vertices
show(s, apt, cpts)

[DRLing2021]

Thank you Mark!