Python trimesh x-cross section produces weird result

Viewed 361

I have a simple box (width: 400, lenght: 400, height: 50):

enter image description here

Here the code to create that box:

import numpy as np
import trimesh

# Vertices
vert = np.array([[-200., -200.,    0.],
              [-200., -200.,   50.],
              [-200.,  200.,    0.],
              [-200.,  200.,   50.],
              [ 200., -200.,    0.],
              [ 200., -200.,   50.],
              [ 200.,  200.,    0.],
              [ 200.,  200.,   50.]])
# Faces
fa = np.array([[6, 0, 4],
              [0, 6, 2],
              [4, 6, 5],
              [6, 7, 5],
              [6, 2, 7],
              [2, 3, 7],
              [2, 0, 3],
              [0, 1, 3],
              [0, 4, 1],
              [4, 5, 1],
              [7, 1, 5],
              [1, 7, 3]])
# MESH
mesh = trimesh.Trimesh(vertices= vert,
                       faces=fa)

The issue

As you can see, the top surface of the box is at z=0, and the lower surface is at z=50.

Now, I would expect to see this clearly, when I do a cross-section, using x as a normal:

# X - Normal => AXIS WRONG
slice_ = mesh.section(plane_origin=(0,0,0), 
                     plane_normal=[1,0,0])

slice_2D, to_3D = slice_.to_planar()
slice_2D.show()

... but what I get is this:

enter image description here

You can clearly see that the box cross-section is not positioned correctly, as it should begin at z=0, and extend to z=50 (horizontal axis of the plot above).

Interestingly, getting cross-sections with z-normal works perfectly:

# Z - Normal => OK
slice_ = mesh.section(plane_origin=(0,0,0), 
                     plane_normal=[0,0,1])

slice_2D, to_3D = slice_.to_planar()
slice_2D.show()

enter image description here

..if I ask for a cross-section at z=-10, where there should be no box, it rightfully complains.

# Z - Normal => OK
slice_ = mesh.section(plane_origin=(0,0,-10), 
                     plane_normal=[0,0,1])

slice_2D, to_3D = slice_.to_planar()
slice_2D.show() 

AttributeError: 'NoneType' object has no attribute 'to_planar'

How can I get the correct cross-section for the x-normal?

Edit

I also posted the question here: https://github.com/mikedh/trimesh/issues/1359

I think you need to specify the matrix explicitly when using to_planar, as if the matrix is unspecified the function has to fit a plane which may not be what you want.

Edit 2

Alternative: maybe one could rotate the mesh by 90 degrees around the x-axis and then do a z-cross, section. I tried that, but somehow it won't work:

from scipy.spatial.transform import Rotation as R
def rotate_vector(vec, rotation_degrees=90, rotation_axis = np.array([1, 0, 0])):
    # Function that turns a vector a given angle about a given axis

    rotation_radians = np.radians(rotation_degrees)
    rotation_vector = rotation_radians * rotation_axis
    rotation = R.from_rotvec(rotation_vector)
    rotated_vec = rotation.apply(vec)
    
    return rotated_vec

# Turn all vertices by 90deg around the x-axis
new_vert = []
for v in vert:
    print(v)
    print(rotate_vector(v))
    new_vert.append(rotate_vector(v))

new_vert = []
for v in vert:
    new_vert.append(rotate_vector(v))

# Create the mesh
mesh_rot = trimesh.Trimesh(vertices= new_vert,
                       faces=fa)

# Create a slice with a z-axis normal
slice_ = mesh_rot .section(plane_origin=(0,0,0), 
                     plane_normal=[0,0,1.0])

slice_2D, to_3D = slice_.to_planar()
slice_2D.show()
    

enter image description here

1 Answers

After long trial and error I found this method to fix it:

def cross_section(mesh, plane_origin=[0,0,0], plane_normal=[1,0,0]):
  
    slice_ = mesh.section(plane_origin=plane_origin, 
                          plane_normal=plane_normal)

    # transformation matrix for to_planar 
    # I don't know why
    to_2D = trimesh.geometry.align_vectors(plane_normal, [0,0,-1])
    
    slice_2D, to_3D = slice_.to_planar(to_2D = to_2D)
    
    return slice_2D, to_3D


slice_2D, to_3D = cross_section(mesh)
slice_2D.show()

enter image description here

Related