How to get proper views of point cloud data in numpy-stl?

Question

I am trying to get front, back, left, right, top, bottom after projecting the point clouds of 3D data. But the results are not good. Is there any better way to get the required views? I have done up to this using numpy-stl documentation and some research for projection.

import numpy as np
import pandas as pd
import os
import time
import scipy
import math
from scipy import ndimage
from stl import mesh
from matplotlib import pyplot as plt
from mpl_toolkits import mplot3d
from matplotlib import cm
%matplotlib inline

# Load an existing stl file:
my_mesh = mesh.Mesh.from_file('Bolt.stl')

# Create a new plot
figure = pyplot.figure()
axes = mplot3d.Axes3D(figure)

# Load the STL files and add the vectors to the plot
axes.add_collection3d(mplot3d.art3d.Poly3DCollection(my_mesh.vectors))

# Auto scale to the mesh size
scale = my_mesh.points.flatten()
axes.auto_scale_xyz(scale, scale, scale)

# Show the plot to the screen
pyplot.show()

I have download that bolt.stl from grabcad.com.

After that, I have tried to project it's different views like:

points = my_mesh.vectors
x = points[:,0].flatten()
y = points[:,1].flatten()
z = points[:,2].flatten()

fig= plt.figure()
ax= fig.add_subplot(111, projection= '3d')
ax.scatter(x,y,z)

ax.plot(x, z, 'r+', zdir='y', zs=100)
ax.plot(y, z, 'g+', zdir='x', zs=-100)
ax.plot(x, y, 'k+', zdir='z', zs=-100)

ax.set_xlabel('X')
ax.set_xlim([-100, 100])
ax.set_ylabel('Y')
ax.set_ylim([-100, 100])
ax.set_zlabel('Z')
ax.set_zlim([-100, 100])

plt.show()

Result is so bad, I cannot even know, where goes the bolt 3D.

I have also tried:

X = points[:,0]
Y = points[:,1]
Z = points[:,2]

fig = plt.figure()
ax = fig.gca(projection='3d')
ax.plot_surface(X, Y, Z, rstride=8, cstride=8, alpha=0.3)
cset = ax.contourf(X, Y, Z, zdir='z', offset=-100, cmap="gray")
cset = ax.contourf(X, Y, Z, zdir='x', offset=-100, cmap="gray")
cset = ax.contourf(X, Y, Z, zdir='y', offset=100, cmap="gray")

ax.set_xlabel('X')
ax.set_xlim(-100, 100)
ax.set_ylabel('Y')
ax.set_ylim(-100, 100)
ax.set_zlabel('Z')
ax.set_zlim(-100, 100)

plt.show()

I have also tried with this code to, which I have found from "MVCNN for 3D shape recognition "

def point_cloud_to_volume(points, size=32):
    vol = np.zeros((size,size,size))
    locations = np.copy(points)
    locations[:] *= (size-1)
    locations = locations.astype(int)
    locations[locations >= size] = 0.0
    locations[locations < 0] = 0.0
    vol[locations[:,0],locations[:,1],locations[:,2]] = 1.0
    return vol

def vol2depthmap(volume_array, bg_val=32):
    vol = volume_array.argmax(2)
    vol[vol == 0] = bg_val
    return vol
v = point_cloud_to_volume(points)
%matplotlib inline
from matplotlib import pyplot as plt
plt.imshow(vol2depthmap(v), interpolation='nearest', cmap='Greys')
plt.colorbar()

Finally, I have tried:

front_proj = points[:, :, :].sum(axis=1)
plt.plot(front_proj)
plt.show()

side_proj = points[:, :, :].sum(axis=0)
plt.plot(side_proj)
plt.show()

top_proj = points[:, :, :].sum(axis=2)
plt.plot(top_proj)
plt.show()

I have tried using slicing, but as you have already guessed what will be the result.

front   = points[:,:,0] 
back   = points[:,:,-1] 
top    = points[0,:,:]  
bottom = points[-1,:,:] 
r_side = points[:,-1,:] 
l_side = points[:,0,:]

How to get the proper views of .stl file like below. Please ignore drafting and dimensions.

Or any other cad format will do. I tried it in .stl file as there is no python-library I found so far, to load other cad formats than stl.

Answer 1

Have you tried this?

figure = plt.figure()
axes = mplot3d.Axes3D(figure)

axes.add_collection3d(mplot3d.art3d.Poly3DCollection(mymesh.vectors))
axes.view_init(90,270) # top view
scale = mymesh.points.flatten('F') #C, F, A, or K
axes.auto_scale_xyz(scale, scale, scale)

plt.show()

Play with the view_init combinations of 0,90,180,270 to get the other projections.

Answer 2

Here's something you can do:

vertices = my_mesh.vectors.reshape(3*len(my_mesh.vectors), 3)
vertices = np.unique(vertices, axis=0)
x, y, z = zip(*vertices)

figure = plt.figure()
axes = mplot3d.Axes3D(figure)
axes.scatter3D(x, y, z)
plt.show()