Plotting PolyData slices with Matplotlib

[banesullivan]

It's possible to plot pyvista.PolyData (vtkPolyData) with matplotlib, so I thought I'd share a snippet to do just that!

In this example, I take a slice that has a Z normal which makes extracting the planar data for Matplotlib very straightforward. In the case where the slice does not have a normal along a cartesian axis, you'd have to rotate the mesh such that it does. I don't have an immediate solution for that so I'll try to post that in the comments below later. If anyone else knows how to rotate meshes from a best-fitting plane to the XY plane, please share! But note that it would mess up the axis labeling for Matplotlib

Adapted from https://perso.univ-rennes1.fr/pierre.navaro/read-vtk-with-python.html

import matplotlib.pyplot as plt
import pyvista as pv
import numpy as np
from pyvista import examples

# Some dataset
mesh = examples.load_uniform()

# Slice and triangulate
#-Note that this slice is on the XY plane
#-for any other orientation, we'd have to rotate it
#-to have a Z normal
#-also note that this requires the data to be on the nodes, not cells
data = mesh.ctp().slice("z", generate_triangles=True)

# Grab data for Matplotlib
x = data.points
tri = data.faces.reshape((-1,4))[:, 1:]
u = data.active_scalar

# Plot it yo!
plt.tricontourf(x[:,0], x[:,1], tri, u)
plt.gca().set_aspect('equal')

[banesullivan]

And here is another interesitng way to plot orthographic slices with Matplotlib. The axes alignments and scaling could use some work.

import matplotlib.pyplot as plt
import pyvista as pv
import numpy as np
from pyvista import examples

def slice_ortho(mesh, x=None, y=None, z=None, scalars=None,
                fig=None):
    mesh.set_active_scalar(scalars)
    # order: YZ, XZ, XY
    slices = mesh.ctp().slice_orthogonal(x=x, y=y, z=z, 
                                   generate_triangles=True, 
                                   contour=False)
    
    p = pv.Plotter(off_screen=True, notebook=False)
    p.add_mesh(slices)
    _, perspective = p.show(screenshot=True)
    
    def plot_mpl(ax, slc, u, v):
        pts = slc.points
        tri = slc.faces.reshape((-1,4))[:, 1:]
        val = slc.active_scalar
        
        ax.tricontourf(pts[:,u], pts[:,v], tri, val)
        return
        
    axii = [(1,2), (0,2), (0,1)]
    
    if fig is None:
        fig = plt.figure()
    
    ax1 = fig.add_subplot(2, 2, 1)
    #ax1.set_aspect('equal')
    ax2 = fig.add_subplot(2, 2, 2, sharey=ax1)
    ax3 = fig.add_subplot(2, 2, 3, sharex=ax1)
    plts = [ax1, ax2, ax3]

    for i, slc in enumerate(slices):
        plot_mpl(plts[i], slc, *axii[i])
        
    ax4 = fig.add_subplot(2, 2, 4, )
    ax4.imshow(perspective)
    ax4.get_xaxis().set_visible(False)
    ax4.get_yaxis().set_visible(False)
        
    return fig


mesh = examples.load_channels()

fig = plt.figure(figsize=(10,6))
slice_ortho(mesh, fig=fig)
plt.tight_layout()
plt.show()

[banesullivan]

I imagine that we could easily take @prisae's awesome Matplotlib slicer tool in discretize and add callbacks to the slice method in PyVista for an interactive, matplotlib based orthographic slicer of any PyVista dataset

[banesullivan]

So I just created a hacked version of your slicer, @prisae. It's over in https://github.com/banesullivan/pyvista-mpl-slicer

The callbacks don't seem to be working.... not sure what exactly is broken, but it's mostly there!

[prisae]

Awesome!