Intro to 3D Visualization in python

pyvista

pyvista is your best bet to create informative visualizations of 3D data from within python. Yes, matplotlib and veusz have beautiful 3D plots as well but these are mostly restricted to relatively simple types of visualizations.

Your best bet to install pyvista maybe

pip install 'jupyterlab>=3' ipywidgets 'pyvista[all,trame]'

as described at their userguide.

```{python}
import numpy as np
import pyvista as pv
import ipywidgets as widgets
import gala.potential as gp
import astropy.units as u
pv.set_jupyter_backend('trame')  # or 'server'

nx, ny, nz = 140, 140, 140
extent = 60. 
x = np.linspace(-extent, extent, nx)
y = np.linspace(-extent, extent, ny)
z = np.linspace(-extent, extent, nz)
X, Y, Z = np.meshgrid(x, y, z, indexing='ij')


potential = gp.MilkyWayPotential()
rho = potential.density([X, Y, Z] * u.kpc).to_value(u.Msun / u.pc**3) #.reshape(nx, ny, nz)
r = rho.flatten()

grid = pv.StructuredGrid(X, Y, Z)
grid["rho"] = rho.ravel()

plotter = pv.Plotter(notebook=True,lighting='three lights')
plotter.enable_shadows()

def plot_iso(threshold=np.log10(np.percentile(r, 75.))):
    plotter.clear()
    iso = grid.contour(isosurfaces=[10**threshold/2,10**threshold,10**threshold*2], scalars='rho')
    
    # Compute per-vertex normals
    iso = iso.compute_normals(cell_normals=False, split_vertices=True, auto_orient_normals=True)
    
    # Add the isosurface mesh
    plotter.add_mesh(
        iso, opacity=0.3,
        smooth_shading=False, 
        show_edges=False, cmap="plasma")

    
    # Eye-dome lighting can help
#    plotter.enable_eye_dome_lighting()

    # Update the scene in the same cell
    plotter.show(interactive_update=True)

widgets.interact(plot_iso, threshold=((np.log10(np.percentile(r, 0.1)), np.log10(np.percentile(r, 99.9)), 
                                    (np.log10(np.percentile(r, 99.9))-np.log10(np.percentile(r, 0.1)))/100)));
```