Use transform matrix when getting/setting camera position

[banesullivan]

When scaling a scene (.set_scale), the camera coordinates are transformed to a new projected space. This PR makes sure that when you set the position of the camera using data coordinates, the camera will properly handle those coordinates in the camera's projected space.

One place where many folks noticed this issue is when setting the camera location or focal point which is handled by the orbit_on_path method. The path given corresponds to the data coordinates and not the camera's projected coordinates so if the scene is scaled, the camera could be in a totally wrong location.

Here is a code snippet to demonstrate:

import pyvista as pv
from pyvista import examples

mesh = examples.download_lidar()
z_scale = 2.0

p = pv.Plotter(notebook=False)
p.add_mesh(mesh, render_points_as_spheres=True)

p.set_scale(1, 1, z_scale)
p.show_axes()

p.show(auto_close=False)
path = p.generate_orbital_path(n_points=36, shift=mesh.length)
p.open_gif("~/Desktop/orbit.gif")
p.orbit_on_path(path, write_frames=True)
p.close()

Master Branch

The camera's position is totally away from the mesh because the coordinates were not properly transformed.

This Branch

Ah, much better.

[akaszynski]

Ah, much better.

Nice @banesullivan.

[akaszynski]

Approved. Thanks @banesullivan!