Using Common Filters

Contents

Using Common Filters#

Using common filters like thresholding and clipping.

import pyvista as pv
from pyvista import examples

PyVista wrapped data objects have a suite of common filters ready for immediate use directly on the object. These filters include the following (see filters_ref for a complete list):

To use these filters, call the method of your choice directly on your data object:

dataset = examples.load_uniform()
dataset.set_active_scalars("Spatial Point Data")

# Apply a threshold over a data range
threshed = dataset.threshold([100, 500])

outline = dataset.outline()

And now there is a thresholded version of the input dataset in the new threshed object. To learn more about what keyword arguments are available to alter how filters are executed, print the docstring for any filter attached to PyVista objects with either help(dataset.threshold) or using shift+tab in an IPython environment.

We can now plot this filtered dataset along side an outline of the original dataset

a lesson filters

What about other filters? Let’s collect a few filter results and compare them:

contours = dataset.contour()
slices = dataset.slice_orthogonal()
glyphs = dataset.glyph(factor=1e-3, geom=pv.Sphere(), orient=False)

p = pv.Plotter(shape=(2, 2))
# Show the threshold
p.add_mesh(outline, color="k")
p.add_mesh(threshed, show_scalar_bar=False)
p.camera_position = [-2, 5, 3]
# Show the contour
p.subplot(0, 1)
p.add_mesh(outline, color="k")
p.add_mesh(contours, show_scalar_bar=False)
p.camera_position = [-2, 5, 3]
# Show the slices
p.subplot(1, 0)
p.add_mesh(outline, color="k")
p.add_mesh(slices, show_scalar_bar=False)
p.camera_position = [-2, 5, 3]
# Show the glyphs
p.subplot(1, 1)
p.add_mesh(outline, color="k")
p.add_mesh(glyphs, show_scalar_bar=False)
p.camera_position = [-2, 5, 3]

p.link_views()
p.show()
a lesson filters

Filter Pipeline#

In VTK, filters are often used in a pipeline where each algorithm passes its output to the next filtering algorithm. In PyVista, we can mimic the filtering pipeline through a chain; attaching each filter to the last filter. In the following example, several filters are chained together:

  1. First, and empty threshold filter to clean out any NaN values.

  2. Use an elevation filter to generate scalar values corresponding to height.

  3. Use the clip filter to cut the dataset in half.

  4. Create three slices along each axial plane using the slice_orthogonal filter.

Apply a filtering chain

result = dataset.threshold().elevation().clip(normal="z").slice_orthogonal()

And to view this filtered data, simply call the plot method (result.plot()) or create a rendering scene:

p = pv.Plotter()
p.add_mesh(outline, color="k")
p.add_mesh(result, scalars="Elevation")
p.view_isometric()
p.show()
a lesson filters
Open In Colab

Total running time of the script: (0 minutes 6.748 seconds)

Gallery generated by Sphinx-Gallery