Pointcloud using Point Gaussians

[Findus23]

Hi,

First of all thanks for this amazing library! It is really easy to get great plots in a short amount of time.

There is just one thing I am missing to reproduce a visualisation from paraview:
I am trying to plot a large amount (>2M) points as simple points with low opacity and point_size=1. This works fine, but depends on being on exactly the right zoom level to be able to see the density of the points (as the point size seems to be independent of zoom).

In paraview a great way around this that also makes the visualisation better is using the "Point Gaussian" representation with "shader preset" gaussian blur instead.

Is there any way to get something similar with pyvista?

[banesullivan]

We don't (yet) support the points gaussian representation in PyVista - I'd love to see a PR to the end though.

One workaround is to manually perform the gaussian sampling of the points prior to plotting.

Could you perhaps provide this fantastic sample dataset for us to work on this challenge?

[kmarchais]

Hi,
Thank you for this feature.
I would like to be able to scale point size according to the diameter of my spheres.
In paraview this option is available with "Use Scale Array" : https://github.com/Kitware/ParaView/blob/master/Remoting/Views/vtkPointGaussianRepresentation.cxx
Is there a way to implement this in PyVista if it's not already available?

[banesullivan]

Take a look at glyphing in PyVista:

• https://docs.pyvista.org/examples/01-filter/glyphs.html
• https://docs.pyvista.org/examples/02-plot/movie_glyphs.html
• https://docs.pyvista.org/api/core/_autosummary/pyvista.DataSetFilters.glyph.html

[akaszynski]

BTW, point cloud gaussians is now implemented. See Plotting Point Clouds.

[kmarchais]

Thank you for your answers, I thought point gaussian representation would have better performance than glyphing.

[banesullivan]

It will have better performance yes - my bad. Though I'm not sure if the point size can be scaled by an array with it? Can it, @akaszynski ?