vtk in jupyter

Hi.I’ve integrated Python vtk in Jupyter lab, however, executing the code opens the renderer window at the server machine not embedded inside Jupyter like: https://youtu.be/s-g3E3aAgEo . Examples from: https://nbviewer.jupyter.org/gist/certik/5723420
All good with ITK and load properly in the Jupyter notebook as output.
Same for ParaView. It calls the application to load at the server side not embedded. However, in case of ParaView, it has been designed to fully control the application remotely by Jupyter notebook (https://gitlab.kitware.com/paraview/paraviewjupyterkernel).
Is there anything missed or done wrong that couldn’t load vyk renderings embedded in Jupyter notebook?


Hi Ali,

They are several approach that are moving forward the integration of VTK/ParaView into Jupyter.

Here is a list of what we currently have available



As @Sebastien_Jourdain mentioned, there are lots of ways that people have been using jupyter with vtk; I can speak to the last one he mentioned.

Panel is a neat project that already has some nice features of vtk exposed. I have been working on improving Panel’s communication between VTK and vtk.js (which is how server-side VTK data gets rendered on the client side). This work will hopefully be completed soon after VTK’s next release (within the next month or so).

@ben.boeckel will a VTK 9 binary Python package (wheel) be available that supports remote rendering?

What module enables remote rendering?

There is not a single module. Configuration support for headless rendering in a CPU-only Docker image, e.g with osmesa.

Ah. We can certainly have an OSMesa-enabled package I think. I imagine users would have to bring their own libOSMesa though.

1 Like

That would be very helpful!