Possible next steps

[jourdain]

The idea is to solve the various pain points that exist from getting the data to the actual visualization in and outside of Jupyter.

• [ ] Getting data
  • [ ] Zarr store?
  • [ ] Data Pane https://datapane.com/
  • [ ] Girder like?
• [ ] From data to vtk
  • [ ] Ingest numpy, xarray, dataframe, ? into vtk meshes
    • [ ] binding Schema
    • [ ] manual binding
    • [ ] Helper Source (zarr, else?)
  • [ ] Leverage new Python inheritance to expand python API
    • [ ] Figure out ingest API per mesh type
    • [ ] Data Field API
      • [ ] numpy, xarray, dataframe (in + out)
    • [ ] Utility classes to extract info to streamline viewer input
• [ ] From vtk to viz
  • [ ] Do we extend VTK rendering helper (source/representation/view)?
  • [ ] Extend pyvista plots to support pipelines?
    • [ ] Add trame backend
  • [ ] Add a couple of dedicated + tunable light weight viewers in trame
• [ ] Pan3D specific code
  • [ ] Some small trame viewers?
  • [ ] Some commonly used UI component for viewer?
  • [ ] Some utility function/tools that can't be in vtk, pyvista, ...

I would suggest to tackle "from data to vtk" first.

[jourdain]

To help in the further discussions @aashish24 @banesullivan @johnkit

[jcfr]

@thewtex

[aashish24]

thanks, @jourdain @johnkit and I had our idea for "data to vtk" and "Pan3D specific code" and we will share our idea shortly

[johnkit]

With regard to SEB's "From vtk to viz" section, Aashish and I discussed the idea of implementing a Pan3D viewer in Phase I of the SBIR. This would be an ipywidget with proposed features:

• Accepts different types of 3-D data including vtkDataObject and Xarray.DataArray instances. Maybe Pandas or numpy arrays with some TBD accompanying schema.
• The internal conversion of input data to vtkDataObject instances for rendering should be done in the vtk python module.
• Can use Trame and/or PyVista underneath (TBD separate or switchable implementations?)
• Can work with ipywidgets for interaction, e.g., time selection, pan/zoom, opacity.
• Works with Jupyter kernel save-restore state
• Provides both client and server-side rendering
• Can be used to capture and export animations

A question that comes to mind is how would this viewer relate to ipyvtklink and maybe ipyparaview and whether or not we can expand or reuse code from either of those projects.