Inline visualtions of vedo plots in ipynb

[clemens-fricke]

Das ist nur ein nice to have und was Jae und ich schon durch trial und error herrausgefunden haben, funktioniert es auch nicht perfekt out of the box.

Wenn man das richtige vedo backend eingestellt hat (ipyvtk) diese funktionalität müsste auch noch hinzugefügt werden muss bei show das keyword '''close=False''' gesetzt werden und dann auf dem plotter der dann zurück gegeben wird noch mal show() ausführen.

What needs to be done

• Add settings option for vedo backend
  • gustaf.vedo.settings.VEDO_BACKEND = None
  • Optional VEDO_BACKEND = "ipyvtk"
• Either check if VEDO_BACKEND == "ipyvtk" than always return plot without closing it or give warning that close=false should be set so that .show() can be called on the return value inside the ipynb.

[clemens-fricke]

After some more investigations, I could not get it to work in the same manner jae and I had while we tried it as a spike.

A consistent way to display interactive vedo graphics inline in ipynb is to create the plotter in the notebook and to collect the vedo objects by calling gus_obj.showable() and calling the plotter directly with the returned objects. See example below.

import gustaf as gus
gus.utils.log.configure(debug=True)

# create gus example object
v_res = [20, 30, 40]
vertices = gus.create.vertices.raster(bounds=[[-1, 0, 5], [4, 1, 10]], resolutions=v_res)
connec = gus.utils.connec.make_hexa_volumes(v_res)
v = gus.Volumes(vertices.vertices, connec)

# collect showables (vedo objects)
a = v.showable()

# create the plotter with the correct backend
plt = Plotter(backend="ipyvtk")

# call the show function directly on the plotter with the needed objects
plt.show(a, at=0, interactive=True, camera=None)

# close the plotter to not create a memory leak
# this can be in another cell to keep the figure interactive until you are done with it
plt.close()

[j042]

• Either check if VEDO_BACKEND == "ipyvtk" than always return plot without closing it or give warning that close=false should be set so that .show() can be called on the return value inside the ipynb.

Sounds good. I'd say we can add a function notebook_plotter (or some other name) that wraps show_vedo.

...
plt = gus.show.notebook_plotter([toshow1, toshow2])
plt.show()

in case of multiple plots at once:

>>> plt1, plt2, plt3 = gus.show.notebook_plotter(toshow1, toshow2, toshow3)
>>> plotters = gus.show.notebook_plotter(toshow1, toshow2, toshow3)
>>> isinstance(plotters, list)
True

[j042]

closing for now

[j042]

@clemens-fricke I think we are ready to implement this