Point 3D models with jupyter notebooks

[enordstrom]

Hi! A beginner here, but very excited.

I'm interested in the 3D gltf model feature. It works great in kepler.gl/demo and is just awesome for putting some additional tiny assets on the map. But not so great when I try jupyter lab notebooks (pip installed). The default duck is not loaded, nothing is shown and the browser notes an 'invalid scenegraph' from deck.

Do you have any tips and tricks here? Maybe CORS-related?

[B13rg]

Are you able to load config?
I have had taken a slice of data and loading it into the demo to roughly setup the data bindings.
Then you can export the map as json, and load it into your notebook via json parsing to dict.
You will need to update the data ID in the config or the python config in your KeplerGL() func

If you don't have a map token, be sure to set the map style to plain "mapStyle": {"styleType": "no_map"},
I did this in jupyterlab 4.4.3 with !{sys.executable} -m pip install keplergl, and have success using the notebook to export html.

If you have the html file, it should be all self contained so you just need to serve it, since you should already have python installed: python3 -m http.server