for an example from geospatial data. There are probably a few routes forward:
A feature in vitessce which they would likely accept. This is very heavyweight, but could mean the whole thing is decoupled from this package, aside from potentially a dependency here. This could be a good, less intensive starting point for vitessce - scverse integration (scanpy could be too much)
Import vitessce layer for scatterplot and then use https://deckgl.readthedocs.io/en/latest/ over it to generate stream plots (maybe a second layer would be needed)
It would be very cool to somehow do dynamic streamline plots. See
https://github.com/theislab/scvelo/assets/43999641/334d0577-5c14-4b8f-9416-9bd19850b787
for an example from geospatial data. There are probably a few routes forward:
I think each option has an upside and a downside. I think this would make a great hackathon project for me.