Looking for a density_mapbox without a map

I am using a plotly express scatterplot (opengl) to render a large field of data points.
I let the user navigate around using input sliders and changing the center of the graph.
This makes the number of visible data points manageable, and the graph responsive.
Example https://signs.2b2t.website/

I am looking for a better way to pan around the map. I can use config=dict({‘scrollZoom’: True}) and panning to move around, but since I the data is cropped, zooming (scrolling) out does not do much. Meanwhile, zooming in does not suddenly improve the performance, as all those points are still being calculated.

Now, I have been experimenting with relayoutData and a resampling function, to update the graph while zooming / panning and ‘downsize’ the dataframe when a lot of points are shown. The resampling works well, but the relayoutData feels very buggy.

Now i see how mapbox density heatmap plots have basically everything I need (zoomable, performant, some sort of resampling), but I cannot figure out if I can use this without an actual map. I need something like the mapbox density heatmap, but with a normal blank 2D plane.

Is it possible to tune a mapbox plot the way I need it, or should I keep trying with relayoutData?

Hello @Entropy5,

Welcome to the community!

Did you check out here?

Thank you for the response
I have no idea why i had disregarded this option but it is indeed well applicable
The way points aggregate in the density mapbox still looks cooler to me but I think this will do
Just to be sure, if i want to zoom in really far on the density heatmap and retain detail, i should still write something that redoes the figure on relayoutData change, right? I need to have an okay figure both at x-range 60million and x-range 6000.

To achieve this kind of behavior, you might want to take a look at HoloViews,