I plotted a geographic map with datashader. Everything works fine, however the geographic projection seems to be incorrect. The lengths and aspect ratios do not match, probably because no geographical projection is set? My Code looks like this:
agg = ds.Canvas().points(df, 'lon',"lat")
karte=ds.tf.set_background(ds.tf.shade(agg, cmap=cc.fire,how="log"), "black")
utils.export_image(img=karte,filename='output', fmt=".png", background=None)
That's a nice-looking amoeba! :-)
Datashader is a general-purpose tool, not tied to geographical data in any way. In particular, if you want the data to be projected into some other coordinate system, you need to project it before calling Datashader; all Datashader does is render what it is given.
Datashader does include one utility function for the special case of projecting lon/lat data into Web Mercator (datashader.utils.lnglat_to_meters), for convenience with plotting libraries that use Web Mercator mapping tiles, but if you are visualizing the data on its own (not overlaid on a map) you shouldn't even need that utility; bare lon/lat values should be ok.
If you want some specific aspect ratio, you can determine it with parameters to the Canvas constructor. Specifically, set set the plot_height and plot_width to whatever image size you want, then set x_range and y_range to the ranges of data space you want in that image size (e.g. longitude and latitude ranges like xrange=(270, 290), yrange=(30,40)).
Or you can use the high-level interfaces at hvplot.holoviz.org or holoviews.org to overlay onto a map, which will force a geographic aspect ratio.