Skip to content

Big networks

Keep a map fast as the network grows: inline data, embedded vector tiles, or the GPU.

import roadstyle as rs

rs.render_edges(edges).save("map.html")               # up to ~10⁴ edges
rs.render_edges(edges, tiles=True).save("map.html")   # ~10⁵ edges, still one file
rs.render_edges(edges, backend="lonboard",            # millions, in a notebook
                color_by="maxspeed_kmh", cmap="magma", width_by=(1, 5))

roadstyle edges.gpkg -o map.html            # inline
roadstyle edges.gpkg --tiles -o map.html    # embedded vector tiles

Up to ~10⁴ edges: the default

The saved page inlines every edge (gzipped) and opens from disk with no server. Nothing to set. Below ~10⁴ edges this is the simplest and just as fast as the options below.

Towards ~10⁵ edges: tiles=True

tiles=True embeds the roads as a PMTiles vector tileset in the same HTML file.

  • The browser parses only the tiles in view, so a ~100k-edge map opens in seconds, not ~10 s.
  • Low zooms carry simplified geometry.
  • The JavaScript API, popups and selection work exactly as with inline data.
  • The file is somewhat larger and the Python build is slower (about a minute at ~100k edges).
  • Needs the tiles extra: pip install "roadstyle[tiles]".

Millions of edges: lonboard

backend="lonboard" draws on the GPU (deck.gl) in a Jupyter notebook. Same call and colours, but fewer cartographic touches: one colour and width per edge, no casing, labels, arrows, legend or JavaScript API. Needs the lonboard extra. Explore at scale there, then ship a styled subset with the default backend.

Choose a backend

web (default) folium lonboard
Comfortable size ~10⁴ edges, ~10⁵ with tiles=True ~10³-10⁴ 10⁵-10⁶+
Road widths scale with zoom fixed pixels fixed pixels
Two-way lanes, arrows, street names yes no no
Bridges and tunnels grade separation, 3D draw order only no
Legends yes yes no
Colour menu (color_options), overlays, JS API yes no no
selected= pre-highlighted edges no (use rsSelect) yes no
Output one offline HTML file HTML (Leaflet) notebook widget

folium is for maps that must live inside a folium workflow. selected= is folium-only and takes a GeoDataFrame of the edges to highlight:

rs.render_edges(edges, backend="folium", selected=edges[edges.name == "Götgatan"])

See also: Every parameter · Put it on a website