The Sunny Coffee Map

There are lots of maps that can help you find a good cafe in Paris. But Sunny Coffee answers a much more important question: which café terraces are actually in the sun right now?

Created by developer Paul Baron, Sunny Coffee is an interactive map that shows which Paris café terraces are currently sunny - and which are sitting in the shade. The map combines Paris open data, 3D building models, weather forecasts and real-time shadow calculations to help users find the perfect terrace for a coffee in the sun.

Parisians flock to terraces whenever the sun appears, but urban shadows move constantly throughout the day. A café that is sunny at 2pm may be completely shaded by 3pm. Sunny Coffee attempts to solve that problem by calculating sunlight exposure across the city in near real-time.

The technical stack behind the project is particularly interesting from a mapping perspective. Terrace locations come from the City of Paris open data portal, while building footprints and heights are sourced from France’s official IGN BD TOPO dataset. Baron explains that he initially experimented with OpenStreetMap building data, but switched to IGN because of its far more complete and accurate height coverage - essential for reliable shadow modelling.

To calculate shadows, the project uses two different approaches. A lightweight server-side geometric ray-casting system (using SunCalc) precomputes sun exposure for all terraces every day, allowing the overview map to load instantly. When users click on an individual terrace, the application switches to a much more detailed live WebGL shadow simulation using ShadeMap, producing an interactive timeline of sunlight conditions throughout the day.

The blog post documenting the project is also a fascinating look at the realities of building geospatial applications. Baron details a series of debugging challenges involving MultiPolygon geometries, rooftop pixel sampling errors, misaligned terrace footprints and even SunCalc’s unconventional astronomical azimuth system. It’s an excellent reminder that the hardest part of many mapping projects isn’t the visualisation itself - it’s dealing with messy real-world geometry and edge cases.

One of the most compelling aspects of Sunny Coffee is how effectively it hides all of this complexity behind a very simple interface. Users don’t need to understand ray-casting, WebGL shadow maps or cadastral geometry. They simply open the map and discover where they can enjoy a sunny coffee in Paris right now.

You can read Paul Baron’s full write-up here: Sunny Coffee: Building a real-time map of sunny Paris café terraces.