Many applications involve constructing possibly non-convex polygons, also called “footprints”, characterizing the shape of a set of points in the plane. Our firefront tracking algorithm, for example, constructs the perimeter of wildfires from the distribution of curated wildfire reports. When the point set contains pronounced clusters and outliers, “good” footprints should consist of disconnected shapes and exclude outliers. Peter and Matt developed and tested a new footprint (above right), extending the well-known characteristic shapes algorithm. Compared with the most widely used alternatives, the algorithm requires fewer parameters and is more robust to certain types of non-uniform point densities. A draft journal paper is now complete, aiming to be submitted in the coming week.