Pounce Atlas is a free, open-source explorer for geographic datasets. The same selected records can be viewed on a South African map, grouped by industry, or explored as dated activity.
Demo: Pounce Atlas — explore business, place and activity
Source: GitHub - zaiqltd/pounce-atlas: Open-source Three.js explorer for business and event datasets: map, industry and activity views, search, share links and local CSV/JSON import. Demo uses fictional data. · GitHub
The released examples contain fictional businesses and venues, with invented names and events placed at approximate town centres. They are demonstrations, not claims about real companies or national economic activity. No Pounce production data is included. Compatible CSV and JSON files can also be imported locally in the browser.
Some Three.js implementation details:
- The dataset records use a single
THREE.Pointscloud and a custom shader with per-point sizes. The dataset points use one draw call, not the entire scene. - Moving between Map and Industries interpolates the position and size buffers instead of rebuilding the geometry. With reduced motion enabled, the points move directly to the new layout.
- An orthographic camera and OrbitControls provide pan and zoom, with a 2D/3D toggle for a tilted view.
- Search and filters carry across the views. A list alternative keeps the records accessible without relying on the WebGL canvas.
The application is TypeScript and Three.js, deployed as static files. The source release needs no Pounce account, API key or private database. Documentation covers the data contract, local imports and deployment.
The application code is MIT licensed. Natural Earth geometry, GeoNames place references and Inter retain their separate notices. Feedback on the interactions and reuse experience is welcome.