tilegen is a tool that slices high-res images into tiles across multiple zoom levels, similar to Google Maps.
It uses multi-threaded tile generation, high-performance image manipulation, and Bun as its runtime.
Inspired by Google Maps, tilegen cuts images into 256×256 PNG tiles for seamless zooming and panning.
Unlike Google Maps, tilegen does not use Mercator Projection, operating on raw pixel grids instead.
The tool's architecture involves metadata read, zoom level calculation, task generation, worker thread execution, and file output.
MAXIMUM_MAGNIFICATION controls how deep users can zoom in relative to the base image's pixel density.
tilegen adjusts tile quality for different zoom levels to optimize disk space without compromising quality.
Performance optimizations include spinning up worker threads and managing tasks efficiently.
Future plans for tilegen include CLI support, WebP and AVIF tile support, vector tile generation, JSON tile manifests, and potential Web Mercator pre-projection.
tilegen aims to be a fast, flexible tool for generating tiles for any large static image, offering simplicity and power.