finds.dev← search

// the find

OvertureMaps/overture-tiles

★ 162 · Java · MIT · updated Jun 2026

Create PMTiles from Overture data

Overture Tiles is a Docker-based pipeline that converts Overture Maps open map data into PMTiles format using Planetiler, then pushes results to S3. It's for mapping engineers who want self-hosted vector tiles from Overture's global dataset without building the pipeline from scratch. The official Overture Maps org maintains it, so it tracks releases directly.

Covers all six Overture themes (base, transportation, buildings, addresses, places, divisions) in one image rather than requiring separate pipelines per theme. BBOX filtering means you can generate regional extracts instead of processing the full global dataset, which matters when Overture parquet files run into the hundreds of GB. The AWS Batch architecture is the right call for this workload — burst compute, no idle infra cost, and it sidesteps the memory constraints you'd hit on a fixed instance. The Terraform companion module means you can stand up the full stack in a reproducible way rather than clicking through the console.

Profiles are baked into the Docker image, so you can't customize layer definitions without rebuilding — fine for consuming Overture data as-is, but a hard wall if you need to filter attributes, rename fields, or adjust zoom ranges for your use case. The pipeline is AWS-only; there's no GCS or Azure Blob output path, and the download stage pulls directly from Overture's S3, so running this outside AWS will cost you egress fees. With 162 stars and 22 forks it's lightly adopted, which means limited community bug reports and the issue tracker probably doesn't surface edge cases across the full diversity of Overture releases. No local-first workflow for the full pipeline — local testing requires Docker and hits real S3 for the download stage unless you use SOURCE_OVERRIDE with your own pre-staged data.

View on GitHub → Homepage ↗

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →