Diving Deeper

Once the Getting Started tutorial has given you a working mental model, these topics go one level down — each is self-contained, so read the ones that match what you are doing.

• Preparing input for tiling — the gpio-optimized GeoParquet contract (WGS84, Hilbert sorting, row-group sizing) and why each part matters for speed and memory.
• Working with the overview file — the geo:overviews format: level bands, metadata, and how the file stays valid, SQL-queryable GeoParquet you can inspect with DuckDB.
• Tuning what appears at each zoom — the quality ladder as one mental model: class ranking, visibility gates, the density budget, clustering, line coalescing, and simplification.
• Tiling remote and multi-file inputs — s3:// / https:// / gs:// byte-range reads, --bbox row-group pushdown, --files-from multi-partition input, and --filter attribute pushdown.
• Keeping memory bounded — the streaming model (memory ≈ O(row group)), the two-pass structure, and what to do when a file is too big for RAM.
• How tylertoo relates to tippecanoe — a factual capability comparison: what each tool does, what only tylertoo does, and which quality-ladder concepts are shared.

For the exhaustive list of every flag and option, see the Reference.