Skip to content

Contributing

Intended contributions

  • add new conversion runners
  • improve preprocessing options
  • expand metric definitions
  • improve docs and onboarding

Adding a new runner

  1. Add a runner script or module under scripts/ following existing style.
  2. Accept image inputs from CLI and expose deterministic outputs.
  3. Extend experiment orchestration with clear preset labels and metadata fields.
  4. Ensure outputs are placed under output/ so the benchmark report can ingest them.

Documentation expectations

  • add or update English docs
  • keep mirrored Japanese docs under docs/ja/
  • keep command examples compatible with UV-powered Python workflows

Review workflow

Open PRs with:

  • clear purpose and scope
  • sample command output (or snippet)
  • explicit note on backward-compatibility for existing outputs

Built for reproducible image-to-SVG benchmarking.