zodiacal

Find your place among the stars.

A sky image has an address, even when you do not know where the camera was pointing. zodiacal matches four-star patterns against a prebuilt catalog index, fits a sky projection, and verifies the proposed alignment.

A field of dots. A place in the sky.

Choose a synthetic camera field, rotate it, and ask zodiacal to find its pointing. The solver receives only pixel positions and brightness; the known coordinates stay outside the solver for comparison afterward.

For a field anywhere on the sky, open the sky explorer →

Rotate the pattern. Keep the match.

Rotate or resize this four-star pattern. The distances change together, so their ratios stay the same. Geometric invariants make a pattern recognizable before its orientation or scale is known.

An illustration of geometric invariance, not a plate solve. Zodiacal uses normalized four-star hash codes, catalog indexes, projection fitting, and verification. A distance ratio alone is not enough to identify a field.

Add a plate solver to your stack

cargo add zodiacal

To use the published command-line solver, install with the CLI feature:

cargo install zodiacal --features cli --locked

For browser WebAssembly, use the same published solver with native data services disabled. Load the index in memory and run the solver in a worker, as this page does:

zodiacal = { version = "0.5", default-features = false }

Solve an image against a catalog

A solve needs an image and a matching-scale index. Index files contain catalog star positions and geometric hashes; the package does not bundle an all-sky index.

zodiacal solve image.png \
  -i my_index_00.zdcl \
  -i my_index_01.zdcl \
  --timeout 60

1. Extract sources

   Find candidate point sources in the image.

2. Match patterns

   Form quads and search a kd-tree of catalog patterns.

3. Fit the sky

   Use matched correspondences to compute a TAN-WCS projection.

4. Verify

   Check the proposed alignment against known catalog positions.

From source detection to a sky solution

Index choice matters: its angular scales must cover the field you want to solve. Build index files from a starfield binary catalog, and keep field size, limiting magnitude, memory, and search timeout in mind.

Build your first index → · Read the benchmark conditions and results →