An astrophotography processing pipeline that shows its work.

RAW/FITS ingest, blind plate solving, measured lens and atmospheric correction, color calibration, PSF diagnostics with windowed deconvolution, FITS/ASDF and tabular Arrow export, and a database of every measurement — each value measured, labeled approximate, or absent.

Download View on GitHub

Windows installer. Star data (~1.3 GB total) downloads in-app on first run.

Stars indexed

6.49M

Gaia DR3, G < 15

Quad patterns

20.9M

all-sky index

Scale bands

0.25° to 45.25°

Automated tests

2,400+

every merge gated

as of 2026-07-22

Your rig, measured.

Measures your lens on every frame: distortion, a SIP layer with atmospheric refraction separated out where site and time are known, and a final thin-plate-spline pass for what the parametric models miss.

See how SkyCruncher works.

The full pipeline as an interactive graph — every module, every data seam, every evidence state. Three views: force web, adjacency matrix, and a step-by-step procedure walk.

The SkyCruncher pipeline DAG: an interactive force-directed graph of modules and data seams, with edge layers and evidence states

909 nodes, 1,534 edges, extracted from the codebase and drift-gated.Open the Pipeline DAG →

The measurement corpus.

Every frame processed leaves a measurement receipt, and the receipts add up to a database. These are the project corpus numbers, published as they stand: today's snapshot is the pre-cutover baseline — the honest before. New snapshots land here as the corpus re-runs on the current solver core; the story is in the data arriving.

as of 2026-07-12 · pre-cutover solver · a fresh regeneration replaces this snapshot

Blind solve rate over time

series begins 2026-07-12 · pre-cutover solver baseline · rate axis 0–100%

Solve wall time over time

series begins 2026-07-12 · blind lane, n=82 attempts · median and p90

What it shows you.

Every panel below is rendered from a real measurement record — no mockups, no sample data.

Solve flowchart widget: the end-to-end solve pipeline as a receipt-driven diagram, each stage annotated from the actual run — Solve flowchart — the path this solve actually took, drawn from the receipt. Stages with no evidence read NOT RUN.

Distortion cascade widget: wireframe displacement surfaces for each fitted lens model, with an absent stage shown as NOT MEASURED — Distortion cascade — displacement surfaces for each model the pipeline fitted. An absent stage says NOT MEASURED, not zero.

PSF field widget: per-region median star width and ellipticity across the frame — PSF field — per-region star sharpness across the frame, with approximations labeled as such.

Culling waterfall widget: the detection funnel from raw detections through each cull bucket to matched stars — Culling waterfall — where detections get thrown away, counted at the moment of rejection.

Lens profile — your actual copy of the lens, measured. The 3D surface draws live in-app; this static export carries the measured stats.

Local processing, shared sky.

Solving runs on your machine and your images never leave it. Star data streams down on demand. An opt-in channel for contributing solve receipts and star measurements back to the community database is in development — shared measurements, never your photos.

Contact

[email protected]