Documentation

Field guide

Capture a useful site survey, correct uncertain sectors, and preserve the diagnostics needed to review a real-world field test.

Requirements

Skyfence requires an iPhone with camera, motion, and location access. It works without LiDAR, but a supported Pro iPhone provides metric scene depth and richer nearby-obstacle geometry. Capture outdoors from the actual telescope or camera position whenever possible.

Capture a site

  1. Start a new survey and add a short site note.
  2. Stand near the intended observing position and hold the iPhone steadily.
  3. Rotate through the full horizon at a comfortable pace, including overhead eaves or branches.
  4. Pause when tracking quality drops or the device asks for a slower pass.
  5. Return to weak bearings for targeted recapture.
Do not treat the first sweep as permanent. Skyfence uses revisable evidence. A careful clear rescan can remove a transient false obstacle without discarding good sectors.

Coverage and quality

Unscanned bearings remain gaps. Confidence reflects source agreement, recency, tracking quality, and observation geometry. A survey can still be useful before reaching 100% coverage, but exports clearly identify incomplete or interpolated sectors.

Corrections and rescans

Use sector correction when the automatic skyline is visibly wrong. A manual correction overrides automatic evidence for that sector until you return it to automatic. Reset starts a new survey and removes the saved model, so Skyfence asks for confirmation.

Planning views

The horizon profile is best for reading altitude by bearing. The all-sky view shows angular coverage and gaps. Planning 3D combines skyline, depth, mesh, point cloud, overhang ribbons, and a synchronized sightline ray. Optics planning checks a selected target direction against the same corrected model.

Export a field test

For a useful field report, include a descriptive note and export the full package. It contains the horizon model, visual-dome geometry, notes, and diagnostics JSON. Diagnostics include frame and sample counts, coverage, sensor quality, thermal observations, Vision timing/errors, busy skips, and a bounded event history.

If capture fails before producing horizon coverage, export diagnostics anyway. Zero-coverage reports are intentionally supported because permission, tracking, thermal, and provider failures are valuable evidence.