Skip to content

Tab 3 — Recording

Synchronised recording across the selected cameras, with automatic download and synchronisation (hand claps or LED flash, per the bottom-bar Sync setting).

🚧 TODO: screenshot.

Recording a trial

  1. Enter a trial name and select the cameras.
  2. Start — then perform two loud hand claps within the first 3 seconds. (Two claps enable a consistency check; one clap works, but without cross-validation.) With Light sync, the LED flashes instead — no clapping; see LED light sync.
  3. Perform the movement, then Stop.
  4. Files are downloaded from each camera into [project]/sessions/[session]/[trial]/video/, and synchronisation runs automatically with the method set in the bottom bar — synced files appear in video/synced/.

A session/trial tree view at the bottom shows all recorded trials.

Sync outputs (per trial)

  • Trimmed MP4s in synced/ — start-aligned and end-trimmed to identical duration. Originals are never modified.
  • sync_onsets.png — detected clap onsets on each camera's audio envelope (audio sync), or sync_led_signal.png — ROI brightness with the detected LED on/off instants (light sync).
  • stitched_videos.mp4 — a 2×2 grid preview for quick visual verification of sync.

If the sync fails

A red SYNC ISSUE — TRIAL DISCARDED popup shows the sync table and the reasons. Clicking OK deletes the trial folder so you can simply re-record. A WARN in the sync table always leads to this: the two claps (or the LED on and off instants) gave offsets that disagree. See Known issues → Synchronisation for claps and LED light sync → If it fails for the LED.

Sound source position (speed-of-sound compensation)

At high frame rates, differences in camera-to-clap distance cause sub-frame sync errors (~3 m ≈ 8.8 ms). Set Sound source X / Y / Z (metres, in the calibration coordinate system); with a calibration loaded, sync automatically subtracts the differential propagation delay. Values persist in go2kin_config.json. Not applied during extrinsic calibration or Set Origin.

🚧 TODO: when to bother (e.g. ≥100 fps with metres of distance difference) and how to measure the source position.