Hands: session 20261005-160713 from 08466c63-aa49-43f6-848f-f3c9086a2992
Contribution to the Frametop hand dataset, uploaded from the hand recorder.
- Session: 20261005-160713
- Contributor: 08466c63-aa49-43f6-848f-f3c9086a2992
- Takes: 10 (8.9 minutes, 5384 frame sets)
- Lighting: indoor
- Objects: keyboard, mouse, gamepad, small (+6 of their own)
- Controllers: straps
- Consent version: 2026-10-03
- Tool: ft-handrec v0.2.0-9-g608ad60
- Size: 9.9 GB
- SHA256 of SHA256SUMS: 42ef2f60ed51f19ccac7063a3a8d140cd0986da82a73de9f8f88fe62dd9eed82
Checked with hands/rec/validate.py: no errors.
Thanks for this session! One small change from the dataset's maintainer, and nothing for you to do or upload again: the session's lighting label now says daylight (it said "indoor").
The recorder measured indoor: its test reads the infrared the room gives off with the headset's lights off, and sunlit windows are too small a part of each picture to move it. Daylight shows in the hands instead: here a hand is a median 1.15 times as bright as what's around it, against 1.5 to 1.7 in lamp-lit sessions, because daylight lights the room about as much as the headset's own infrared lights do. That makes this the dataset's first daylight round, the light hand tracking finds hardest.
Only manifest.json (its session.lighting, with the old value kept under replaced) and SHA256SUMS changed. No pixels changed.
Thank you! Merged into the dataset after review.
A small correction to my earlier comment, from the dataset's maintainer. The daylight label stays: it's based on what your recording shows, big windows lit by the sun in the afternoon. But I gave the hand contrast (hands about 1.15 times as bright as the room around them) as the evidence, and that was too strong: a later session recorded at night in a pale room read almost as low. It goes with daylight, but doesn't show it on its own. Nothing for you to do, and thanks again for the session.