Sample data outputs · Stereo
What a stereo recording looks like.
- Views
- Left + right, calibrated
- Sync
- Single hardware clock
- Extras
- Depth · IMU · hand pose
- Formats
- RLDS · HDF5 · zarr
Sample clip
A real stereo clip, about 14 seconds.
Downscaled to 1920 × 1080 so it loads on a web page. Delivered episodes carry the full-resolution views with their calibration.
Download the sample
egocentric_stereo_sample.mp4 ↓The web preview of a stereo recording, exactly as shown here. No form required.For the mono 360 recordings with matching IMU rows, see the main sample outputs.
Per stereo episode
What makes stereo data trainable.
Each stereo episode is delivered with the streams and metadata below, so depth and 3D labels are recoverable rather than estimated.
Left + right video
Two lenses at a human interpupillary baseline, so depth is recoverable from every frame pair.
Calibration
Intrinsics, extrinsics and the stereo baseline are measured every session and shipped with the episode, so rectification is exact rather than assumed.
Hardware sync
Left and right frames trigger on one clock alongside depth, IMU and audio, so there is no drift to correct between eyes or between sensors.
3D labels
3D hand joints, object poses and contact events are annotated against the stereo geometry, not guessed from a single frame.
Spec
Stereo capture at a glance.
- Streams
- Stereo RGB · depth · IMU · audio
- Calibration
- Per session, shipped with episode
- Annotation
- 3D hand pose · object pose · contacts
- Stereo delivery
- Raw frames + calibration; rectified pairs and disparity on request
- Formats
- RLDS · HDF5 · zarr
- License
- Perpetual, buyer-owned
Stereo gives dense depth aligned pixel-for-pixel with the RGB view and works outdoors where active depth sensors struggle. Many programs capture both, synced on one clock.
Want stereo data for your task?
Send the activity and object list you need. Stereo episodes come back calibrated, synced and in the format your pipeline reads.