Pricing
Priced per source video-hour, prorated by measured duration. Annotate is the default. Review your plan's cost before processing. Geometry is quoted per job. A
fixed-price tier always costs less than its parts. You only pay for delivered
outputs. Hand boxes, Hand keypoints: a free Preview on sampled frames.
3D hand pose: Not available yet.
-
Inspect
$15.00/ source video-hour
Parts bought one by one: $30.00/source video-hour
- Base (upload check, decode, cleaning, exports)
- Objects
- Face blur (on request through the assistant: a separate blurred copy of an uploaded video, not part of a run; not charged)
Default sampled hands use FastCrest Hand Landmarks v1, with hand boxes as downloadable JSON files. Optional FastCrest Hand Placement v1 requires a Geometry quote before processing.
-
Annotate
$40.00/ source video-hour
Parts bought one by one: $120.00/source video-hour
- Base (upload check, decode, cleaning, exports)
- Objects
- Labels
- Face blur (on request through the assistant: a separate blurred copy of an uploaded video, not part of a run; not charged)
- Hand keypoints (Preview: free on sampled frames)
Default sampled hands use FastCrest Hand Landmarks v1, with hand boxes as downloadable JSON files. Optional FastCrest Hand Placement v1 requires a Geometry quote before processing.
Camera path and depth need a job quote; metric accuracy is not established.
-
Dense 2D
$90.00/ source video-hour
Parts bought one by one: $255.00/source video-hour
- Base (upload check, decode, cleaning, exports)
- Objects
- Labels
- Masks and tracks
- Face blur (on request through the assistant: a separate blurred copy of an uploaded video, not part of a run; not charged)
- Hand keypoints (Preview: free on sampled frames)
Default sampled hands use FastCrest Hand Landmarks v1, with hand boxes as downloadable JSON files. Optional FastCrest Hand Placement v1 requires a Geometry quote before processing.
Camera path and depth need a job quote; metric accuracy is not established.
FastCrest Latent Actions v1 and optional FastCrest Motion Tokens v1 are Preview outputs from selected frame pairs. These are not robot commands; policy improvement has not been measured.
-
Geometry
Quoted per job
Camera, depth and 3D scope require a job quote and qualification.
Per-deliverable add-ons
| Deliverable | $/source video-hour |
| Base (upload check, decode, cleaning, exports) | $24.00 |
| Camera poses | Quoted per job |
| Depth | Quoted per job |
| Objects | $6.00 |
| Labels | $90.00 |
| Face blur | on request through the assistant: a separate blurred copy of an uploaded video, not part of a run; not charged |
| Masks and tracks | $135.00 |
| Hand boxes | Preview: free on sampled frames |
| Hand keypoints | Preview: free on sampled frames |
| 3D hand pose | Not available yet |
Optional outputs and scale estimates
FastCrest Device Map Depth v1 uses a compatible source-bound device map. Embedded IMU scale estimates need motion sensors in the original recording, a compatible camera path and calibration. These inputs may be unavailable; detected sensors alone do not establish metric scale.
Wearer height can help estimate hand size. A declared hand length takes priority. Neither input establishes 3D accuracy. 3D hand pose: Not available yet.
Robot-log upload accepts Parquet today for a free local channel and alignment preview. Full-size support, fitting, sensor calibration and contact verification are not qualified or available here. Review the processing estimate before starting any extra processing.
Free sample: your first 3 clips of up to 120 seconds each, on any tier, at no charge. Partial runs are
charged only for delivered deliverables. Hand outputs are never charged.