7/16
Public progress
Evaluation sample available
CP FaceMotion-16 v1.0
A structured face-centric video dataset designed for facial motion analysis, head pose estimation, gaze target research, lip-motion analysis, audiovisual speech research and avatar/animation workflows.
Sample available; full dataset in production
The current public production update covers 7 of the 16 target participants. Final package counts will be published after capture, QA, remote verification and reshoots are complete.
FaceMotion-16
6K source capture
Facial motion
Mouth motion
32
Prompts
2
Repeats
6K
Capture
Dataset card
Verified product facts
A consented face-centric audiovisual video dataset for controlled facial motion, gaze targets, head pose, mouth motion and short speech research.
- Status
- Production in progress. Public progress: 7/16; a 64-clip evaluation sample is available by request.
- Version
- 1.0
- Updated
- 2026-07-14
- Delivery
- Reviewed private delivery
Specifications
- Public production progress: 7 of 16 target participants
- Full target: 16 participants / 2,048 clips
- Per participant: 128 clips
- Prompt design: 32 prompts x 2 lights x 2 repetitions
- Native source: 5888 x 3312 HEVC Main 10 at 25 fps
- Audio: PCM 48 kHz stereo from the capture protocol
- Evaluation sample: 1 pseudonymous participant / 64 clips
- Sample design: 32 prompts x 2 lights x R01
- Sample media: 1080p H.264/AAC / 600.48 seconds / about 83 MiB
- Public previews: 10 stills and 4 preview clips
Capture and provenance
Captured by Continental People under a standardized face-centric protocol with participant releases, controlled lighting, prompt identifiers and synchronized audiovisual recording.
Metadata included
- Pseudonymous participant identifier
- Prompt identifier
- Lighting condition and repetition
- Clip manifest and checksums
- Package documentation
- No landmarks, tracking labels or prepared crops in the current sample
Intended uses
- Facial motion analysis
- Gaze estimation research
- Head pose estimation
- Lip-motion and audiovisual speech research
- Avatar animation evaluation
- Multimodal model training under the commercial license
Rights and commercial terms
The evaluation sample and commercial/presale package use separate published licenses. The current sample is evaluation-only; commercial use requires the commercial license and order terms.
Authoritative terms are in the applicable order and license documentation. Review the buyer license.
Restrictions and product boundaries
- No facial recognition, identity verification, surveillance or biometric identification.
- No emotion scoring, voice cloning, impersonation or deceptive synthetic identity use.
- No claim that landmarks, tracking, biometric labels or prepared facial crops are included.
- No redistribution or public upload of raw media.
Quality assurance
- Current published production progress: 7/16 target participants
- The 64-clip sample covers all 32 prompts in two lighting conditions
- The sample runs 600.48 seconds in a documented 1080p H.264/AAC profile
- 10 stills and 4 short preview clips prepared
- Full-package counts remain targets until production and QA are complete
Package availability
Sample and full-package status
Available by qualified free request
64-clip evaluation sample
One released pseudonymous participant across all 32 prompts and two lighting conditions; evaluation-only license.
In production
16-participant full package
Target 2,048 clips. Commercial presale discussions cover scope, delivery schedule and order-specific terms.
Approved recipients receive the evaluation sample or future commercial package privately after request review and the applicable license acceptance.
FAQ
Buyer questions
Is the full FaceMotion dataset complete?
No. The full 16-participant package is in production. The current public progress is 7/16, while the separate 64-clip evaluation sample is available now.
Does the sample include speech audio?
Yes. Synchronized speech audio is part of the FaceMotion protocol and is retained in the approved sample and preview clips.
Are landmarks, tracking labels or face crops included?
No. The current sample does not include landmarks, biometric tracking labels or prepared face, eye or mouth crops. The high-resolution source can support buyer-created crops where the commercial scope permits.
Next step
Review a package or define a custom capture
Start from the appropriate dataset request, or discuss a project-specific protocol with our production team.
Evaluation preview
See facial motion and audiovisual speech before requesting the sample
These short approved clips preserve synchronized speech audio because audio is part of the FaceMotion capture protocol.
FaceMotion in action
Short video previews with synchronized speech audio. Request the 64-clip sample to evaluate the complete prompt range.
Neutral baselineSpeech audio
Facial motionSpeech audio
Head poseSpeech audio
Mouth motion + audioSpeech audio
Available now
64-clip evaluation sample
- 64 accepted video clips
- 1 released adult participant
- 32 standardized prompts
- 2 controlled lighting conditions
- Synchronized wireless-lavalier audio
- Evaluation-only sample license
Full dataset in production
16-participant capture design
- Target: 16 adult participants
- 32 standardized prompts
- 2 repetitions per prompt
- 2 controlled lighting conditions
- Native 6K source capture at 25 fps
- Synchronized wireless-lavalier audio
- 4K and 1080p delivery editions
- 6K source access scoped separately
Designed for
Research and evaluation workflows
- Facial motion analysis
- Gaze estimation
- Head pose estimation
- Lip motion analysis
- Audiovisual speech research
- Avatar animation systems
- Multimodal model training from high-resolution source video
Product boundaries
What the package does not claim
- The high-resolution source supports buyer-created face, eye-region and mouth-region crops. Prepared crops are available only when separately scoped.
- Landmark annotations, biometric tracking and pre-labeled facial-analysis outputs are not included.
- Not intended for facial recognition, identity verification, surveillance, emotion scoring or voice cloning.
Licensing and access
Start with the sample or discuss the complete package
The evaluation sample is licensed for internal evaluation only. Commercial presale discussions cover the planned complete package, delivery schedule and order-specific terms. Every request is reviewed before controlled delivery.