Start with nonnegotiable evidence
List the tasks, viewpoint, visible interactions, and metadata your model requires. Existing footage may fit part of the brief while lacking a critical angle or task outcome. New capture can address specified conditions, but feasibility still needs review. Separate essential evidence from preferences before comparing the two routes so a large archive does not distract from missing requirements.
Inspect what can be recovered
Some gaps can be addressed through annotation or documentation; others cannot. A reviewer can label an observable action after recording, but cannot create an unseen contact event or a missing synchronized signal. Ask for representative full sequences and available capture records. Treat inferred metadata differently from information recorded during the original session.
Evaluate control and naturalness
A newly designed task can provide deliberate variation and consistent recording units. It can also introduce staging if participants adapt their movements to the camera. Existing naturalistic footage may offer unplanned behavior while providing less control over coverage. State which tradeoffs matter for your model and record how each source meets them, rather than assuming one route is universally better.
Review the proposed use separately
Technical suitability does not establish permission for commercial training, evaluation, redistribution, or another use. Ask what source documentation is available for the specific proposal. Send Fiund both the essential capture requirements and the intended use so the discussion can distinguish sourcing fit, possible additional collection, and outstanding review questions before any delivery commitment.
Review checklist
- Rank essential evidence separately
- Identify recoverable and missing information
- Document staging and capture control
- Review source-specific use requirements