ClipWhat is a subject?
A named person, item or place associated with a project and its clips. One record can connect several examples: an athlete’s face and uniform, a dish from different angles, or a venue. You review the proposed appearances and confirm the matches. Try a restaurant shoot.
How does ClipLogger know who's in a clip?
ClipLogger can group similar face evidence and use reviewed subject records, visible appearance, text, and project context to propose associations. A person becomes a known subject because you name, import, or confirm that record. It does not establish an unknown person's real-world identity. Helmets, occlusion, distant numbers, and similar uniforms can leave a match ambiguous.
How do subjects get created?
Two ways: you create them, or a roster seeds them. Importing a roster CSV (jersey number, name, position…) mints a person-subject per row with the number bound. New-subject creation is deliberately roster/lookup-locked for AI flows: the model can propose linking an existing subject but can't invent a brand-new person, which is what keeps hallucinated names out of your cast.
How does the jersey→athlete join work?
A roster lookup maps a visible code to a known record, while the review workflow lets you evaluate the evidence and confirm the association. Duplicate numbers need context such as team, season, or role. A helmet does not provide a face match, and a number alone is not always unique. Keep uncertain appearances pending or reject them rather than forcing a named result.
What does "pending" mean on a subject or suggestion?
It needs review. Inspect the crop, the clip, and the candidate subject before confirming or rejecting it. Some evidence and proposal records are saved for review and recovery even while the final association is unconfirmed. The meaning of pending must not be simplified into a promise that no data has been written.
How do I review subjects efficiently?
Open Subject Review and work through one subject and its instances at a time. Arrow keys move focus; Return or A confirms the focused instance, and X rejects it. Shift-A and Shift-X apply the corresponding action to the focused subject's instances, so inspect the scope before using them. Review Matches and crop review can show related evidence without requiring another identical presence decision.
Can the same person exist in two projects?
A subject may be relevant to multiple projects, but each project's cast, field values, and reviewed associations have their own context. Check for duplicate subject records before linking more appearances. Use the available merge and review controls deliberately; a matching display name is not proof that two records represent the same person.
Do I get bounding boxes?
Where spatial evidence is available, the review surfaces can show a crop or box around the relevant subject. Inspect it to see what the proposal is grounded in. A box is evidence for a particular frame or occurrence, not a guarantee that the same person has been tracked perfectly through every frame of the clip.