Define the repeated relationships
The same footage can appear as a full episode, a highlight, a vertical crop, and a subtitled clip. Identical file checks will miss many of these relationships. Ask for source identifiers and known derivative links, then distinguish exact duplicates from shared intervals and alternate views. Decide what your task considers useful repetition before removing or counting related material.
Track overlapping intervals
When clips derive from a longer recording, preserve start and end offsets where available. Two clips may overlap only partly or repeat the same event from different edits. Keep uncertainty visible if the source mapping is incomplete. A collection’s total runtime should explain whether it counts delivered minutes, unique source intervals, or another agreed unit rather than allowing those units to blur together.
Review similarity findings
Automated screening can suggest repeated frames or sound, but edits, crops, and unrelated recurring graphics complicate interpretation. Inspect candidate groups and separate publisher branding from shared event content. Do not delete every visually similar segment without considering whether it is a new take or a different camera. Retain reviewed relationships so future acquisition rounds can be checked consistently.
Apply grouping to evaluation
Alternate edits and overlapping source intervals should follow the split policy required by the evaluation question. Report exceptions and unknown relationships. Preserve group identifiers after filtering or filename changes so later additions do not reopen a resolved leakage path. Fiund can review a proposed collection’s source coverage and documentation; runtime totals alone should not be presented as independent or commercially cleared inventory.
Review checklist
- Distinguish duplicate files from shared source footage.
- Preserve clip offsets and derivative relationships.
- Review similarity groups before filtering.
- Keep related material grouped under the split policy.