Compare the four formats on one requirements list

Write the decisions the record must support before choosing a format: source provenance, rights scope, stable traits, controlled variation, timeline state, attempt findings, approvals, version history, and restricted access. A narrative document explains nuance well but can hide stale values. A spreadsheet makes rows, owners, and filters explicit but handles visual relationships awkwardly. A card board helps teams scan shots and references yet may weaken field consistency. An asset-linked database offers traceable relationships and permissions but costs more to configure and govern. Score only these workflow properties; do not infer generation quality from the record system.

• Traceability and version control
• Timeline and visual context
• Reviewer effort
• Access and retention risk

Test each candidate with the same change scenario

Use an authorized fictional example and introduce a controlled episode change: new wardrobe, temporary injury, revised prop, and one expired reference. Ask an editor to locate the current rule, identify affected shots, record an exception, approve the update, and produce a minimized handoff. Measure steps, ambiguous fields, missed dependencies, duplicate values, access exposure, and the ability to reconstruct the decision. Keep the exercise about information design. It does not require a video output, a current AI service, or evidence that any format preserves a character automatically.

Choose a primary record and define its failure controls

Select the smallest format that remains clear under the expected cast size, episode count, team structure, and approval burden. Name one canonical record, define who may change it, set version and archive rules, and forbid untracked local copies from becoming authoritative. If visual boards or documents remain useful, link them to stable IDs in the primary register and mark their update responsibility. Schedule a migration trigger for scale, security, or collaboration changes. The comparison ends with a reversible editorial decision and known trade-offs, not a ranking of named software or a claim of tested integration.