First: can the viewer read the beat?
Show the sequence to someone who has not read the prompt. Ask what happened, where attention moved, and what the character wanted. The answer should come from picture and sound, not a spoken explanation by the filmmaker. Record confusion as an editing or coverage issue before reaching for higher resolution.
Look at the first and last frame of each shot. Does the action enter and leave in ways the adjacent shots can use? An image that looks strongest halfway through may provide a poor handoff. If the scene's purpose is unclear, revise the beat and coverage; polishing texture cannot repair missing information.
Second: inspect continuity at the seam
Watch at normal speed, then step through each cut. Check screen direction, position and shape of props, wardrobe, eyelines, shadow direction, camera height and the scale of movement. Generated hands or background details deserve attention only where the audience's eye goes; a tiny artifact outside the point of interest may be less damaging than a beautiful shot that reverses the action.
Run a sound-only pass. Is dialogue intelligible? Do space and perspective remain stable? Does a generated ambience mask an edit? For a hybrid scene, compare photographed and generated material for colour, grain, motion cadence and depth cues. The aim is not uniformity at any cost; a deliberate visual break can be part of the storytelling.
Third: review the release object
Open the actual export at its intended size and with captions. Check titles, aspect ratio, audio levels and accessibility. Then inspect a separate provenance ledger: source images, model and version, licensed music, performer consent, voices, stock footage, and any disclosure requirements. A good cut is not a complete release file if its ingredients are unaccounted for.
Use red/amber/green notes by issue: red blocks the scene's meaning or lawful release; amber needs a documented choice; green is acceptable. Those labels are a team communication device, not a numerical quality score. Keep the rejected takes and reason codes. They make the next scene's production plan more honest.
Sources & context
- Screen Method · Does it hold across the cut? (companion method)
- C2PA · Content Credentials explainer (provenance scope; checked 16 Sep 2026)
The rubric is editorial synthesis. C2PA provenance can document an asset's history but does not certify the factual truth of its content.