How to Decide Whether a Bad AI Video Needs a Prompt Fix or a Full Regeneration
Brightory creates practical AI guides and workflows based on research, testing, and documented AI usage patterns.
Diagnose the failure before regenerating. Keep a known-good version, change one variable, compare the result, and only widen the intervention when the evidence shows the problem is not local.
Diagnose the failure before regenerating. Keep a known-good version, change one variable, compare the result, and only widen the intervention when the evidence shows the problem is not local.
What is actually failing
This is a generation efficiency problem, not a signal that you need a longer prompt. Modern video models can follow references and motion instructions, but every new angle, occlusion, contact point or camera move introduces visual information the model must infer. The safest workflow is to identify which state must remain fixed and which state is allowed to change.
A local artifact does not justify rebuilding a good shot. Regenerate fully only when the underlying composition, reference choice or motion plan is wrong.
Use a locked-state workflow
- Choose the last known-good state. Save the approved reference or clean frame before changing anything.
- Write down the invariants. Keep a short list of details that must not change; do not bury them inside prose.
- Change one production variable. Add the new action, angle, camera move or interaction without redesigning the subject.
- Reduce ambiguity. If the shot fails, simplify motion, shorten the clip, expose the subject more clearly or add the missing reference angle.
- Verify before extending. Only use a clip as the next reference after it passes a frame-by-frame continuity check.
A prompt structure that is easier to debug
LOCKED: - subject / product identity: [unchanged] - wardrobe / materials / colors: [unchanged] - location anchors: [unchanged] - lighting direction: [unchanged] CHANGE ONLY: - action: [...] - camera: [...] - start state: [...] - end state: [...] DO NOT INTRODUCE: - new wardrobe, props, cuts, camera angles or redesigns - changes to logo, facial identity, product proportions or scene layout
When another reference helps — and when it hurts
Keep motion complexity below the fidelity limit
Verification checklist
- Does the face or product silhouette remain stable at the hardest point of motion?
- Do hair, wardrobe, labels and accessories survive the whole clip?
- Are start position, end position and screen direction correct?
- Did the model add an unrequested cut, prop, person or camera move?
- Can the final clean frame safely become the starting reference for the next shot?
Where the AI Video Kit fits
Want to make this workflow repeatable?
If deciding Whether a Bad AI Video Needs a Prompt Fix or a Full Regeneration is something you need to do more than once, the expensive part is rebuilding the process every time. The Brightory AI Video Consistency & Production Kit packages this kind of work into continuity sheets, shot-planning templates, reusable prompt blocks and production QA checks, so you can carry the method into the next project instead of starting over.
Controlled-generation benchmark
Diagnose the failing visual property before rewriting the entire prompt. The most expensive loops change several variables at once and erase evidence about which control mattered.
- Name the failing property.
- Return to the last known-good state.
- Change one variable.
- Run the shortest useful test.
- Promote only verified results.
Acceptance test
The targeted failure must be fixed without introducing a new continuity problem.
Decision rule
Treat regeneration as a controlled experiment: one failure, one changed variable, one acceptance test.
Diagnostic matrix
| Failure class | First thing to hold constant | First variable to test |
|---|---|---|
| Identity | Reference set | Motion/angle complexity |
| Geometry | Object reference and framing | Contact/occlusion |
| Motion | Subject state | Camera path |
| Environment | Spatial anchors | One shot delta |
Worked example: isolate one failure before spending another full render
If a shot fails in several ways, choose the first property that makes it unusable and hold the rest of the setup stable. Run a shorter, simpler version that tests only that property. Promote the result only after the specific acceptance check passes.
This creates evidence about cause. A full regeneration with a rewritten paragraph may look different, but it tells you little about which control actually solved—or worsened—the problem.
Operational notes for this workflow
Keep a generation log with reference set, prompt, model/settings, the single variable changed and the pass/fail result. When a test fails, return to the last known-good state rather than stacking another fix on top of the failure.
This log is especially useful when credits are limited: short diagnostic generations produce more information per attempt than repeatedly rendering the full shot with several simultaneous changes.
Final continuity checklist
- Known-good state: did this test begin from an approved reference, frame or clip?
- Single delta: can you name the one visual or motion variable that was allowed to change?
- Weakest frame: did you inspect the highest-motion, most occluded or most oblique frame rather than only the opening image?
- Boundary: does the shot connect cleanly to the state before and after it?
- Promotion rule: would you trust this exact result as the reference for the next generation?
If the answer to the final question is no, the clip has not passed continuity even if several frames look excellent. Repair the failed state before extending the sequence.
- Runway — AI Character References: Tips & Techniques
- Runway — Gen-4 Image References
- Google AI for Developers — Veo 3.1 reference images
- Adobe Firefly — Motion reference
- Adobe Firefly — Generate videos using images / first and last frames
AI video controls differ by model and can change quickly. The workflow below distinguishes durable production practice from model-specific features; re-check the linked official documentation before relying on a current limit or control.