Two generations confront each other in a kitchen while the younger person holds the product and something on the stove goes wrong.
Conflict, a physical incident, and the product share the first frame, creating immediate curiosity.
This is more than reverse-prompting. AI reads the hook, conflict, shot intent, and product proof, then delivers three production-ready remake directions.

AI REMAKE STUDIO
Ten keyframes are extracted in your browser before Kimi analysis. The full video is not uploaded as Base64 during analysis.
Analysis sends compressed keyframes, video metadata, and your product brief to the configured Kimi API. It does not send the original video file.
Real analysis example
Two generations confront each other in a kitchen while the younger person holds the product and something on the stove goes wrong.
Conflict, a physical incident, and the product share the first frame, creating immediate curiosity.
Let the toughest family member treat the product as the problem, then prove it is useful with one visible demonstration.
High-heat cooking with steam but no choking smoke
Ten-second realistic family kitchen ad. A mother and daughter argue because the cooking oil is disappearing too fast. The daughter stir-fries greens on high heat to prove the high smoke point; the mother smells the air and places the bottle back by the stove. Fixed natural light, continuous wardrobe, text-free geometric label.
Run a live analysis first. The example cards only preview the structure.
Verified critical path
Analysis sends compressed keyframes only. The reference video is uploaded to the generator only after you approve a plan.
The browser extracts ten keyframes locally instead of uploading a Base64 copy of the whole video.
Kimi returns the hook, timeline, retention devices, continuity locks, and three remake prompts.
Choose a plan, confirm the credit charge, and ask Seedance 2.0 Mini for a ten-second sample.
Identify the hook, escalation, product proof, and payoff. Reuse the effective structure without copying the original content.
Define framing, camera, blocking, product visibility, dialogue, and transitions for every time range.
Generate voiceover, caption rhythm, and music direction, ready for MixVoice presets or authorized cloned voices.
Turn one proven structure into three creative angles, then send the selected plan to Seedance.
Frequently asked questions
No. It reconstructs an executable creative blueprint from visible evidence and never claims to recover the creator's original prompt.
No. The browser extracts compressed keyframes locally and sends only those frames and metadata to Kimi. The video is uploaded only when you click Generate.
The prompt forbids copying real likenesses, trademarks, original dialogue, or distinctive expression. It reuses abstract structure such as pacing and shot intent.
Seedance currently accepts about 1.8–15.2 seconds of reference video. Longer videos can still be analyzed, but you should trim the strongest segment before generating.
Your next product video