promptvv
Turn any video into a structured AI video prompt
Start new thread

promptvv - Turn any video into a structured AI video prompt

by•
promptvv turns an uploaded video—or a Douyin, Xiaohongshu, or Bilibili link—into a structured prompt you can use with Jimeng, Seedance, Kling, and other AI video models. It breaks the source into events, shots, camera moves, actions, performance, and dialogue, and supports follow-up refinements. No manual downloading is needed for supported links. New accounts include 30 seconds of free analysis.

Add a comment

Replies

Best
Maker
📌
I built promptvv while working on AI video remakes with a friend. Rewriting prompts by watching a clip frame by frame was tedious, and ordinary video descriptions were not structured enough to use directly in models such as Jimeng and Seedance. The product accepts a local video or a supported Douyin, Xiaohongshu, or Bilibili link, then breaks it into events, tasks, shots, camera movement, actions, performance, and dialogue. I spent nearly a week refining the output format so the result could be copied into an AI video workflow with less cleanup. New accounts include 30 seconds of free analysis. I’d love feedback on where the reverse-engineered prompt misses the source video or which output format you find most useful.

Would love a side-by-side preview that compares the generated prompt against the original clip with timestamps, so I can quickly spot which event or camera move the parser missed before sending it to Kling.

Maker

 Great suggestion — a side-by-side timeline would make it much easier to verify each event and camera move before exporting the prompt to Kling. This is both useful and feasible, so I’ve added it to our roadmap. Thanks for the thoughtful feedback!

It would be great if promptvv could let you export the generated prompt as a side-by-side timeline overlay with the original video clip thumbnails. That way it's much easier to verify each shot breakdown matches what was in the source before feeding it into Jimeng or Kling.

Dropped in a Xiaohongshu link and it picked out the camera angles and beat-by-beat actions without me touching anything. The structured breakdown saved me a lot of guesswork when I fed it into Kling.

dropped a bilibili link in and it pulled the shots and camera moves without me downloading anything, which saved me a step i didn't even realize i hated