promptvv

promptvv is a Chinese-language web tool that reverse-engineers existing videos into structured prompts for AI video generators. Users upload a clip or paste a public share link from Douyin, Xiaohongshu or Bilibili, and get prompts formatted for Jimeng, Seedance, Kling, Hailuo and Doubao. It is built for short-video account owners, e-commerce sellers, short-drama writers and Chinese content teams who want to study how a reference video is put together.

A single multimodal model reads the picture and the audio along one timeline. It tracks people, products, scenes and props, and picks up voiceover, dialogue, music and ambient sound in context. Before starting a job, the user chooses a video type such as dance, product promotion or short drama, one of three output formats, and the target model's clip length limit of 15 seconds, 30 seconds or a custom value.

Each result holds an event summary, image prompts for scenes, characters and props, and video prompts broken into tasks and shots with framing, camera movement, timing, action and dialogue. Videos longer than the model limit are split into several tasks automatically. A follow-up chat can rework the prompt with new characters, products, plot, lines or style. Results download as TXT or Markdown, and batch jobs, history search, an API and MCP access are available.

Usage is measured in seconds of video, and failed jobs are not charged. New accounts get 30 seconds free, and paid packs cost ¥19.9 for 600 seconds, ¥29.9 for 1800 seconds and ¥56.9 for 6000 seconds, each valid for 365 days. Batch submission comes with the two larger packs and a priority queue with the largest. Local files can be MP4, MOV, WebM or M4V, up to 500 MB and 10 minutes long.