Video to Prompt
Paste a YouTube link. The AI watches the video and gives you a style prompt, every shot with its framing, camera and light, and a ready-to-use prompt for each one.
- 1
Copy a YouTube link
Any public video or Short works: youtube.com/watch, youtu.be or youtube.com/shorts links.
- 2
Pick your video model
Choose Veo, Kling, Sora, Runway or "Any model". The prompts are written the way that model reads them best.
- 3
Break it down
The AI watches the picture and listens to the sound. A short video takes 10 to 30 seconds.
- 4
Copy what you need
Copy the style prompt, one shot's prompt or the whole breakdown, or download it as Markdown or JSON.
From a video you like to prompts you can use
Describing a video well enough for an AI model to reproduce its feel is hard. You need the shot sizes, how the camera moves, what the light does, how long each shot lasts and what the sound is doing. Video to Prompt watches the video for you and writes all of that down, shot by shot, with a prompt for each shot that you can paste straight into a video model.
What you get
- A style prompt: one paragraph that captures the overall look, color, pacing, camera language and sound. Use it as the base of every prompt in your project.
- Every shot: start and end time, framing (close-up, wide…), camera (static, pan, dolly, handheld…), light and color, the action, and any text on screen. Each timestamp links to that moment on YouTube.
- A prompt per shot: 40 to 80 words that recreate the shot with original subjects, written for the model you picked.
- How it was made: live action, AI generated, 3D or 2D animation, screen recording or motion graphics, with the clues the AI used, such as a "Made with Veo" watermark or physics that don't look real.
- Sound: whether there is voice, music and sound effects, and how the rhythm works.
Prompts written for your video model
Each model reads prompts a little differently, so pick yours before you start:
| Model | How the prompts are written |
|---|---|
| Veo | One cinematic paragraph with camera, lens and light, plus a short line about sound. Clips are up to 8 seconds. |
| Kling | Subject and motion first, then scene, camera and style. Clips are 5 or 10 seconds. |
| Sora | Like a line from a film script: camera, subject, action, time of day, look and mood. |
| Runway | Camera move first, then subject, action and style, in one or two sentences. |
| Any model | Plain sentences covering subject, action, setting, camera, light and style. |
A long shot in the original may be longer than one generated clip. Split it into two prompts, or let the model's own extend feature continue it.
Free and paid
Everyone gets one free breakdown a day, no account needed. The free breakdown covers the first minute of the video, which is the whole thing for most Shorts.
With credits, you get the whole video up to 10 minutes for one credit per started minute, so a 3-minute video costs 3 credits. Detailed mode costs two credits per minute and reads up to 2 minutes. Credits are the same ones used for transcripts: packs start at $5 for 30 and never expire. A breakdown that fails costs nothing, and opening one you already paid for again within a week is free.
Limits worth knowing
- Only public YouTube videos and Shorts for now. TikTok and Instagram are next.
- Shot timestamps are close but can be off by a second or so on fast cuts.
- The AI may merge very short, near-identical shots into one.
- Videos the AI's safety filter flags can't be analysed, and you're not charged for them.
ReelToScript is an independent tool and has no affiliation with YouTube, Google, OpenAI, Kuaishou or Runway.
Use cases
Frequently asked questions
It turns an existing video into text prompts that describe it: the overall style, and each shot's framing, camera move, light and action. You paste those prompts into an AI video model to make something in the same style.
Public YouTube videos and Shorts, from youtube.com/watch, youtu.be or youtube.com/shorts links. Private, members-only and age-restricted videos can't be read. TikTok and Instagram are not supported yet.
Every visitor gets one free breakdown a day, without an account. It covers the first minute of the video. With credits you get the whole video, up to 10 minutes, at one credit per started minute. Credit packs start at $5 for 30.
Detailed mode uses a stronger model that splits shots more finely and describes camera moves more precisely. It reads up to 2 minutes and costs two credits per minute.
The prompts describe the style and the shots, and they are written to use your own original subjects, without real people's names or brands. Use them for inspiration and your own work, and respect the original creator's rights.
No. The AI reads the public YouTube link directly; we never download, store or republish the video. We keep the text breakdown for a week so opening it again is instant.
The breakdown covers what you see and hear, not a word-for-word script. For the exact words, use our YouTube transcript generator; there's a link under every breakdown.