Upload a clip and the AI watches all of it, then writes the prompt you would need to make something like it: who is in it, what happens in what order, how the camera moves, the light, and the sound. Copy it, or generate the video right here.
Sign in to start — 3 free prompts a day, no card.
Video to prompt is the reverse of text to video. Instead of writing a description and getting a clip, you give it a clip and get the description back - written the way AI video models read it: the subject, the actions in order, camera movement, lighting, style and audio. It cannot recover the exact prompt someone else typed (no tool can; the same video can come from many different prompts), but it gives you one that makes something close.
MP4, MOV or WebM, up to 60 seconds. Trim longer videos to the part you want first.
Any model, Veo 3, Kling, Seedance or MiniMax H3. Each one reads prompts a little differently, so the wording changes to suit it.
Cuts, camera moves, the order of actions and the final frame are all checked - not just a thumbnail.
Edit the prompt if you like, copy it anywhere, or press Generate this video to make it here.
Tools that grab one frame miss the motion. This one reads the full clip, so it knows a continuous zoom from a real cut and keeps actions in the order they happen.
If the clip cuts off mid-action, the prompt stops there too. Details it could not see clearly are listed separately instead of being guessed into the prompt.
Veo 3 prompts include dialogue and sound, Seedance prompts are split by shot, MiniMax H3 prompts come in its three-part format.
The prompt opens in our generator with the matching model already selected - no copying between sites.
Saw an AI clip on TikTok or Reels and want one like it with your own subject? Get the prompt, change the subject, generate.
See how a clip you like breaks down into subject, action, camera and light - the fastest way to get better at prompting.
Turn a reference clip from a client or a moodboard into a written brief you can reuse and tweak.
Yes. Sign in and you get 3 prompts a day free, no card needed. After that each prompt costs 1 credit. Making the video from the prompt is a separate generation priced by the model you pick; new accounts get 18 free credits for that.
No, and no tool can. The original prompt is not stored in the video, and many different prompts can produce similar footage. What you get is a prompt written from what is actually on screen, which gets you close - then you adjust it.
Not yet. Save the video to your device first, then upload the file. Only use videos you have the right to use.
Up to 60 seconds and 50 MB. For a longer video, trim it to the part you want to recreate - most AI video models only make 5 to 15 second clips anyway.
Pick Any model for a general prompt that works with Veo, Kling, Sora, Runway, Seedance and others, or pick a specific model and the wording is adjusted to how that model reads prompts.
Yes, if the video has audio. It only writes down what it actually hears - speech, music, sound effects. It does not guess lyrics, and if it cannot hear anything it says so.
Video to prompt gives you a description, and a new video made from it will have the same idea but new footage. Video to video keeps your actual clip - the same motion and timing - and only changes the look.
No. The video is deleted as soon as the prompt is written. We keep the text result so you can see what you made.
Anything with nudity, sexual content, graphic violence or minors in unsafe situations is refused, the same rules as the rest of the site.
3 free prompts a day once you sign in, no card. New accounts also get 18 credits to turn the prompt into a video.