What does Video to Prompt give you?
A structured breakdown and three prompts you can use straight away.
| Part | What it describes |
|---|---|
| Subject | Who or what is on screen, by appearance and action — never by identity |
| Scene | Location, time of day, weather and set details |
| Action | What happens, shot by shot with timings |
| Camera | Movement, speed, lens and depth of field |
| Composition | Framing, angle, height, layers, rule of thirds or symmetry |
| Light | Light quality, direction and colour palette |
| Sound | Music mood, sound effects and ambience |
| Negative prompt | Artifacts to avoid for this kind of shot |
How do you turn a video into a prompt?
- Upload a short clip (MP4, MOV or WebM, up to 60 seconds and 100 MB). Trim to the shot you care about.
- Pick a style: MiniMax H3 (the main person becomes “Image 1”), Detailed (80–160 words) or Short (one sentence).
- Choose a language: English, Chinese or Spanish.
- Copy, edit, or send it to Image to Video to make a new clip of the scene with your own photo.
What does a result look like?
From a 5-second clip of a man at a window table raising a mug (MiniMax H3 style):
Image 1 sits at a wooden table by a bright window, holding a ceramic cup. Image 1 turns toward the camera, smiles softly, and raises the cup in a subtle toast. Medium close-up, soft natural daylight from left, shallow depth of field, subtle push-in, quiet room ambience.
This is a real output from our test of the tool on October 16, 2026.
Is it private and safe?
The clip is analysed by Google Gemini into text and deleted; it is never edited and never used to generate a video. People are described by appearance, not identified. Clips showing sexual content, graphic violence or hate are refused, and prompts naming real public figures are withheld. See the content policy and privacy policy.
FAQ
What does Video to Prompt do?
Google Gemini watches your clip at two frames a second and writes a prompt describing the subject, scene, action, camera movement, composition, light and sound, plus a MiniMax H3 version and a short version.
Is it free?
This tool is completing verification and is not available yet. Planned free limits will be shown before use.
Does it identify the people in my video?
No. People are described by appearance and action only. Comparisons with known people are removed, and real public figures are refused.
Is my video used to make another video?
No. The clip is analysed into text and deleted; it is never sent to a video generation model.