AI Video to Prompt
Upload a clip and get a time-coded prompt for every shot. Edit it, then recreate the shot with Seedance, Kling or Veo.
See the shot breakdown and time-coded prompt our analysis returned for each sample clip. Compare the camera movement, lighting and action it describes with what you see.
Red shoe (single)
A stationary side view of a vibrant red leather casual sneaker with cream-colored stitching details around the eyelets and heel tab, paired with off-white textured rubber midsole featuring intricate patterned grooves; captured in soft natural light against a plain grey backdrop as subtle horizontal panning reveals more context of its placement on a smooth concrete floor.
A tabby cat with a striped coat and green eyes
A playful brown tiger-striped housecat stands tall indoors against an unfocused home interior backdrop under gentle illumination; it gracefully shifts between standing fully erect or partially bent at hips while making expressive hand movements—lifting right forepaw to touch nose then swiftly retracting it, alternately moving left rear leg upward before returning back down, finally ending by holding up two outstretched hands.
A man in a black t-shirt and jeans holding a microphone
A stand-up comedian performs live under bright spotlights wearing a plain black T-shirt while addressing engaged crowd, raising right hand for comedic effect.
Red toy sports car
A small bright red sporty model car enters a white platform stage from the right edge moving steadily across its length until stopping at mid-frame oriented front-facing before reversing course smoothly heading off-screen again to the opposite end while text captions appear sequentially below describing each motion step.
The red car moves to the left.
It changes direction.
Then it returns to the right.
This video to prompt generator returns a shot list, a separate list of on-screen text and a one-sentence summary of the overall style.
The clip is described as a list of shots, each with start and end times that stay within the video's real length. It works best on a single scene or a few slow cuts, so check the shot boundaries before you recreate.
Each shot describes framing, camera movement, subject, action, setting, lighting, color grade and mood. A 40–80 word prompt brings those details together.
Captions and other visible text appear in their own list. You decide which words, if any, belong in your next prompt.
Copy an individual prompt, use Copy all for the full sequence, or download the structured shot list as JSON with its timecodes and visual details.
{
"duration_sec": 5,
"overall_style": "Simple, minimalistic",
"on_screen_text": [
"The red car moves to the left.",
"It changes direction.",
"Then it returns to the right."
],
"shots": [
{
"start": "00:00.0",
"end": "00:05.0",
"shot_type": "Wide shot",
"camera_movement": "Static",
"subject": "Red toy sports car",
"action": "Enters from right frame, drives towards left side of screen, stops in center facing forward, then starts driving back toward right frame.",
"lighting": "Even, soft diffused lighting"
}
]
}Use AI video to prompt to carry a reference shot's camera movement and lighting into a new scene. Adapt the prompt, then take it straight to generation.
Choose Recreate on a shot to open the video workspace with its prompt and your selected model already filled in. Generate from that starting point.
Enter a new subject or product in Replace subject with… and choose Adapt. The rewrite preserves camera movement, pacing and lighting, while removing descriptions of captions, watermarks and logos.
Analyze the reference and generate your version on the same site. Your chosen shot goes directly to the workspace, so you can keep working from its prompt.
Go from a reference clip to a prompt you can edit in three steps.
Choose an MP4, MOV or WebM file up to 60 seconds long. Sign in when you are ready to analyze it.
Review each shot's timecodes and visual details alongside its prompt. Check the overall style and the separate on-screen text list.
Edit a shot's prompt or use Adapt to rewrite it for your chosen model. Add a subject replacement if needed, then choose Recreate to continue in the video workspace.
Choose a target model and use Adapt to rewrite an individual shot in its format. Choose Generic for a prompt you can use across models.
For Seedance 2.0, describe the shot's action timing and camera path within your sequence.
For Kling 3.0, spell out how the subject moves and how the camera follows.
For Veo 3.1, describe the framing, scene lighting and order of actions in the shot.
For Grok Imagine, put the subject, setting and visible action into a direct scene description.
For MiniMax H3, describe the camera movement, action rhythm and mood you want to carry forward.
Use video to prompt to study how a reference is filmed and plan which parts to carry into your own scene.
Start from a reference product shot, then replace the item with your own. Keep the camera path and lighting as the basis for a new product scene.
Break a short clip into shots and actions to study its rhythm. Adapt individual shots around your own subject before generating a new version.
Use the time-coded shot list as a written breakdown of framing, movement and scene changes. Export the JSON to keep those details with your planning notes.
Compare what you see in a clip with the words used to describe it. Each shot shows how camera direction, action and lighting fit into a single prompt.
Upload one image to get a prompt for GPT Image, Nano Banana or Seedream. Take the result directly to the image workspace to generate a new version.
Image to PromptUpload a clip and sign in to analyze it. AI video to prompt describes the clip as a list of shots, each with start and end times, visual details and a 40–80 word prompt. Shot boundaries are estimates, so review them, especially on clips with many fast cuts. You can then edit, copy or adapt any shot before taking it to the video workspace.
Upload an MP4, MOV or WebM video up to 60 seconds long. If your reference is longer, trim it to the part you want to analyze before uploading. A single scene or a few slow cuts gives the most accurate shot list.
Currently, only file uploads are supported. You cannot paste a YouTube or TikTok link; download the clip first, then upload the video file.
Use Adapt to format a shot's prompt for Seedance 2.0, Kling 3.0, Veo 3.1, Grok Imagine or MiniMax H3. You can also choose Generic for a model-independent prompt.
The prompt is a reference starting point, so an exact match is not guaranteed. Use the shot's camera movement, action and lighting to guide a new version, then edit the prompt around the details that matter to you.
Yes. Enter your new subject or product in Replace subject with… on an individual shot, then choose Adapt. The rewrite keeps camera movement, pacing and lighting while removing descriptions of captions, watermarks and logos.
Yes. Visible text is listed separately from the shot details, so you can decide whether to include it in a prompt. Adapt removes descriptions of captions, watermarks and logos from the rewritten prompt.
New users receive signup credits to get started. You need to sign in to analyze a video, and both analysis and video generation use credits. See Pricing for the available plans.
You can study a reference's framing, camera movement and pacing to inform your own shots. Use material you have the right to use, and avoid copying someone else's logos, identifiable likenesses or copyrighted content. You are responsible for having permission to use your source material.
Your uploaded video is used only to generate your results. It is not added to the public showcase.
Start with one clip and choose the shot you want to build on. Use the video to prompt generator to get its prompt, replace the subject and continue in the video workspace.