Video to Prompt Generator

Turn any video link into AI video prompts

Paste a link. We cut the video into shots, read every frame and write a ready-to-use prompt for each shot — plus the hook, structure and why it works.

Works with TikTok, YouTube Shorts, Instagram Reels, X, Facebook and Xiaohongshu. First 60 seconds analyzed.

Want the finished video, not just prompts? Remake it with AI

What is video to prompt?

Video to prompt is reverse prompting for video: you give it an existing video and it writes the text prompts that would recreate it with an AI video model. Felo's free video to prompt generator takes a TikTok, YouTube, Instagram, X, Facebook or Xiaohongshu link, splits the first 60 seconds into shots, reads the frames and returns one prompt per shot (subject, action, camera, setting, lighting, style), one prompt for the whole video, and a breakdown of the hook, structure and why it works.

How video to prompt works

1
Paste a link
Any public TikTok, YouTube Shorts, Instagram Reel, X or Facebook video.
2
We break it down
Scene cuts, timing, camera, subject and action for every shot.
3
Copy the prompts
One prompt per shot plus one for the whole video, written for the model you pick.

Example: a viral miniature-building Short

9 shots in 10 seconds, 1.1s per shot. Three of the generated prompts:

Shot 10.0–0.7s
Close-up overhead shot of two hands carving and assembling a small wooden craft on a workbench, working with a wooden board and a window-shaped cutout. Outdoors in a forest with soft natural light and a blurred green background; realistic, tactile, handheld documentary style.
Shot 20.7–1.4s
Close-up side view of two hands using a hot glue gun on a small cardboard house on an outdoor wooden table, cardboard parts scattered nearby. Softly blurred green forest background, natural daylight, realistic satisfying-craft style.
Shot 99.1–10.0s
Wide high-angle reveal of a finished miniature thatched cottage with a warm glowing window, a pebble path and a tiny wooden bridge over a stream on the forest floor. Soft dusk light, cozy, highly detailed miniature diorama, slow gentle push-in.

Source: a 10-second YouTube Short of hands building a tiny forest cabin.

Reverse engineer any video prompt

Video to prompt works like a reverse prompt: instead of writing a prompt and hoping for the look, you start from a video that already works and get the prompts behind each shot. Use them in Veo, Sora, Kling, Seedance, Runway or any image-to-video tool.

Viral remake

Felo Video to Prompt at a glance

Input	A public video link — TikTok, YouTube (Shorts and long videos), Instagram Reels, X, Facebook, Xiaohongshu
What's analyzed	The first 60 seconds, up to 24 shots
Output	One English prompt per shot (40–80 words), one prompt for the whole video, plus hook, structure and why it works
Written for	Veo 3, Sora 2, Kling, Seedance or any model — you pick before you start
Breakdown language	English, Chinese, Japanese, Korean, Russian, Hebrew or Arabic (prompts stay in English)
Time	Usually 1–2 minutes
Price	Free with a Felo account, 5 videos a day, no credit card
Want the video itself?	Felo Viral Remake turns the same link into a finished AI video

Video to prompt for Veo 3, Sora 2, Kling and Seedance

Pick the model before you paste the link and every prompt is written for it. Then copy the prompts into the tool you use:

Veo 3

Paste the shot prompts into Google Flow or Gemini. Generate shot by shot and cut them together in the original order to keep the pacing.

Sora 2

Use the whole-video prompt for a single generation, or the shot prompts one by one in the Sora app for more control.

Kling

Works with Kling text-to-video, or add a reference image and use the shot prompt as the motion description in image-to-video.

Seedance

Use the shot prompts in Seedance through Dreamina, CapCut or the API. Short, single-action shots work best.

Video to prompt FAQ

Is the video to prompt generator free?

Yes. Log in and you get free analyses every day. No credit card needed.

Can I paste a TikTok link?

Yes. Paste the TikTok video link directly. YouTube Shorts and long videos, Instagram Reels, X, Facebook and Xiaohongshu links work too. Private or login-only posts can't be read.

Which AI video models are the prompts for?

Pick Veo 3, Sora 2, Kling, Seedance or any model. Prompts describe subject, action, camera, setting, lighting and style, so they also work in Runway, Hailuo or Wan.

How long can the video be?

Any length, but only the first 60 seconds are analyzed — that's where the hook and pacing are.

Does it copy the original video?

No. You get text prompts that describe each shot. Faces, names, logos and on-screen text from the original are never copied into the prompts.

Can I turn a dance or AI video into a prompt?

Yes. Paste the dance video link and you'll get the moves, camera and style as prompts you can reuse with your own photo in an image-to-video model.

Can I upload a video file instead of a link?

Not yet — paste a public link. If your video is only on your phone, post it as unlisted on YouTube or private-link on TikTok first, then paste that link.

What's the difference between video to prompt and Viral Remake?

Video to prompt gives you text prompts to use in other AI video tools. Felo Viral Remake skips that step: it keeps the video's hook, cuts and pacing and generates the finished AI video for you.