Aan de slag
Reel2Prompt — Video to Prompt, Pay Once

Reel2Prompt — Video to Prompt, Pay Once

Found a reel you wish you'd made? Reel2Prompt reverse-engineers any video or image into a ready-to-use, Seedance 2.0-style generation prompt — shot by shot, with real timings. One-time purchase. Runs inside Claude Code on your own machine; your videos never leave it.
#Video#Media#Sociale media
Beoordeling
Meer beoordelingen nodig
Verkocht
0
Gebruikswijze
Downloaden

Stop paying monthly to ask "how was this made?"

Higgsfield, Runway, and every video-AI suite will analyze videos for you — for another monthly subscription. Reel2Prompt does video-to-prompt as a one-time purchase: your Claude Code subscription is the engine, your machine does the work, and there is nothing else to pay. Ever.

What it does

  • Video → Prompt. Point it at any local video (a reel you saved, a Short, an ad, an AI clip). It detects the scene cuts, samples frames through the whole video — not three screenshots — and writes a complete Seedance 2.0-style prompt: global style, shot-by-shot timeline with real timings, camera language, motion, mood, sound, plus consistency constraints. Includes a self-contained prompt for every individual shot.
  • Image → Prompt. Any image becomes one polished, paste-ready generation prompt: subject, composition, style, lighting, palette, mood, medium.
  • Prompt Enhancer. One-line idea in, full structured video or image prompt out.
  • Deep mode (optional). Add your own free Gemini key and Reel2Prompt sends the actual video for native motion + audio analysis. Entirely optional — frame mode works without any API key.

Output is written in Seedance 2.0 prompt structure and works as a strong starting prompt for any video model.

⚠️ Requirements — read before buying

This is a downloadable skill package for Claude Code, not a hosted chat agent.

  • Claude Code required (or a compatible agent client). Built and tested for Claude Code — Anthropic's coding agent (CLI, desktop app, or IDE extension) on a plan that includes it. Also works in other skill-compatible agent clients with a shell and image viewing (Codex, OpenClaw, Hermes). It does NOT run in claude.ai web chat, ChatGPT, or on your phone.
  • ffmpeg required (free, open source). The included setup assistant installs it for you on Windows (winget), macOS (brew), or Linux (apt) — allow ~2 minutes on first run.
  • Your videos must be local files. Download the reel to your machine first; Reel2Prompt does not download from TikTok/Instagram/YouTube for you.
  • Analysis tokens are billed to your own Claude plan (a typical reel costs a few cents of your existing subscription usage — nothing paid to us).

If any of that is unfamiliar, a hosted agent will fit you better than this download.

How it works (90 seconds after install)

  1. Save a reel you love to your machine
  2. In Claude Code: "Turn this video into a prompt: ~/Downloads/reel.mp4"
  3. Get reel-prompt.md: master prompt + per-shot prompts + timing map
  4. Paste into Seedance 2.0 (or any video model) and generate your own version

What you get

  • reel2prompt skill — ffmpeg scene-detection + frame-sampling scripts, Seedance 2.0-style prompt formats
  • reel2prompt-setup skill — guided installer: paste one line into Claude Code and it installs everything, checks dependencies, and runs a test with you
  • Free updates to the v1 package
  • MIT-licensed foundations, full attribution included

Why one-time works

No server, no metering, no quota. The skill is instructions + scripts; your Claude does the thinking locally. That's why this is $12.99 once instead of $15–30 every month.

FAQ

Does it clone or copy the video? No. It writes a prompt describing the style, structure, and motion. You generate your own original content from it.

Does it work on videos of people? It describes people generically. Recreating a real person's likeness needs their consent — the skill flags this automatically.

Videos longer than 60s? Works, best under ~90s (reels/Shorts territory). Longer videos get sparser sampling.

Do my videos get uploaded anywhere? Frame mode: no — everything runs locally. Optional deep mode sends the video to Google's Gemini API under your own key, only if you enable it.


Externe API’s

ffmpeg (local install) · generativelanguage.googleapis.com (optional)