Video Analyzer

Analyze videos and generate storyboard breakdowns. Extracts frames via scene detection, transcribes audio with Whisper, analyzes visuals with Claude Vision, and computes 8-field stylistic fingerprints.

Features

Frame Extraction — Scene-detection-based keyframe selection (or fixed intervals)
Audio Transcription — Timestamped transcription via OpenAI Whisper
Visual Analysis — Per-frame descriptions using Claude Vision
Stylistic Fingerprint v3 — 8-field deterministic classification (rendering class, world type, character strategy, narrative structure, visual abstraction, visual density, camera language, tonal positioning)
Storyboard Output — Combined shot-by-shot breakdown as .docx or .md

MCP Server

This project includes an MCP (Model Context Protocol) server so you can use video analysis directly from Claude Desktop or Claude Code.

Install via Claude Code

claude mcp add video-analyzer -s user -- uvx video-analyzer-mcp

Install via Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "video-analyzer": {
      "command": "uvx",
      "args": ["video-analyzer-mcp"],
      "env": {
        "ANTHROPIC_API_KEY": "your-key-here"
      }
    }
  }
}

MCP Tools

Tool	Description
`video_analyze`	Full pipeline: download, extract frames, transcribe, analyze, fingerprint, storyboard
`video_extract_frames`	Extract representative frames using scene detection or fixed intervals
`video_transcribe`	Transcribe audio with OpenAI Whisper
`video_fingerprint`	Generate 8-field Stylistic Fingerprint v3 classification
`video_check_deps`	Verify all required dependencies are installed

Standalone Usage

pip install -r requirements.txt
python3 analyze_video.py "https://youtube.com/watch?v=..." --output-dir ./output

Prerequisites

Python 3.10+
FFmpeg — brew install ffmpeg (macOS) or apt-get install ffmpeg (Linux)
ANTHROPIC_API_KEY — set as an environment variable

Stylistic Fingerprint Fields

Rendering Class — Stylized 3D, Flat 2D, Minimalist Line Art, Textured 2D, Mixed Media, Photoreal
World Type — Stylized Real-World, Abstract Concept Space, Data/Presentation Space, Fictional Metaphor Universe
Character Strategy — None, Mascot-Led, Single Narrator, Single Protagonist Arc, Ensemble Cast
Narrative Structure — Direct Explanation, Step-by-Step, Problem-Solution, Analogy, Myth-Busting, etc.
Visual Abstraction Index — 1 (Photorealistic) to 5 (Maximum Abstraction)
Visual Density — Minimal, Sparse, Moderate, High
Camera/Editing Language — Cinematic, Social Vertical Punch, Presentation Deck, Static Slides, etc.
Tonal Positioning — Institutional, Corporate Professional, Gen Z Social, Child-Friendly, Dark Editorial

License

MIT

video-analyzer

Quick Install