AIMDB – AI-Powered Movie Analyzer 🎬

An intelligent system that analyzes movies through both visual and audio content, generating detailed expert reviews and ratings using GPT-4o-mini for visual analysis and Claude 3.5 Sonnet for screenplay (transcribed via whisper API) evaluation.

• Frame Analysis: Extracts keyframes and creates visual summaries
• Audio Processing: Transcribes movie dialogue using OpenAI's Whisper API
• Expert Panel: Generates diverse AI expert reviews with unique perspectives
• Comprehensive Analysis:
  • Visual aesthetics (GPT-4o-mini)
  • Screenplay quality (Claude 3.5 Sonnet)
  • Narrative structure
  • Technical proficiency
  • Cultural impact
  • And more!
• IMDB-Style Reviews: Includes punchy, memorable comments from each expert
• Beautiful Output: Colorful, emoji-rich formatting for engaging results

• Python 3.8+ (recommended: Python 3.9-3.11 for best compatibility)
• FFmpeg installed on your system
• OpenAI API key (for GPT-4o-mini and Whisper)
• Anthropic API key (for Claude 3.5 Sonnet)

1. Clone the repository:

git clone https://github.com/yourusername/aimdb.git
cd aimdb

2. Create and activate a virtual environment (recommended):

# For Python 3.9-3.11 (recommended)
python -m venv venv
source venv/bin/activate  # On Unix/macOS
venv\Scripts\activate     # On Windows

3. Install dependencies:

# For Python 3.9-3.11
pip install -r requirements.txt

# For Python 3.13+ (experimental)
pip install --break-system-packages -r requirements.txt

4. Download NLTK data (will be done automatically on first run, but you can do it manually):

python -m nltk.downloader punkt averaged_perceptron_tagger maxent_ne_chunker words

5. Set up your API keys:

# On Unix/macOS
export OPENAI_API_KEY='your-openai-api-key'
export ANTHROPIC_API_KEY='your-claude-api-key'

# On Windows
set OPENAI_API_KEY=your-openai-api-key
set ANTHROPIC_API_KEY=your-claude-api-key

If you encounter installation issues:

1. Try using Python 3.9, 3.10, or 3.11 for best compatibility
2. Install system dependencies:

# On Ubuntu/Debian
sudo apt-get install python3-dev ffmpeg

# On macOS
brew install ffmpeg

# On Windows
choco install ffmpeg

3. If you're using Python 3.13+, you might need to install from source:

pip install --no-binary :all: -r requirements.txt

• Python 3.8+
• FFmpeg installed on your system
• OpenAI API key (for GPT-4o-mini and Whisper)
• Anthropic API key (for Claude 3.5 Sonnet)
• Sufficient disk space for frame extraction

Basic usage:

python main.py movie.mp4

Advanced options:

# Specify number of experts
python main.py movie.mp4 --experts 15

# Skip audio analysis (faster, visual-only review)
python main.py movie.mp4 --no-audio

# Full example with all options
python main.py path/to/movie.mp4 --experts 12 --no-audio

• Video: .mp4, .mkv, .avi, .mov
• Audio: Automatically extracted from video

The system will:

1. Extract keyframes from your movie (2 frames per second)
2. Create 10x10 grid sheets of frames
3. Extract and transcribe audio using Whisper (unless --no-audio is specified)
4. Generate expert reviews using both GPT-4o-mini and Claude 3.5 Sonnet
5. Display detailed analysis and IMDB-style comments

The system creates the following directory structure:

moviename/
├── frames/         # Individual keyframes
├── sheets/        # 10x10 grid visualizations
└── audio/         # Extracted audio and transcriptions
    └── chunks/    # Audio chunks for processing

Each expert provides:

• Score out of 100 points
• Confidence interval
• IMDB-style comment
• Detailed review
• Category-by-category breakdown:
  • Visual Aesthetics (15 points)
  • Screenplay Quality (15 points)
  • Narrative Structure (15 points)
  • Technical Proficiency (10 points)
  • Innovation (10 points)
  • Cultural Impact (10 points)
  • Audience Appeal (10 points)

The final score is calculated by averaging expert ratings, with:

• Confidence intervals
• Genre bonuses
• Tier classification from "Critically Flawed" to "Timeless Masterpiece"

You can modify:

• frame_extractor.py: Adjust frame extraction rate
• expert_generator.py: Customize expert profiles
• rating_system.py: Modify rating categories and weights

• Processing time depends on movie length and number of experts
• API costs will vary based on usage:
  • GPT-4o-mini for visual analysis
  • Claude 3.5 Sonnet for screenplay analysis
  • Whisper for audio transcription
• Ensure sufficient disk space for frame extraction
• Consider API rate limits when processing multiple movies

Contributions are welcome! Please feel free to submit a Pull Request.