v0.1.3 - Memvid

🎉 v0.1.3 Release

🐳 Docker Support for Advanced Codecs

• Cross-platform H.265/HEVC encoding - No more codec dependency nightmares!
• Automated Docker container management for non-MP4 codecs
• Works seamlessly on Windows (WSL), macOS, and Linux
• Handles all FFmpeg operations in isolated environment

🤖 Multi-LLM Provider Support

• Added Google Gemini support - Use provider='google' in MemvidChat
• Added Anthropic Claude support - Use provider='anthropic' in MemvidChat
• New modular LLMClient class for easy provider management
• Consistent interface across all LLM providers

⚙️ Enhanced Configuration System

• Centralized configuration management via config.py
• Per-codec configuration profiles for optimal compression
• Flexible FFmpeg parameter customization
• Support for different video container formats (MP4, MKV, AVI)

✨ New Examples

codec_comparison.py

Compare different video codecs side-by-side:

• Test H.264, H.265, and MP4V compression ratios
• Benchmark encoding/decoding performance
• Find the optimal codec for your use case

file_chat.py

Enhanced document processing and chat:

• Process entire directories or specific files
• Configurable chunking parameters
• Support for PDF, EPUB, HTML, and text files
• Load and chat with existing memories
• Graceful FAISS index fallback for small datasets

🔧 Improvements

Better Error Handling

]- Improved error messages for missing dependencies

• Better handling of codec-specific issues

Configuration Flexibility

• Customizable chunk sizes and overlap
• Per-codec video parameters (CRF, preset, profile)
• Configurable frame rates and sizes

Package Structure

• Moved LLM providers to optional dependencies: pip install memvid[llm]
• Added EPUB support as optional: pip install memvid[epub]
• Core dependencies remain minimal

📦 Installation

# Basic installation
pip install memvid==0.1.3

API Keys

Set your API keys as environment variables:

export OPENAI_API_KEY="sk-..."
export GOOGLE_API_KEY="AIza..."
export ANTHROPIC_API_KEY="sk-ant-..."

🙏 Acknowledgments

Special thanks to our contributors who made this release possible with Docker support, codec testing, and multi-LLM integration!

@TyJK

---