What is open-director?

OpenDirector is an open source, self hosted AI video production studio. A person types a single sentence describing a video idea, and the system runs that idea through a pipeline of nine specialized AI agents that work together to produce a finished video complete with a script, storyboard, character designs, voiceover, background music, and rendered output. The whole system is designed to run through Docker, so getting started is meant to be as simple as running docker compose up after adding API keys to an environment file. The nine agent pipeline includes a Research Agent that can look up known stories, facts, and brand references, a Script Agent that writes the story outline, an Art Style Agent that picks from a library of built in visual styles, a Storyboard Agent that breaks the story into scenes and shots, a Character Agent that designs characters and assigns voices, a Location Agent that creates settings for each scene, a Voice Agent and BGM Agent that handle text to speech and music, and a Media Agent that pulls everything together into final assets, followed by a render worker that outputs the finished video. OpenDirector offers two main modes. Creative mode runs the full nine agent pipeline from one sentence to a complete video with adjustable aspect ratios and export resolutions. Batch mode is built for producing many short videos at once from a list of topics, with configurable clip length, subtitle styling, and background music, using royalty free stock footage or local video files. The project supports multiple AI providers for text generation, including OpenAI, Google Gemini, DeepSeek, Qwen, MiniMax, and locally run Ollama models, plus pluggable image and media generation providers. The application itself is built with Next.js, React, and TypeScript, and stores data using MySQL, Redis, and MinIO object storage, all wired together through Docker Compose. Both English and Chinese interfaces are supported.