vimax

ViMax is an open-source agentic video generation framework developed by HKUDS. It automates the end-to-end video creation pipeline, from narrative input to final video output, integrating the roles of director, screenwriter, producer, and video generator into a single workflow. Key features include: - Idea2Video: Converts a short concept into structured stories, characters, scripts, storyboards, shots, and a finished video. - Script2Video: Converts an explicit screenplay into controllable multi-scene, multi-shot video while preserving creative intent. - Novel2Video: Adapts long-form fiction into episodic visual narratives with narrative compression, character tracking, and scene planning. - AutoCameo: Places a person or pet from a reference photo into generated stories while maintaining consistent appearance. - Agent Loop and TUI: An interactive workspace for discussing ideas, revising plans, resuming sessions, reviewing artifacts, and controlling rendering. - Web UI: A browser-based interface for managing named projects, agent conversations, file uploads, artifact inspection, storyboard previews, and provider configuration. - Consistent production pipeline: Coordinates references, first frames, camera continuity, and final assembly end-to-end. - Parallelized generation: Generates compatible shots and media assets concurrently to accelerate multi-shot video production. The framework supports multiple model providers including OpenRouter, Google Omni, MiniMax, Seedance 2.0 Fast, and GPT Image 2. It is built with Python 3.12, uses uv for environment management, and requires Node.js 18+ for the Web UI. The project is MIT-licensed and runs on Linux and Windows.