HKUDS /ViMax

About ViMax

ViMax runs the whole pipeline, from story idea to finished shots. Hand it an idea, a screenplay, or a novel chapter. It plans characters, storyboards and shots, then generates the clips. Paid API keys are needed for the chat, image and video models.

Best for: Independent filmmakers who want a multi-shot AI video without running every step by hand.

• Turns a short idea into stories, characters, scripts and a final video
• Converts a written screenplay into multi-scene, multi-shot video
• Adapts long fiction into episodes and tracks characters across scenes
• Drops a person or pet from a photo into the story as a cameo
• Lets you plan and revise in a terminal or browser chat with the agent
• Generates compatible shots at the same time to save time

1. 1
   Install uv, then clone and sync: the command below, cd ViMax, uv sync

   git clone https://github.com/HKUDS/ViMax.git

   Copy
2. 2
   Copy the example config: the command below

   cp configs/agent.example.yaml configs/agent.local.yaml

   Copy
3. 3
   Add your LLM, image and video model, base_url and api_key in configs/agent.local.yaml (or set VIMAX_LLM_API_KEY, VIMAX_IMAGE_API_KEY, VIMAX_VIDEO_API_KEY)
4. 4
   Start the terminal UI: vimax tui, or vimax tui new / vimax tui resume
5. 5
   For the Web UI (Node.js 18+): cd web, npm install, npm run dev, then open http://127.0.0.1:4173
6. 6
   Or run a direct pipeline: set up configs/idea2video.yaml or configs/script2video.yaml, then run main_idea2video.py or main_script2video.py

• ViMax vs OpenMontage