seesee75-commits/ComfyUI-MiniMaxH3-Director

What it solves

MiniMax H3 is a powerful model that generates joint video and audio, but using it in ComfyUI typically requires manual, complex prompt writing and frame counting. This project provides a timeline-based editor that replaces manual storyboard prompting with a visual interface for managing shots, keyframes, and audio tracks.

How it works

The project introduces a set of nodes for ComfyUI that transform the generation process into a storyboard workflow. Users can drag and drop images, videos, and audio clips onto a timeline to define shots and timestamps. The system then automatically compiles these visual arrangements into the specific prompt format required by MiniMax H3, allowing users to see the exact prompt before rendering.

Who it’s for

Content creators and AI video artists using ComfyUI who want a more intuitive, professional editing-style workflow for MiniMax H3 instead of writing long, manual text prompts.

Highlights

• Timeline Editor: Visual management of shots, keyframes, and audio tracks with a ruler in seconds or frames.
• Automated Prompting: Automatically compiles a storyboard prompt from timeline segments and reference assets.
• Reference Management: Dedicated slots for subjects (people, scenes, props) and support for reference videos and audio.
• Audio Integration: Support for reference audio, muxed soundtracks, and specific voice clip assignments per subject.
• VLM Integration: Includes an "Analyze" button and an Enhance Prompt node that uses local vision models to describe reference images.
• Retake System: A dedicated node to splice regenerated ranges back into a base video.