AIMixer/ComfyUI_MiniMaxH3_Director

What it solves

This project provides a comprehensive "director's console" for the MiniMax H3 video generation model within ComfyUI. It simplifies the creation of long-form videos by consolidating segment planning, condition encoding, sampling, decoding, and exporting into a single node, replacing the need for complex, fragmented workflows.

How it works

It acts as a high-level orchestrator for the official MiniMax H3 nodes. It supports multiple generation modes including text-to-video (t2v), image-to-video (i2v), first-and-last-frame-to-video (fl2v), reference-to-video (r2v), and video-to-video (v2v). The plugin integrates a visual timeline for segmenting videos, managing keyframes, and assigning prompts. It also implements advanced features like "Inter-segment Guidance" for smooth transitions between clips and a "Semantic Bridge" for better control over spatial and action logic.

Who it’s for

AI video creators and ComfyUI users who want professional-grade control over long-form video generation, multi-segment storytelling, and high-resolution refining using the MiniMax H3 model.

Highlights

• Multi-segment Timeline: Visual timeline for cutting, splitting, and intelligent scene detection (via PySceneDetect) to manage long videos.
• Advanced Generation Modes: Supports a wide array of tasks from simple text-to-video to complex reference-based video generation (r2v) with shared parameters.
• Inter-segment Guidance: Ensures motion and audio continuity by using the end of a previous segment to guide the next.
• Refine & Upscale: Integrated second-pass sampling for high-definition refining and latent upscaling.
• Director Packs: Ability to import and export project files (.mmxpack.zip) containing timelines, prompts, and assets.
• Native Stereo Audio: Generates synchronized stereo audio alongside the video frames.