nkxx188 /ComfyUI-MiniMaxH3-Easy

About ComfyUI-MiniMaxH3-Easy

Wire MiniMax H3 video generation into ComfyUI with a handful of nodes. ComfyUI is a node-based app for running image and video models. You get text-to-video, image-to-video, first/last-frame, reference and digital-human modes. Bring the official MiniMax H3 nodes and the H3 models on your own machine.

Best for: ComfyUI users who already run MiniMax H3 and want a small node setup for multi-shot clips.

• Runs text, image, first/last-frame, reference and digital-human modes from one node
• Manages images, video and audio in one Media Loader with drag, paste and reorder
• Type @ in a prompt to add an image, video or audio reference; # makes a dialogue block
• Chains shots with Context Segments, each with its own prompt, duration and media
• Refines each segment with Pixel Resize, 3D Latent Upscale or Low VRAM Tile
• Rewrites prompts with an optional optimizer, whole sequence or per segment

1. 1
   Update ComfyUI to a version that includes the official MiniMax H3 nodes.
2. 2
   Clone the repo into ComfyUI/custom_nodes: the command below, then restart ComfyUI. In ComfyUI Manager, pick the Nightly version.

   git clone https://github.com/nkxx188/ComfyUI-MiniMaxH3-Easy.git

   Copy
3. 3
   Put diffusion models, text encoders and VAEs in the matching ComfyUI/models folders. For latent upscale, add the H3 3D latent upscaler to ComfyUI/models/latent_upscale_models/.
4. 4
   Import workflow/1.MiniMax_H3_Easy.json, pick your models in MiniMax H3 Easy Loader, choose a mode, and enter a prompt.
5. 5
   Connect your media, choose resolution, aspect ratio and duration, then queue the workflow.