Feedback
AI Ad Video Example
Loading...
comfyui minimax h3
ComfyUI MiniMax H3 brings open-weight generation to your own GPU — text, stills, or clips in, 2K video with matched stereo sound out.
All Tools
Discover our comprehensive AI-powered animation toolkit
Nano Pro
Advanced AI Image Generator
Veo3.1
Create Stunning Videos with Veo3.1
Kling 2.6
See the Sound, Hear the Visual.
Sora2 Video Generator
Advanced AI Video Generator for High-Quality Videos
Kling Motion Control
Turn reference images into amazing motion videos in minutes
Nano Banana Image Generator
Advanced AI Image Generator
Grok Image Generator
Create Sharp Visuals with AI
Runway Video Generator
Create Videos from Text or Images with AI
What Makes the comfyui minimax h3 Setup Different
Built on MiniMax's omni-modal model published as open weights, the comfyui minimax h3 setup runs entirely inside ComfyUI. Text, images, video, and audio are read together in one shared context, so dialogue, effects, and music arrive already matched to the picture. Renders reach roughly 15 seconds at up to 2K and 24fps, and every diffusion knob stays exposed at the node level.
- Stereo Sound, Generated In-PassDialogue, effects, and music are produced alongside the picture and muxed into a single MP4 by the comfyui minimax h3 graph — no separate audio pass, no dubbing step.
- Runs on Your Own HardwareBecause the comfyui minimax h3 weights are open, you decide resolution, clip length, and sampling parameters yourself — no queue, no metered API.
- Mix Text, Image, Video, and VoiceFeed several reference types into one comfyui minimax h3 run to hold a face, a look, a camera path, or a speaking voice steady across the whole clip.
Three Steps to Your First comfyui minimax h3 Clip
From a fresh ComfyUI install to an open-weight clip with matching audio — here is the shortest path through the comfyui minimax h3 workflow.
Core Capabilities of the comfyui minimax h3 Toolchain
Native templates, omni-modal inputs, in-pass stereo sound, reference locking, and optional Sage Attention acceleration — together the comfyui minimax h3 toolchain covers the whole local video pipeline, from prompt to finished file.
Three Ready-Made Templates
The comfyui minimax h3 template library includes text-to-video, image-to-video, and reference-to-video examples, so each generation mode works the moment you open it.
One Shared Multimodal Context
Text, stills, footage, and audio all land in the same context window inside the comfyui minimax h3 model, letting you combine reference types within a single run.
Lock Down Identity, Style, and Motion
Pull a character, a look, a movement, a camera path, or a voice from your references — the comfyui minimax h3 R2V node accepts up to 9 images, 3 videos, and 3 audio clips.
Clean On-Screen Text and Logos
Lettering and brand marks come out sharp with the comfyui minimax h3 model, and natural-language instructions can describe how each reference relates to the others.
Sage Attention for Faster Runs
Drop a Patch Sage Attention KJ node into the comfyui minimax h3 graph to roughly double throughput while giving up very little image quality.
Resolution and Duration Controls
The comfyui minimax h3 Resolution Selector derives width and height from your aspect ratio and megapixel target, snapped to the model's 32-pixel grid and 17-frame blocks at 24fps.
comfyui minimax h3: Common Questions
The questions people ask most about the comfyui minimax h3 setup, from install steps to output specifications.
What exactly is the comfyui minimax h3 workflow?
It is ComfyUI's built-in support for MiniMax H3, an omni-modal generation model that MiniMax published as open weights. One forward pass reads your text, images, video, and audio references and returns video with its own stereo soundtrack.
What output quality does it support?
Clips from the comfyui minimax h3 graph top out near 2K at 24fps and run about 15 seconds. The native canvas keeps a 768px short edge, maxes out at 768x1344, and rounds dimensions to a multiple of 32.
Which generation modes are included?
You get three sample graphs with the comfyui minimax h3 pack: text-to-video, image-to-video with optional first- and last-frame anchors, and reference-to-video, which pins a character, style, motion, camera angle, or voice.
Does it generate audio?
It does. Voice, sound effects, and music are all modeled by the comfyui minimax h3 model in the same pass as the picture, then delivered already synchronized inside one MP4.
How do I get started?
Install ComfyUI 0.30.0 or newer, open Template Library > Video, select a comfyui minimax h3 template, and accept the prompt to pull weights from the Comfy-Org/MiniMax-H3 repo on Hugging Face.
Can I speed up generation?
Add SageAttention plus the KJNodes pack, then place a Patch Sage Attention KJ node between UNETLoader and BasicGuider in the comfyui minimax h3 graph — throughput roughly doubles at little visible cost.
Put the comfyui minimax h3 Workflow to Work
Generate on your own GPU with open weights, synced stereo sound, and every parameter in reach. Text-to-video, image-to-video, and reference-to-video graphs are ready the moment you open the comfyui minimax h3 pack.