MiniMax H3 Infinite Video: Make 2-Hour Movies in ComfyUI (0-Shot)

Full tutorial link > https://www.youtube.com/watch?v=1580ZDX-60Q

• Turn one folder of scene prompts into a long, coherent AI video with MiniMax H3 - locally, 0-shot and without babysitting every clip. This ComfyUI walkthrough shows how to match references, queue scenes, generate clips and automatically merge everything into one movie.

• The opening is the raw workflow result. Then we rebuild it from installation to playback: models, presets, VRAM modes, prompt creation, folder batching, reference syntax, draft settings, troubleshooting, selective regeneration and consistency. It can scale to very long projects, including the 2-hour movie shown here.

COMFYUI INSTALLERS + PRESETS:

• https://www.patreon.com/posts/comfyui-installers-presets-105023709

SWARMUI MODEL DOWNLOADER:

• https://www.patreon.com/SECourses/posts/model-downloader-114517862

DISCORD HELP & SUPPORT:

• https://discord.com/invite/software-engineering-courses-secourses-772774097734074388

Zip File Content Screenshots

Some Other Screenshots From Workflows

Ultra advanced multiple reference upload and mention system

Optional Low VRAM optimization with just a toggle enable disable

Trim and upload references feature

Automatically download necessary files ultra fast and SHA256 verified

Sana Labs published 4x speed up implementation

Video Chapters

• [00:00:00] 0-shot generated movie showcase

• [00:00:58] MiniMax H3 local workflow reveal

• [00:01:10] Entire movies, prompts and scenes automated

• [00:01:22] Audio references and folder-batch strategy

• [00:01:40] Local desktop vs cloud deployment

• [00:01:50] Download ComfyUI and the model downloader

• [00:02:01] Fresh install and recommended Python 3.12

• [00:02:17] Verify installation and install bundle 100

• [00:02:37] Troubleshooting and support

• [00:02:47] Final checks before launch

• [00:02:57] Launch ComfyUI with run_gpu.bat

• [00:03:11] 6 GB GPU support and speed tradeoffs

• [00:03:22] Logs, errors and starting the preset

• [00:03:33] Find the MiniMax H3 presets

• [00:03:44] References-to-video 4x workflow

• [00:03:54] Compilation, 20 steps and speed LoRA

• [00:04:12] Change logs and Windows requirements

• [00:04:24] Frequent workflow and custom-node updates

• [00:04:36] Missing models and downloader setup

• [00:04:48] Launch downloader and share model paths

• [00:05:03] Core vs low-VRAM MiniMax H3 bundles

• [00:05:16] INT4 vs recommended INT8 quality

• [00:05:29] Options for 6-12 GB GPUs

• [00:05:43] Select the correct models path

• [00:05:55] Folder structure and model download

• [00:06:07] Fix path warnings and select both VAEs

• [00:06:23] Reference manager and default interface

• [00:06:35] The enhanced prompt helper file

• [00:06:47] Improve prompts with any major LLM

• [00:06:59] ChatGPT setup with five voice attachments

• [00:07:17] Example prompts and downloadable ZIP

• [00:07:28] Prompt construction and easy referencing

• [00:07:42] Upload the audio references

• [00:07:52] Reference syntax and voice samples

• [00:08:08] Listen to the supplied voice references

• [00:08:18] Audio and image reference options

• [00:08:28] Reorder image references by drag and drop

• [00:08:38] Batch folder mode needs no manual prompt

• [00:08:50] Set the folder path and draft resolution

• [00:09:01] Aspect ratios and divisible dimensions

• [00:09:15] Recommended 1344x768 for 16:9

• [00:09:25] 30-second limit and 15-second sweet spot

• [00:09:35] Best parameters: ready to run

• [00:09:47] Run a quick 0.4 MP draft

• [00:09:57] Automatic merging and queue monitoring

• [00:10:08] Folder prompts and reference matching

• [00:10:20] Load many references; use only the matches

• [00:10:30] Per-generation reference limits

• [00:10:40] Batch power for full videos and animations

• [00:10:50] Iterate prompts with your favorite GPT

• [00:11:00] Draft low resolution, review, then refine

• [00:11:13] The key file for better prompts

• [00:11:23] Single-clip mode and included presets

• [00:11:33] Image, text and references-to-video presets

• [00:11:46] Lightricks speed-up LoRA implementation

• [00:11:59] LoRA version notes and future updates

• [00:12:17] SwarmUI support and future advanced tutorial

• [00:12:28] Read the docs and enable low-VRAM mode

• [00:12:38] Save 15-40%+ VRAM

• [00:12:54] Built-in node help and first output

• [00:13:08] Regenerate only a weak section

• [00:13:26] Play the generated result

• [00:13:36] Current audio-reference limitation

• [00:13:46] Improve voice and character consistency

• [00:13:56] Match image IDs to characters

• [00:14:06] Included reference and consistency guide

• [00:14:16] 0-shot proof: ChatGPT authored all prompts, the movie script and rough draft

• [00:14:35] Automated setup and broad GPU support

• [00:14:45] Final requirements reminder and outro

• MiniMax H3 supports 30-second segments; 15 seconds is the sweet spot. This demo uses 10-second scenes for easy review and regeneration.

• For 16:9, use 1344x768. Draft around 0.4 MP, review the sequence, improve prompts, then render final quality.

• Load up to 99 references; each prompt uses only matched IDs. Per generation: up to 3 audio, 3 video and 9 image references. Extra audio/video refs are discarded.

• Use INT8 for speed and near-BF16 quality; INT4 is for lower VRAM. It runs on 6 GB GPUs, but more slowly. Low-VRAM mode saves about 15%, reaching 40%+ in some cases.

• Audio references are less reliable than image references. If voice quality drops, try no audio reference. Map image IDs clearly for consistency.

• Covers local Windows, Massed Compute, RunPod and SimplePod. Use a fresh install with Python 3.12 where recommended. Read requirements and changelogs; nodes and workflows update often.

• If one scene fails, regenerate only that section. Join the Discord above for setup help.

#MiniMaxH3 #ComfyUI #AIVideo #LocalAI #GenerativeAI

MiniMax H3 Infinite Video: Make 2-Hour Movies in ComfyUI (0-Shot) (YouTube)