ComfyUI-MiniMax-H3-LongMedia README

The project's README from github.com/vizart-vj/ComfyUI-MiniMax-H3-LongMedia. As one markdown file: README.md.

ComfyUI-MiniMax-H3-LongMedia

Production-oriented ComfyUI nodes for MiniMax H3 long-form video/audio generation, native reference editing, MultiClip planning, camera direction, fixed segmentation, lip-sync/redubbing, latent hi-res refinement, and adaptive low-VRAM execution.

screenshot

Current release: 0.5.40

Main Nodes

Legacy internal MiniMaxH3LatentLab... class identifiers remain registered for workflow compatibility.

Installation

Install into:

ComfyUI/custom_nodes/ComfyUI-MiniMax-H3-LongMedia

or install the package from the Comfy Registry.

Package identity:

GitHub:             vizart-vj/ComfyUI-MiniMax-H3-LongMedia
Comfy PublisherId:  noise

Restart ComfyUI after installation/update.

Current Setup Model

New workflows use independent semantic controls:

control_mode
h3_mode
timeline_mode
duration_source
audio_mode

This is the main change in how the project should be understood compared with the public 0.4.40 documentation.

See Operating Modes.

H3 Conditioning

t2va
fl2va
ref2va
hybrid
video_ref_edit

Timeline

single
segmented
multiclip

Duration Ownership

auto
video
audio
manual
longest_input

duration_source controls timeline length only. It does not remove audio references or change final-audio policy.

video_ref_edit

Typical source-character replacement:

video_1 = source video frames
image_1 = replacement identity
audio_1 = source soundtrack or new dub

A Video input is an IMAGE batch and never contains soundtrack data.

For preserve-style modes, Video1+Audio1 can be presented as a native paired source-performance reference while Audio1 also owns the target timing/output waveform.

For audio_mode=lip_sync, Audio1 is intentionally independent from Video1's original facial performance so completely new dialogue or singing can drive the replacement character.

Audio2/Audio3 remain prompt-addressable references for music, percussion, bass, ambience, or other semantic timing. In video_ref_edit, they are conditioning references only; Audio1 remains the sole preserved/passthrough source soundtrack.

See Audio Modes and video_ref_edit.

MultiClip + Cameras

Recommended connection:

Long Media Planner
        ↓ clip_plan
Long Media Cameras
        ↓ clip_plan
Long Media Setup

Use:

timeline_mode = multiclip

Planner owns diegetic scene/action prompts, durations, names, and optional seeds. Cameras owns framing, rig/lens, movement, speed, spatial relation, entity continuity, and transitions.

See:

Segmented Long Form

Use:

timeline_mode = segmented

for one continuous semantic movie split into fixed-duration internal units for VRAM/stability. It is not a storyboard scheduler.

See Fixed Segmentation Prompting.

Two-Stage H3 / Latent Hi-Res

The Long Media Sampler can:

  1. generate a low-resolution Stage-1 denoised x0;
  2. learned-upscale the video latent only;
  3. preserve the audio latent;
  4. rebuild target-grid conditioning;
  5. optionally run an independent same-seed fresh-noise high-resolution H3 pass.

Without Latent Hi-Res, the Refiner remains a continuous zero-noise low-sigma tail.

See Two-Stage Sampling, Latent Hi-Res and Refiner.

Sampler / Memory

Production starting point:

sampler_mode   = auto
memory_mode    = auto
attention_mode = auto

Keep ComfyUI Dynamic VRAM enabled.

Current memory-safety work includes exact Comfy Kitchen query streaming for structurally impossible fused-QKV workloads, repeat-run memory isolation, guarded native INT8 VBAR prefetch on constrained GPUs, and RAM-pressure-aware pinned host memory.

See Sampler, VRAM and Performance Guide.

FastH3 / FastVideo VSA

LongMedia includes isolated compatibility paths for supported H3ddle/PulpCut FastH3 VSA and Kijai FastVideo VSA packages.

These paths use strict structural detection and reset their runtime state when switching back to ordinary H3 checkpoints.

Loop Closure

Loop Closure returns the generated tail toward the opening macro-state in latent/H3 space. It is independent from conditioning/timeline mode and does not use an RGB crossfade as its primary mechanism.

Documentation

Start at docs/README.md.

Important guides:

Example Workflows

Release workflows contain neutral media placeholders rather than user media or local preview paths.

See workflows/README.md.

Third-Party Code

This repository contains adapted third-party components under their respective licenses, including:

See THIRD_PARTY_NOTICE.md and THIRD_PARTY_APACHE_2_0.txt.

Release History

0.5.40 is the release consolidation after public v0.4.40.

Historical release notes remain in docs/ for compatibility/reference. They may use the old workflow_mode terminology.

License

See LICENSE.

ComfyUI-MiniMax-H3-LongMedia (vizart-vj/ComfyUI-MiniMax-H3-LongMedia)

Searching Database…

294 ·GitHub
ComfyUI-MiniMax-H3-LongMedia
这是什么东西?一套面向生产环境的 ComfyUI 自定义节点,用于运行 MiniMax H3 进行长视频和音频生成:参考编辑、MultiClip 规划、镜头调度、固定分段、口型同步和重新配音、潜空间高分辨率精修,以及对低显存友好的安全运行。
它是一个独立应用,还是要装到 ComfyUI 里的东西?它是一个插件,不是独立应用:节点安装到现有的 ComfyUI 环境(custom_nodes 文件夹,或通过 Comfy Registry 安装包)后,会作为节点出现在你的工作流中。
怎么把它装进 ComfyUI?安装到 ComfyUI 的 custom_nodes 目录,或从 Comfy Registry 获取该包;安装或更新后需要重启 ComfyUI。
除了插件本身,我还需要准备什么?需要一个能正常运行的 ComfyUI 环境,并有一个可供采样的 H3 模型;README 把保持启用 ComfyUI 的 Dynamic VRAM 选项作为生产环境的起点。
以前的 0.4.40 教程还适用吗?部分适用:0.5.x 是在公开的 0.4.40 之后做的整合,新工作流采用独立的语义控制节点(Setup、Planner、Cameras、Sampler、Decode),而旧资料可能还在用 workflowmode 这套旧术语。
我需要显存很大的高端显卡吗?不需要——低配置显卡是明确支持的目标:针对装不下的 fused-QKV 负载提供查询流式处理、重复运行的内存隔离、带保护的 INT8 VBAR 预取,以及能感知内存压力的锁页主机内存。
用这些节点我实际能生成什么?带配乐的长视频:可以把一部连续的影片按固定时长的内部单元切分(以保证显存占用和稳定性),也可以做 MultiClip 序列——由 Planner 节点负责场景提示词、时长、命名和种子,由 Cameras 节点处理构图、镜头、运镜和转场。Loop Closure 会把生成的结尾拉回到开头的宏观状态。
角色替换的口型同步模式是怎么工作的?在 videorefedit 模式下,Video1 和 Audio1 可以作为一对配套的源表演参考,同时 Audio1 仍然负责目标时间轴和输出波形。在 lipsync 模式下,Audio1 刻意与原始面部表演解耦,因此全新的台词或歌声都可以驱动替换后的角色。
Latent Hi-Res 的两阶段采样是做什么的?Long Media Sampler 先对低分辨率的 Stage-1 x0 做去噪,在保留音频潜空间的同时对视频潜空间进行学习式上采样,重建目标网格的条件输入,并可选地以相同种子用全新噪声独立跑一次高分辨率采样;不开启的话,Refiner 只是一个零噪声、低 sigma 的收尾阶段。
有现成的工作流可以上手吗?有——发行版自带示例 JSON 工作流,比如 'MiniMax-H3-LongMedia-SAFE-1080p-15s.json'(生产环境的起点)和 'MiniMax-H3-LongMedia-LatentUpscale-Detailer.json',里面用的是中性的媒体占位符,而不是用户媒体或本地预览路径。
What are similar projects?ComfyUI-MiniMaxH3-TimelineDirector, ComfyUI-ALLinONE-MinimaxH3, ComfyUI_MiniMax_H3_Extender, FreeVideo, Comfyui-MMH3-UltimateUpscale, lanshu-create-ai-presenter-video, DiffSynth-Studio, LTX-2
Which topics is it in?lipsync, kv-cache, minimax, memory-management, upscaling, ai-video, diffusion, comfyui, audio-generation, video-editing, text-to-video, content-creation
Creator
vizart-vj
Size
6.9 MB
Stars
294
Updated
CommitSep 29Releasev0.6.63Sep 29
Age
2mo
Commits · 90 Days