文档 /
教程 /
MiniMax (Hailuo) - AI video generation
MiniMax (Hailuo) - AI video generation
Generate high-quality video from text, images, and subject references using MiniMax's T2V-01, I2V-01, and S2V-01 models in ComfyUI
教程
overview 更新于 2026-10-25
MiniMax (Hailuo) - AI video generation
Generate high-quality video from text, images, and subject references using MiniMax's T2V-01, I2V-01, and S2V-01 models in ComfyUI
MiniMax (Hailuo) is a leading AI video generation platform renowned for exceptional motion quality and cinematic output, now inside ComfyUI. It offers two model lines — MiniMax-01 (T2V-01 text-to-video and I2V-01 image-to-video with Director/Live modes) and MiniMax S2V-01 (subject reference-to-video) — covering everything from film-grade narrative to character-driven content.
Key capabilities
- T2V-01 with Director mode — 15 cinematic camera movement types (pan, tilt, dolly, crane, tracking, aerial, handheld, roll, zoom-in, zoom-out, push-in, pull-out, orbit, first-person, static) for precise shot-level direction
- I2V-01-Director — Apply cinematic camera controls to image-initiated videos, combining the visual foundation of a starting image with the expressive language of Director mode
- I2V-01-Live — Generate natural, lively motion from a single image with organic movement dynamics — ideal for bringing portraits, landscapes, and product shots to life
- S2V-01 subject reference — Maintain character, face, or object identity across the generated video using a single reference image; perfect for serial content
- High motion quality — Industry-leading realism in character movement, physics, and fluid dynamics across all model variants
提示 使用 API 节点需要保证你已经正常登录,并在受许可的网络环境下使用,请参考API 节点总览部分文档来了解使用 API 节点的具体使用要求。
提示 本地用户 请确保你的 ComfyUI 已经更新。 ComfyUI 下载 ComfyUI 更新教程 本指南里的工作流可以在工作流模板中找到。如果找不到,可能是 ComfyUI 没有更新。 如果加载工作流时有节点缺失,可能原因有: 1. 你用的不是最新版(每夜版)。 2. 启动时有些节点导入失败。 云端用户 * Cloud 会在 ComfyUI 稳定版本发布后更新。 所以,如果你发现本文档中有任何核心节点缺失,可能是因为新核心节点尚未在最新稳定版中发布。请等待下一个稳定版发布。
Models
MiniMax-01 T2V-01 with Director mode and I2V-01 with Director/Live modes — the first-generation MiniMax architecture for text and image-based video generation.
MiniMax S2V-01 Subject reference-to-video for maintaining character, face, or object identity across generated clips.
MiniMax H3 Current flagship omni-modal video generation with native stereo audio (see full H3 docs).