ComfyUI 使用 MiniMax H3 进行文生视频

📅 发布时间:2026/8/13 10:49:06
ComfyUI 使用 MiniMax H3 进行文生视频 ComfyUI 使用 MiniMax H3 进行文生视频环境准备硬件要求软件要求获取 MiniMax H3 模型文件方法一使用 ModelScope 下载国内推荐方法二使用 Hugging Face 下载备选步骤 1安装并升级 huggingface_hub步骤 2设置镜像源可选国内加速步骤 3获取 Hugging Face Access Token如果遇到 401 错误步骤 4下载所有模型文件备选如果 hf 命令不可用使用 Python 脚本ComfyUI 使用 MiniMax H3 进行文生视频参考由于本人水平有限难免出现错漏敬请批评改正。更多精彩内容可点击进入我的个人主页查看MiniMax H3 是一个强大的开源“全模态”模型能根据文字生成带有原生立体声音频的视频。环境准备硬件要求GPU至少16GB 显存的 NVIDIA 显卡推荐 RTX 4090 或更高。系统内存建议32GB 以上。磁盘空间模型文件约需40GB请确保有足够空间。软件要求ComfyUI需要v0.30.0 或更高版本。Python 环境确保 ComfyUI 的 Python 环境已正确安装推荐 Python 3.10。Git用于下载节点和模型可选。获取 MiniMax H3 模型文件你需要下载以下 5 个核心模型文件总大小约 40GB并放置到 ComfyUI 的models/目录下的对应子文件夹中。模型文件存放目录 (在ComfyUI/models/下)用途minimax_h3_fl2va_pruned_int8_convrot.safetensorsdiffusion_models/文生视频(T2V)和图生视频(I2V)主模型minimax_h3_ref2va_pruned_int8_convrot.safetensorsdiffusion_models/参考生视频(R2V)等qwen3vl_32b_minimax_h3_nvfp4_awq.safetensorstext_encoders/文本编码器minimax_h3_video_vae_fp16.safetensorsvae/视频 VAEminimax_h3_audio_vae_fp32.safetensorsvae/音频 VAE方法一使用 ModelScope 下载国内推荐ModelScope 国内访问速度快且该模型为公开模型无需认证。安装 ModelScope 命令行工具如果尚未安装pipinstallmodelscope一次性下载所有文件到 ComfyUI 的模型目录# 请将 /path/to/ComfyUI/models 替换为你的 ComfyUI 模型根目录的实际路径modelscope download--modelComfy-Org/MiniMax-H3\--local_dir/path/to/ComfyUI/models\vae/minimax_h3_audio_vae_fp32.safetensors\vae/minimax_h3_video_vae_fp16.safetensors\text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors\diffusion_models/minimax_h3_fl2va_pruned_int8_convrot.safetensors\diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors说明命令中的--local_dir会直接存放文件到指定目录且会保持vae/、text_encoders/等子目录结构与 ComfyUI 的模型目录结构完全匹配。验证下载检查对应子目录下是否已存在上述文件。方法二使用 Hugging Face 下载备选如果 ModelScope 无法使用或者你偏好从 Hugging Face 官方源下载也可以使用hf或huggingface-cli命令。注意Comfy-Org/MiniMax-H3是公开模型但有时 Hugging Face 可能要求登录才能下载返回 401 错误。你需要配置 Access Token。步骤 1安装并升级 huggingface_hubpipinstall-Uhuggingface_hub步骤 2设置镜像源可选国内加速如果你在中国大陆可以设置镜像源加速下载exportHF_ENDPOINThttps://hf-mirror.com步骤 3获取 Hugging Face Access Token如果遇到 401 错误登录 Hugging Face进入Settings → Access Tokens。生成一个Read权限的 token复制保存。在终端中设置环境变量exportHF_TOKENhf_xxxxxxxxxxxxxxxxxxxxx# 替换为你的真实 token步骤 4下载所有模型文件使用hf download命令注意hf是huggingface-cli的替代命令如果你的版本较旧可能还是huggingface-cli但推荐使用hf# 一次性下载所有文件到指定目录建议直接指定到 ComfyUI 的 models 根目录hf download Comfy-Org/MiniMax-H3\vae/minimax_h3_audio_vae_fp32.safetensors\vae/minimax_h3_video_vae_fp16.safetensors\text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors\diffusion_models/minimax_h3_fl2va_pruned_int8_convrot.safetensors\diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors\--local-dir /path/to/ComfyUI/models说明--local-dir会保留仓库中的目录结构即文件会下载到/path/to/ComfyUI/models/vae/、/path/to/ComfyUI/models/text_encoders/等与 ComfyUI 的模型目录结构一致。如果下载中断重新执行相同命令hf download会自动断点续传。如果你已设置HF_ENDPOINT镜像无需额外操作。备选如果hf命令不可用使用 Python 脚本fromhuggingface_hubimportsnapshot_download snapshot_download(repo_idComfy-Org/MiniMax-H3,local_dir/path/to/ComfyUI/models,local_dir_use_symlinksFalse,allow_patterns[*.safetensors],# 只下载 safetensors 文件resume_downloadTrue,)ComfyUI 使用 MiniMax H3 进行文生视频Realistic live-action cinematic look, sports event teaser: practical film photography style, golden-hour coastal stadium at dusk, anamorphic lens, shallow depth of field, film grain, sea breeze haze catching sunset light, restrained gradingfora premium feel, powerful natural movement. Scene overview: at dusk along waterfront athletic complex, athletes and citizens intertwineinrhythmic motion as the city awakens to the spirit of the2026Guangdong Provincial Games. This is the emotional prelude of an official teaser: every gesture honorslocalheritage and athletic aspiration, grounded and uplifting. Storyboard(each shot a separate scene, rapid cuts, all landing on the musical beats):[0s-1.5s]Shot1: wide drone-descending-to-eye-level: silhouette of a young sprinter exploding from starting blocks on the track, ocean glinting behind the stadium curves, wind lifting her hair and the corner of a red provincial games banner.[1s-2.5s]Shot2: medium tracking shot alongside: elderly fisherman on the adjacent pier pauses his net-mending to watch, weathered hands stilling as a para-swimmer glides through the open-water lane beside him, salt spray catching backlight.[2.5s-4s]Shot3: low-angle close-up: teenager’s worn running shoes pounding wet tarmac past lychee trees, fruit shadows dappled across her determined face, background blur showing children mimicking her stride with makeshift batons.[4s-5s]Shot4: freeze: the instant the sprinter’s hand touches the relay baton mid-transfer, fingers interlaced against the glowing stadium archway, golden light flaring softly, holdinginunified resolve. Camera: each shot its own angle, cuts clean and hard, no dissolves, gentle handheld intimacy during human moments, stabilized only on the final symbolic freeze. Audio: crashing waves, distant crowd murmur, rhythmic footfalls, traditional Cantonese percussion woven beneath modern score, a resonant gong strike on the baton pass, music cresting warmly at 4s, fading into ambient sea breathinthe last 1s. No text, subtitles, logos or watermarks of any kind, no animation or cartoon rendering, no overly-CG look, keep the live-action texture.参考[1] https://docs.comfy.org/tutorials/video/minimax/minimax-h3#minimax-h3-text-to-video-t2v