环境:Ubuntu 22.04.1 LTS + NVIDIA GeForce RTX 3090
一、 安装显卡驱动
访问NVIDIA官方网站(https://developer.nvidia.com/cuda-downloads)获取对应安装命令
https://developer.nvidia.com/cuda-downloads
安装完成后执行nvidia-smi检测安装是否成功
二、 安装Miniconda3,安装Pytorch
在Index of /下载miniconda Linux系统安装包,运行安装包,按提示安装即可
安装Python环境miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-py310_24.5.0-0-Linux-x86_64.sh
sh Miniconda3-py310_24.5.0-0-Linux-x86_64.sh
安装Pytorch
# 换清华源
pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple/
pip install --pre torch torchvision torchaudio
三、下载ComfyUI,安装依赖
安装ComfyUI
git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI
pip install -r requirements.txt
大模型放置目录:ComfyUI/models/checkpoints/
插件放置目录:ComfyUI/models/checkpoints/
放大模型目录:ComfyUI/models/upscale_models
常用模型下载地址:
国内: LiblibAI-哩布哩布AI - 中国领先的AI创作平台
常用工作流下载地址:https://openart.ai/workflows/home
常用插件:
基础管理插件:https://github.com/ltdrdata/ComfyUI-Manager.git
翻译插件:https://github.com/AIGODLIKE/AIGODLIKE-ComfyUI-Translation.git
ControlNet插件:https://github.com/Kosinkadink/ComfyUI-Advanced-ControlNet.git
图像放大插件:https://github.com/ssitu/ComfyUI_UltimateSDUpscale.git
四、启动ComfyUI
# 配置下载模型走镜像站
export HF_ENDPOINT="https://hf-mirror.com"
python3 main.py --listen 0.0.0.0 --port 8188
标签:插件,git,ComfyUI,com,https,Linux,安装
From: https://blog.csdn.net/gnayji/article/details/140928763