ubuntu + conda 的 stable-diffusion-webui 安装笔记

前提：安装好 nvidia 显卡驱动。ubuntu 20.04, miniconda 环境

1. 下载仓库

# clone web ui and go into its directory
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
cd stable-diffusion-webui

2. 创建 conda 虚拟环境

conda create -n sd python=3.10

3. 安装所需的包

# install torch with CUDA support. See https://pytorch.org/get-started/locally/ for more instructions if this fails.  该命令会自动安装好 cuda 和 cudnn
conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia

安装后 测试 pytorch 是否安装好。

# -c cmd : program passed in as string (terminates option list)
python -c "import torch; print(torch.cuda.is_available())"

安装其它包：

# check if torch supports GPU; this must output "True". You need CUDA 11. installed for this. You might be able to use
# a different version, but this is what I tested.

# clone web ui and go into its directory
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
cd stable-diffusion-webui

# clone repositories for Stable Diffusion and (optionally) CodeFormer
mkdir repositories
git clone https://github.com/CompVis/stable-diffusion.git  stable-diffusion-stability-ai
git clone https://github.com/CompVis/taming-transformers.git repositories/taming-transformers
git clone https://github.com/sczhou/CodeFormer.git repositories/CodeFormer
git clone https://github.com/salesforce/BLIP.git repositories/BLIP

# install requirements of Stable Diffusion
pip install transformers==4.19.2 diffusers invisible-watermark --prefer-binary

# install k-diffusion
# pip install git+https://github.com/crowsonkb/k-diffusion.git --prefer-binary
# 估计是需要替换为：
# git clone https://github.com/crowsonkb/k-diffusion.git repositories/k-diffusion

# (optional) install GFPGAN (face restoration)
pip install git+https://github.com/TencentARC/GFPGAN.git --prefer-binary

# (optional) install requirements for CodeFormer (face restoration)
pip install -r repositories/CodeFormer/requirements.txt --prefer-binary

# install requirements of web ui
pip install -r requirements.txt  --prefer-binary

# update numpy to latest version
pip install -U numpy  --prefer-binary

下载 sd 1.5 模型（v1-5-pruned-emaonly.ckpt）放到：stable-diffusion-webui/models/Stable-diffusion/ 下。网址：https://huggingface.co/runwayml/stable-diffusion-v1-5

问题1：Warning: k_diffusion not found at path stable-diffusion-webui/repositories/k-diffusion/k_diffusion/sampling.py

解决：

https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/1272#issuecomment-1262244055

问题2：ImportError: cannot import name 'get_device' from 'basicsr.utils.misc'

解决：https://github.com/sczhou/CodeFormer/issues/193#issuecomment-1503410609

4. 启动程序

python webui.py 

上述命令会报以下问题：

问题3：No module 'xformers'. Proceeding without it.

需要使用如下命令才能使用 xformers：
# 参数列表见：modules/cmd_args.py
python webui.py --xformers

安装 xformers：

坑：不同版本的 torch 需要不同版本的 xformers。以上我们安装的是 2.0.1 版本的 torch，参考：https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/10494#issuecomment-1552353339 ，因此：

pip install xformers==0.0.20

否则你会在生成图像的时候会报如下错误：

NotImplementedError: No operator found for `memory_efficient_attention_forward` with inputs:
	 query       : shape=(2, 4096, 8, 40) (torch.float16)
	 key         : shape=(2, 4096, 8, 40) (torch.float16)
	 value       : shape=(2, 4096, 8, 40) (torch.float16)
	 attn_bias   : <class 'NoneType'>
	 p           : 0.0
`flshattF` is not supported because:
	xFormers wasn't build with CUDA support
	Operator wasn't built - see `python -m xformers.info` for more info
`tritonflashattF` is not supported because:
	xFormers wasn't build with CUDA support
	requires A100 GPU
`cutlassF` is not supported because:
	xFormers wasn't build with CUDA support
`smallkF` is not supported because:
	xFormers wasn't build with CUDA support
	dtype=torch.float16 (supported: {torch.float32})
	max(query.shape[-1] != value.shape[-1]) > 32
	unsupported embed per head: 40

最后我们就可以愉快的使用了：

python webui.py --xformers 

输入 pandas:

5. 一键启动

配置一个 sd_web_ui 脚本：

#!/bin/bash
source activate sd
cd /xx/stable-diffusion-webui
python webui.py --xformers 

然后我们就可以在命令行启动了：

./sd_web_ui 

参考：

【1】https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-NVidia-GPUs ：Manual Installation