【本地化部署Stable Diffusion WebUI（MACOS安装）】

提示：文章写完后，目录可以自动生成，如何生成可参考右边的帮助文档

---

前言

随着人工智能的不断发展，AI绘图热兴起，我也想加入到AI绘图大潮中。

---

一、Stable Diffusion WebUI适配三类芯片

此模型支持三类GPU芯片

1. NVidia (recommended)
2. AMD GPUs.
3. Intel CPUs, Intel GPUs (bothintegrated and discrete) （我的mac属于此情况）

二、安装

说明：期间需要访问外网，自行实现巧妙上网

1.安装git、conda等

brew install cmake protobuf rust python@3.10 git wget miniconda

2.选定目录及下载Github远程仓库文件

mkdir SDW
cd SDW
git clone git@github.com:openvinotoolkit/stable-diffusion-webui.git

3.创建conda虚拟环境

创建虚拟环境

conda env create -p /Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/automatic -f /Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/environment-wsl2.yaml

最后创建完成后显示如下：

# To activate this environment, use                                                                                                                                                                                                 
#                                                                                                                                                                                                                                   
#     $ conda activate /Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/automatic                                                                                                                                                
#                                                                                                                                                                                                                                   
# To deactivate an active environment, use                                                                                                                                                                                          
#                                                                                                                                                                                                                                   
#     $ conda deactivate

激活环境

conda activate /Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/automatic

4.安装依赖

cd stable-diffusion-webui
pip install -r requirements.txt

pip速度慢需添加国内pip源

cd ~
mkdir .pip

cd .pip
touch pip.conf

在pip.conf文件中，添加以下内容：

[global]
index-url = https://pypi.doubanio.com/simple/
trusted-host = pypi.doubanio.com

5.运行

还有未安装的依赖在此步骤执行中会一次安装，当然有报错需要一次解决。

英特尔CPU+GPU的场景执行前环境要生效如下内容：

export PYTORCH_TRACING_MODE=TORCHFX
export COMMANDLINE_ARGS="--skip-torch-cuda-test --precision full --no-half" 
./webui.sh

6.问题处理

运行./webui.sh会有很多报错

缺openai模块等

OSError: Can't load tokenizer for 'openai/clip-vit-large-patch14'. If you were trying to load it from 'https://huggingface.co/models', make sure you don't have a local directory with the same name. Otherwise, make sure 'openai/clip-vit-large-patch14' is the correct path to a directory containing all relevant files for a CLIPTokenizer tokenizer.

手动git clone

可直接git clone
可以指定git代理实现

#指定git代理
git config --global https.proxy socks5://127.0.0.1:1088
或
git config --global https.proxy https://127.0.0.1:1088
#取消git代理
git config --global --unset https.proxy

例如：
git clone https://huggingface.co/openai/clip-vit-large-patch14
或
git clone git@hf.co:openai/clip-vit-large-patch14

类似的v1-5-pruned-emaonly.safetensors模型获取失败

https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned-emaonly.safetensors

此文件较大有4G多，建议通过从源网站下载到云盘，再从云盘下载到本地，然后让入目标文件夹如：stable-diffusion-uebul/models/Stable-diffusion/

RuntimeError: “LayerNormKernelImpl” not implemented for 'Half’等

RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'
/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/torch/nn/functional.py", line 2515, in layer_norm

RuntimeError: "addmm_impl_cpu_" not implemented for 'Half'
/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/torch/nn/modules/linear.py", line 114, in forward

那么执行时添加如下参数：

./webui.sh --precision full --no-half 

正常运行sdw的的话会自动访问http://127.0.0.1:7860/

命令行显示类似如下：

浏览器显示如下：

注意选择支持intel GPU的script，OpenVINO™ Acceleration Script

绘图时报错

类似如下报错
报错显示如下：

OSError: We couldn't connect to 'https://huggingface.co' to load this file, couldn't find it in the cached files and it looks like CompVis/stable-diffusion-safety-checker is not the path to a directory containing a file named config.json. Checkout your internet connection or see how to run the library in offline mode at 'https://huggingface.co/docs/transformers/installation#offline-mode'.

以及

   Traceback (most recent call last):
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connection.py", line 198, in _new_conn
        sock = connection.create_connection(
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/util/connection.py", line 85, in create_connection
        raise err
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/util/connection.py", line 73, in create_connection
        sock.connect(sa)
    TimeoutError: timed out

    The above exception was the direct cause of the following exception:

    Traceback (most recent call last):
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connectionpool.py", line 793, in urlopen
        response = self._make_request(
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connectionpool.py", line 491, in _make_request
        raise new_e
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connectionpool.py", line 467, in _make_request
        self._validate_conn(conn)
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connectionpool.py", line 1099, in _validate_conn
        conn.connect()
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connection.py", line 616, in connect
        self.sock = sock = self._new_conn()
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connection.py", line 207, in _new_conn
        raise ConnectTimeoutError(
    urllib3.exceptions.ConnectTimeoutError: (<urllib3.connection.HTTPSConnection object at 0x7ff2fbebe290>, 'Connection to huggingface.co timed out. (connect timeout=10)')

    The above exception was the direct cause of the following exception:

    Traceback (most recent call last):
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/requests/adapters.py", line 486, in send
        resp = conn.urlopen(
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/connectionpool.py", line 847, in urlopen
        retries = retries.increment(
      File "/Users/zhaoyao/Downloads/SDW/stable-diffusion-webui/venv/lib/python3.10/site-packages/urllib3/util/retry.py", line 515, in increment
        raise MaxRetryError(_pool, url, reason) from reason  # type: ignore[arg-type]
    urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='huggingface.co', port=443): Max retries exceeded with url: /CompVis/stable-diffusion-safety-checker/resolve/main/config.json (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7ff2fbebe290>, 'Connection to huggingface.co timed out. (connect timeout=10)'))

很显然是绘图时需要访问huggingface.co下载模块

考虑直接添加代理的方式，因为不可能每次画图每次需要下载模块的时候每次手动下载

修改webu.sh

在此文件while [[ “$KEEP_GOING” -eq “1” ]]; 上一行添加一行如下内容：

export HF_ENDPOINT=https://hf-mirror.com

绘图时会有如下显示：

绘图结束显示

---

总结

参考于https://github.com/openvinotoolkit/stable-diffusion-webui
https://openai.wiki/stable-diffusion-webui-localization.html。