简易教程：在本地2G显存和CPU环境下部署stable-diffusion-webui、NovelAI、Python和Git，以及CUDA的安装指南
最编程
2024-01-23 19:22:03
...
参考原文2：https://blog.****.net/yefufeng/article/details/127719952
环境准备
Python：3.10.7+
Git
NovelAI模型
CUDA（下载不高于现显卡版本的)
一、安装Python
官网：https://www.python.org
如下图操作，找到需要的版本，下载python安装包：
下载安装包
红框这一列往下拉，找到需要的版本的python（稳定版）（右边是预发布版）
如图，点击下载exe安装程序：
双击打开安装程序,勾选 Add Python 3.10 to PATH，然后点击上面红框安装。（自定义安装到d盘可能运行批处理程序会出问题）
选择安装路径到D盘后一直返回到最初安装页面（直接安装后面运行webui-user.bat文件可能会有问题）
验证安装
在 C:\Users\XXXX\AppData\Roaming下新建pip文件夹，在pip文件夹内新建pip.ini文件，并输入以下内容保存(pip使用清华镜像仓库)。
1 [global] 2 timeout = 60000 3 index-url = https://pypi.tuna.tsinghua.edu.cn/simple 4 5 [install] 6 use-mirrors = true 7 mirrors = https://pypi.tuna.tsinghua.edu.cn
二、安装CUDA
下载对应的安装包
桌面>>右键>>NAVIDIA控制面板 打开如下：
点击上图 系统信息，在打开的界面点击 组件，记住这个版本，下载的CUDA不要高于这个版本
CUDA下载地址：
https://developer.nvidia.com/cuda-toolkit-archive
点击运行程序，基本点击确定和点击下一步就行
三、安装Git
后面一直点击下一步就行
验证安装
四、安装dev-sidecar
内含NovelAI除ckpt和pt文件外的其他文件，ckpt文件太大，没超会上传不了。（自行百度搜索NovelAI模型的相关分享应该也是可以用的，或者可以直接用下面种子链接下载）
链接：https://pan.baidu.com/s/1UxYM2votxxFI0GnPhh9_zw 提取码：hte5
exe傻瓜式安装，安装进入程序按照提示安装证书即可。
五、拉取stable-diffusion-webui
打开dev-sidecar，如下图设置
在d盘新建一个文件夹，拉取项目放这
在空白处右键，选择 git bash here，如下输入拉取项目(复制后点击鼠标右键>>paste粘贴后回车)
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui
六、下载GFPGANv1.4.pth
下载慢的可以在上面百度云盘分享链接下
https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth
七、下载NovelAI模型并复制所需文件到相应位置
NovelAI（novelaileak（50g））模型，磁力链接：
magnet:?xt=urn:btih:5bde442da86265b670a3e5ea3163afad2c6f8ecc&dn=novelaileak
复制或移动下载好的NovelAI模型到项目文件夹 stable-diffusion-webui
1、复制GFPGANv1.4.pth 到 stable-diffusion-webui 根目录。
2、复制novelaileak\stableckpt\animefull-latest\model.ckpt 到 stable-diffusion-webui\models\Stable-diffusion目录下，并改名为novel-ai.ckpt。
3、复制novelaileak\stableckpt\animefull-latest\config.yaml 到 stable-diffusion-webui\models\Stable-diffusion目录下，并改名为novel-ai.yaml 。
4、复制novelaileak\stableckpt\animevae.pt 到 stable-diffusion-webui\models\Stable-diffusion目录下，并改名为novel-ai.vae.pt 。
5、复制novelaileak\stableckpt\modules\modules下的所有文件 到 stable-diffusion-webui\models\hypernetworks目录下，如果hypernetworks目录不存在，新建文件夹即可。
八、编辑webui-user.bat保存并运行
stable-diffusion-webui 根目录找到webui-user.bat文件，右键选中编辑
1 @echo off 2 3 set PYTHON=D:\Programs\Python\Python310\python.exe 4 set GIT= 5 set VENV_DIR= 6 set COMMANDLINE_ARGS=--ckpt .\models\Stable-diffusion\novel-ai.ckpt --lowvram --always-batch-cond-uncond --precision full --no-half --opt-split-attention-v1 --use-cpu sd --autolaunch 7 8 call webui.bat
在set PYTHON=这行代码后添加的是python安装路径，命令提示符下输入where python后回车可得到。
在 set COMMANDLINE_ARGS= 这行代码后：
2G 显存的增加 --lowvram ，再输入空格然后把下面这段加上
--always-batch-cond-uncond --precision full --no-half --opt-split-attention-v1 --use-cpu sd --autolaunch
4G 显存的增加 --medvram
保存后双击运行webui-user.bat，提示Running on local URL时成功，会自动打开浏览器（或手动打开127.0.0.1:7860），结果如下：
ps：#后面再运行不需要再开dev-sidecar
#使用cpu运行，生成图片需要5-10分钟，期间cpu会长时间拉满，电脑可能会有点卡
#请勿短时间内连续过多次使用，以免cpu长时间过热损坏cpu
#如遇到某些bug可以参考：https://blog.****.net/yefufeng/article/details/127719952
#备忘
1 (novelai) E:\workspace\02_Python\novalai\stable-diffusion-webui>python launch.py -h 2 Python 3.10.6 | packaged by conda-forge | (main, Oct 24 2022, 16:02:16) [MSC v.1916 64 bit (AMD64)] 3 Commit hash: b8f2dfed3c0085f1df359b9dc5b3841ddc2196f0 4 Installing requirements for Web UI 5 Launching Web UI with arguments: -h 6 usage: launch.py [-h] [--config CONFIG] [--ckpt CKPT] [--ckpt-dir CKPT_DIR] [--gfpgan-dir GFPGAN_DIR] 7 [--gfpgan-model GFPGAN_MODEL] [--no-half] [--no-half-vae] [--no-progressbar-hiding] 8 [--max-batch-count MAX_BATCH_COUNT] [--embeddings-dir EMBEDDINGS_DIR] 9 [--hypernetwork-dir HYPERNETWORK_DIR] [--localizations-dir LOCALIZATIONS_DIR] [--allow-code] 10 [--medvram] [--lowvram] [--lowram] [--always-batch-cond-uncond] [--unload-gfpgan] 11 [--precision {full,autocast}] [--share] [--ngrok NGROK] [--ngrok-region NGROK_REGION] 12 [--enable-insecure-extension-access] [--codeformer-models-path CODEFORMER_MODELS_PATH] 13 [--gfpgan-models-path GFPGAN_MODELS_PATH] [--esrgan-models-path ESRGAN_MODELS_PATH] 14 [--bsrgan-models-path BSRGAN_MODELS_PATH] [--realesrgan-models-path REALESRGAN_MODELS_PATH] 15 [--scunet-models-path SCUNET_MODELS_PATH] [--swinir-models-path SWINIR_MODELS_PATH] 16 [--ldsr-models-path LDSR_MODELS_PATH] [--clip-models-path CLIP_MODELS_PATH] [--xformers] 17 [--force-enable-xformers] [--deepdanbooru] [--opt-split-attention] [--opt-split-attention-invokeai] 18 [--opt-split-attention-v1] [--disable-opt-split-attention] 19 [--use-cpu {all,sd,interrogate,gfpgan,swinir,esrgan,scunet,codeformer} [{all,sd,interrogate,gfpgan,swinir,esrgan,scunet,codeformer} ...]] 20 [--listen] [--port PORT] [--show-negative-prompt] [--ui-config-file UI_CONFIG_FILE] 21 [--hide-ui-dir-config] [--freeze-settings] [--ui-settings-file UI_SETTINGS_FILE] [--gradio-debug] 22 [--gradio-auth GRADIO_AUTH] [--gradio-img2img-tool {color-sketch,editor}] [--opt-channelslast] 23 [--styles-file STYLES_FILE] [--autolaunch] [--theme THEME] [--use-textbox-seed] 24 [--disable-console-progressbars] [--enable-console-prompts] [--vae-path VAE_PATH] 25 [--disable-safe-unpickle] [--api] [--nowebui] [--ui-debug-mode] [--device-id DEVICE_ID] 26 [--administrator] [--cors-allow-origins CORS_ALLOW_ORIGINS] [--tls-keyfile TLS_KEYFILE] 27 [--tls-certfile TLS_CERTFILE] [--server-name SERVER_NAME] 28 29 options: 30 -h, --help show this help message and exit 31 --config CONFIG path to config which constructs model 32 --ckpt CKPT path to checkpoint of stable diffusion model; if specified, this checkpoint will be added to 33 the list of checkpoints and loaded 34 --ckpt-dir CKPT_DIR Path to directory with stable diffusion checkpoints 35 --gfpgan-dir GFPGAN_DIR 36 GFPGAN directory 37 --gfpgan-model GFPGAN_MODEL 38 GFPGAN model file name 39 --no-half do not switch the model to 16-bit floats 40 --no-half-vae do not switch the VAE model to 16-bit floats 41 --no-progressbar-hiding 42 do not hide progressbar in gradio UI (we hide it because it slows down ML if you have hardware 43 acceleration in browser) 44 --max-batch-count MAX_BATCH_COUNT 45 maximum batch count value for the UI 46 --embeddings-dir EMBEDDINGS_DIR 47 embeddings directory for textual inversion (default: embeddings) 48 --hypernetwork-dir HYPERNETWORK_DIR 49 hypernetwork directory 50 --localizations-dir LOCALIZATIONS_DIR 51 localizations directory 52 --allow-code allow custom script execution from webui 53 --medvram enable stable diffusion model optimizations for sacrificing a little speed for low VRM usage 54 --lowvram enable stable diffusion model optimizations for sacrificing a lot of speed for very low VRM 55 usage 56 --lowram load stable diffusion checkpoint weights to VRAM instead of RAM 57 --always-batch-cond-uncond 58 disables cond/uncond batching that is enabled to save memory with --medvram or --lowvram 59 --unload-gfpgan does not do anything. 60 --precision {full,autocast} 61 evaluate at this precision 62 --share use share=True for gradio and make the UI accessible through their site 63 --ngrok NGROK ngrok authtoken, alternative to gradio --share 64 --ngrok-region NGROK_REGION 65 The region in which ngrok should start. 66 --enable-insecure-extension-access 67 enable extensions tab regardless of other options 68 --codeformer-models-path CODEFORMER_MODELS_PATH 69 Path to directory with codeformer model file(s). 70 --gfpgan-models-path GFPGAN_MODELS_PATH 71 Path to directory with GFPGAN model file(s). 72 --esrgan-models-path ESRGAN_MODELS_PATH 73 Path to directory with ESRGAN model file(s). 74 --bsrgan-models-path BSRGAN_MODELS_PATH 75 Path to directory with BSRGAN model file(s). 76 --realesrgan-models-path REALESRGAN_MODELS_PATH 77 Path to directory with RealESRGAN model file(s). 78 --scunet-models-path SCUNET_MODELS_PATH 79 Path to directory with ScuNET model file(s). 80 --swinir-models-path SWINIR_MODELS_PATH 81 Path to directory with SwinIR model file(s). 82 --ldsr-models-path LDSR_MODELS_PATH 83 Path to directory with LDSR model file(s). 84 --clip-models-path CLIP_MODELS_PATH 85 Path to directory with CLIP model file(s). 86 --xformers enable xformers for cross attention layers 87 --force-enable-xformers 88 enable xformers for cross attention layers regardless of whether the checking code thinks you 89 can run it; do not make bug reports if this fails to work 90 --deepdanbooru enable deepdanbooru interrogator 91 --opt-split-attention 92 force-enables Doggettx's cross-attention layer optimization. By default, it's on for torch 93 cuda. 94 --opt-split-attention-invokeai 95 force-enables InvokeAI's cross-attention layer optimization. By default, it's on when cuda is 96 unavailable. 97 --opt-split-attention-v1 98 enable older version of split attention optimization that does not consume all the VRAM it can 99 find 100 --disable-opt-split-attention 101 force-disables cross-attention layer optimization 102 --use-cpu {all,sd,interrogate,gfpgan,swinir,esrgan,scunet,codeformer} [{all,sd,interrogate,gfpgan,swinir,esrgan,scunet,codeformer} ...] 103 use CPU as torch device for specified modules 104 --listen launch gradio with 0.0.0.0 as server name, allowing to respond to network requests 105 --port PORT launch gradio with given server port, you need root/admin rights for ports < 1024, defaults to 106 7860 if available 107 --show-negative-prompt 108 does not do anything 109 --ui-config-file UI_CONFIG_FILE 110 filename to use for ui configuration 111 --hide-ui-dir-config hide directory configuration from webui 112 --freeze-settings disable editing settings 113 --ui-settings-file UI_SETTINGS_FILE 114 filename to use for ui settings 115 --gradio-debug launch gradio with --debug option 116 --gradio-auth GRADIO_AUTH 117 set gradio authentication like "username:password"; or comma-delimit multiple like 118 "u1:p1,u2:p2,u3:p3" 119 --gradio-img2img-tool {color-sketch,editor} 120 gradio image uploader tool: can be either editor for ctopping, or color-sketch for drawing 121 --opt-channelslast change memory type for stable diffusion to channels last 122 --styles-file STYLES_FILE 123 filename to use for styles 124 --autolaunch open the webui URL in the system's default browser upon launch 125 --theme THEME launches the UI with light or dark theme 126 --use-textbox-seed use textbox for seeds in UI (no up/down, but possible to input long seeds) 127 --disable-console-progressbars 128 do not output progressbars to console 129 --enable-console-prompts 130 print prompts to console when generating with txt2img and img2img 131 --vae-path VAE_PATH Path to Variational Autoencoders model 132 --disable-safe-unpickle 133 disable checking pytorch models for malicious code 134 --api use api=True to launch the api with the webui 135 --nowebui use api=True to launch the api instead of the webui 136 --ui-debug-mode Don't load model to quickly launch UI 137 --device-id DEVICE_ID 138 Select the default CUDA device to use (export CUDA_VISIBLE_DEVICES=0,1,etc might be needed 139 before) 140 --administrator Administrator rights 141 --cors-allow-origins CORS_ALLOW_ORIGINS 142 Allowed CORS origins 143 --tls-keyfile TLS_KEYFILE 144 Partially enables TLS, requires --tls-certfile to fully function 145 --tls-certfile TLS_CERTFILE 146 Partially enables TLS, requires --tls-keyfile to fully function 147 --server-name SERVER_NAME 148 Sets hostname of server
原文地址：https://www.cnblogs.com/zhongbenbayun/p/16879362.html
上一篇： 3个方法阻止BT下载