如果有你幸访问本教程，可以看到文初使用 sd_xl_base_1.0.safetensors 模型生成的图片样张。参数如下：
1 2 3 4 5 6 7 prompt: a photo of a cat, images: 2 , seed: 1 , steps: 20 , cfgscale: 7 , variationseed: 1991156077 , aspectratio: 1 : 1 , width: 1024 , height: 1024 , initimagecreativity: 0.6 , refinercontrolpercentage: 0.2 , refinermethod: PostApply, refinerupscalemethod: pixel-bilinear, model: OfficialStableDiffusion/sd_xl_base_1.0 .safetensors, swarm_version: 0.6 .1 .0 , date: 2024 -04 -05 , generation_time: 2.13 (prep) and 18.11 (gen) seconds
友情提醒 ：如果你将 seed 参数（图像种子）值设置为 -1 则会生成随机图像。如果你想强化学习，可以参考官方文档参数设置，这是一个不错的选择。
获取 StableSwarmUI：StableSwarmUI ，你需要部署 Python 环境。如果你访问 Python 官网有困难，可以从 Python中文网 获取资源。当然也可以从国内各大开源软件镜像网站获取。如果实在有困难，文末我也会给出自己上传的 Windows 平台 Python 安装包。
我个人 Python 环境，仅供参考：
1 2 PS D:\work> python -V Python 3.10 .6
Windows 平台打开 powershell 使用 expand-archive StableSwarmUI-0.6.1-Beta：
1 Expand-Archive .\StableSwarmUI-0 .6.1 -Beta .zip -DestinationPath E:\AI\
进入 StableSwarmUI 根目录：
1 cd E:\AI\StableSwarmUI-0 .6.1 -Beta \
让我们一起愉快的开启 StableSwarmUI 体验之旅吧，在这之前，再次友情提醒安装好 Python 环境哟。
StableSwarmUI 安装与启动 启动 StableSwarmUI 之前我想介绍一下，个人硬件环境（笔电）。不算太新，不算太老，将将能玩（学习）：
CPU：R7-7735H（实际上是 R7-6800H的马甲U，苏妈挤了多少次牙膏了，升级命名鬼才）。8 个纯大核 u，平时也够用了。
显卡(GPU)：NVIDIA GeForce RTX 4060 Laptop GPU, 8.00 GiB
上了独立显卡，这种搭配，AMD 强劲的核显显得有点浪费了。
大学新生开学季，怎么套路换电脑？新思路，老爸啊，你儿子想学 Ai，需要高配一丢丢的 N 卡作为生产力。千万别说是我教的 ，问就是生产力需要。
有点跑题了，继续 StableSwarmUI 介绍， 目前处于 Beta 版本
A Modular Stable Diffusion Web-User-Interface, with an emphasis on making powertools easily accessible, high performance, and extensibility.
安装与启动，Windows 平台执行 launch-windows.bat 脚本，安装过程可能比较缓慢，也许你需要科学上网。
Windows 平台执行 launch-windows.bat 脚本，调用 dotnet 过程。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 PS E:\AI\StableSwarmUI-0 .6.1 -Beta > .\launch-windows .baterror: 指定的源已添加到可用程序包源的列表中。请提供唯一的源。 Usage: dotnet nuget add source [arguments ] [options ] Arguments: PackageSourcePath 包源的路径。 Options: -n |--name 源的名称。 -u |--username 连接到已验证源时要使用的用户名。 -p |--password 连接到已验证源时要使用的密码。 --store-password-in-clear-text 通过禁用密码加密来启用存储可移植包源凭据。 --valid-authentication-types 此源的有效身份验证类型的列表(用逗号分隔)。如果服务器公布 NTLM 或 Negotiate，并且你的凭据必 须使用基本机制发送(例如将 PAT 与本地 Azure DevOps Server 一起使用时)，请将此项设置为 basic。其他有效值包括 negotiate、kerberos、ntlm 和 digest，但这些值很可能没有用。 --protocol-version The NuGet server protocol version to be used. Currently supported versions are 2 and 3 . See https://learn.microsoft.com/nuget/api/overview for information about the version 3 protocol. Defaults to 2 if not specified. --configfile NuGet 配置文件。如果指定，将仅使用此文件中的设置。如果未指定，则将使用当前目录中的配置文件 的层次结构。有关详细信息，请参阅 https://docs.microsoft.com/nuget/consume-packages /configuring-nuget-behavior 。 -h |--help Show help information 适用于 .NET MSBuild 版本 17.9 .6 +a4ecab324 正在确定要还原的项目… 已还原 E:\AI\StableSwarmUI-0 .6.1 -Beta \src\StableSwarmUI.csproj (用时 2.75 sec)。 StableSwarmUI -> E:\AI\StableSwarmUI-0 .6.1 -Beta \src\bin\live_release\StableSwarmUI.dll 已成功生成。 0 个警告 0 个错误 已用时间 00 :00 :10.68
启动 StableSwarmUI 过程。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 17 :11 :09.194 [Init ] === StableSwarmUI v0.6.1 .0 Starting ===17 :11 :09.390 [Init ] Prepping extension: StableSwarmUI.Builtin_StabilityAPIExtension.StabilityAPIExtension...17 :11 :09.400 [Init ] Prepping extension: StableSwarmUI.Builtin_ScorersExtension.ScorersExtension...17 :11 :09.400 [Init ] Prepping extension: StableSwarmUI.Builtin_ImageBatchToolExtension.ImageBatchToolExtension...17 :11 :09.401 [Init ] Prepping extension: StableSwarmUI.Builtin_GridGeneratorExtension.GridGeneratorExtension...17 :11 :09.402 [Init ] Prepping extension: StableSwarmUI.Builtin_DynamicThresholding.DynamicThresholdingExtension...17 :11 :09.403 [Init ] Prepping extension: StableSwarmUI.Builtin_ComfyUIBackend.ComfyUIBackendExtension...17 :11 :09.404 [Init ] Prepping extension: StableSwarmUI.Builtin_AutoWebUIExtension.AutoWebUIBackendExtension...17 :11 :09.439 [Init ] Parsing command line...17 :11 :09.441 [Init ] Loading settings file...17 :11 :09.442 [Init ] No settings file found.17 :11 :09.443 [Init ] Re-saving settings file...17 :11 :09.462 [Init ] Applying command line settings...17 :11 :09.615 [Init ] Prepping options...17 :11 :09.895 [Init ] Loading models list...17 :11 :09.906 [Init ] Loading backends...17 :11 :09.908 [Init ] Loading backends from file...17 :11 :09.909 [Init ] Prepping API...17 :11 :09.910 [Init ] Backend request handler loop ready...17 :11 :09.912 [Init ] Prepping webserver...17 :11 :10.086 [Init ] Scan for web extensions...17 :11 :10.164 [Init ] Readying extensions for launch...17 :11 :10.165 [Init ] Launching server...17 :11 :10.166 [Init ] Starting webserver on http://localhost:7801 17 :11 :10.213 [Init ] Program is running.17 :11 :10.715 [Init ] Launch web browser to install page...17 :11 :13.249 [Info ] Creating new admin session 'local' for ::1 17 :11 :46.067 [Init ] [Installer ] Installation request received, processing...17 :11 :46.069 [Init ] [Installer ] Setting theme to dark_dreams.17 :11 :46.070 [Init ] [Installer ] Configuring settings as 'just yourself' install.17 :11 :46.070 [Init ] [Installer ] Downloading ComfyUI backend... please wait...18 :13 :01.394 [Init ] [Installer ] Downloaded! Extracting... (look in terminal window for details)7 -Zip (a) 23.01 (x86) : Copyright (c) 1999 -2023 Igor Pavlov : 2023 -06-20 Scanning the drive for archives: 1 file, 1446137955 bytes (1380 MiB)Extracting archive: dlbackend\comfyui_dl.7 z -- Path = dlbackend\comfyui_dl.7 z Type = 7 zPhysical Size = 1446137955 Headers Size = 326646 Method = LZMA:25 BCJ2 Solid = + Blocks = 2 Everything is Ok Folders: 2689 Files: 28279 Size: 5157388562 Compressed: 1446137955 18 :14 :49.725 [Init ] [Installer ] Installing prereqs...18 :15 :03.751 [Init ] [Installer ] Enabling ComfyUI...18 :15 :03.757 [Init ] [Installer ] Downloading model from 'https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_base_1.0.safetensors' ... please wait...18 :15 :03.758 [Init ] Initializing backend Cloning into 'sd-dynamic-thresholding' ... remote: Enumerating objects: 373 , done. remote: Counting objects: 100 % (116 /116 ), done. remote: Compressing objects: 100 % (23 /23 ), done. remote: Total 373 (delta 102 ), reused 102 (delta 93 ), pack-reused 257 Receiving objects: 84 % (314 /373 ), 620.00 KiB | 607.00 KRece Receiving objects: 100 % (373 /373 ), 1.00 MiB | 984.00 KiB/s, done. Resolving deltas: 100 % (186 /186 ), done. 18 :15 :13.209 [Info ] Saving backends...18 :15 :15.127 [Info ] Self-Start ComfyUI (Git Pull) exited properly.18 :15 :45.869 [Error ] Failed to download 'https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_base_1.0.safetensors' (HTTP): HttpRequestException: 由于连接方在一段时间后没有正确答复或连接的主机没有反应，连接尝试失败。 (huggingface.co:443 )18 :15 :45.913 [Init ] [Installer ] Model download complete.18 :15 :45.942 [Warning ] Failed to load metadata for OfficialStableDiffusion/sd_xl_base_1.0 .safetensors:System.IO.EndOfStreamException: Unable to read beyond the end of the stream. at System.IO.Stream.ReadAtLeastCore(Span`1 buffer, Int32 minimumBytes, Boolean throwOnEndOfStream) at System.IO.Stream.ReadExactly(Byte[] buffer, Int32 offset, Int32 count) at StableSwarmUI.Text2Image.T2IModel.GetSafetensorsHeaderFrom(String modelPath) in E:\AI\StableSwarmUI-0 .6.1 -Beta \src\Text2Image\T2IModel.cs:line 70 at StableSwarmUI.Text2Image.T2IModelHandler.LoadMetadata(T2IModel model) in E:\AI\StableSwarmUI-0 .6.1 -Beta \src\Text2Image\T2IModelHandler.cs:line 367 at StableSwarmUI.Text2Image.T2IModelHandler.<>c__DisplayClass22_0.<AddAllFromFolder>b__1(String file) in E:\AI\StableSwarmUI-0 .6.1 -Beta \src\Text2Image\T2IModelHandler.cs:line 509 18 :15 :45.951 [Info ] ComfyUI backend 0 shutting down...18 :15 :45.952 [Init ] [Installer ] Installed!18 :15 :48.159 [Info ] Creating new admin session 'local' for ::1 18 :16 :50.589 [Info ] Self-Start ComfyUI (Install rembg) exited properly.18 :16 :52.376 [Info ] Self-Start ComfyUI (Install opencv-python-headless ) exited properly.18 :17 :08.055 [Info ] Self-Start ComfyUI (Install imageio-ffmpeg ) exited properly.18 :17 :08.068 [Init ] Self-Start ComfyUI-0 on port 7821 is loading...18 :18 :22.518 [Init ] Self-Start ComfyUI-0 on port 7821 started.18 :18 :22.521 [Init ] Initializing backend 18 :18 :26.190 [Warning ] Failed to load metadata for OfficialStableDiffusion/sd_xl_base_1.0 .safetensors:System.IO.EndOfStreamException: Unable to read beyond the end of the stream. at System.IO.Stream.ReadAtLeastCore(Span`1 buffer, Int32 minimumBytes, Boolean throwOnEndOfStream) at System.IO.Stream.ReadExactly(Byte[] buffer, Int32 offset, Int32 count) at StableSwarmUI.Text2Image.T2IModel.GetSafetensorsHeaderFrom(String modelPath) in E:\AI\StableSwarmUI-0 .6.1 -Beta \src\Text2Image\T2IModel.cs:line 70 at StableSwarmUI.Text2Image.T2IModelHandler.LoadMetadata(T2IModel model) in E:\AI\StableSwarmUI-0 .6.1 -Beta \src\Text2Image\T2IModelHandler.cs:line 367 at StableSwarmUI.Text2Image.T2IModelHandler.<>c__DisplayClass22_0.<AddAllFromFolder>b__1(String file) in E:\AI\StableSwarmUI-0 .6.1 -Beta \src\Text2Image\T2IModelHandler.cs:line 509 18 :18 :33.285 [Info ] Self-Start ComfyUI (Git Pull) exited properly.fatal: unable to access 'https://github.com/mcmonkeyprojects/sd-dynamic-thresholding/' : Failure when receiving data from the peer 18 :18 :33.320 [Init ] Self-Start ComfyUI-0 on port 7822 is loading...18 :18 :41.465 [Init ] Self-Start ComfyUI-0 on port 7822 started.18 :19 :01.743 [Info ] Shutting down...18 :19 :01.814 [Info ] Backend request handler loop closing...18 :19 :01.815 [Info ] ComfyUI backend 0 shutting down...18 :19 :01.815 [Info ] Shutting down self-start ComfyUI (port=7822 ) process 18 :19 :01.866 [Info ] Self-Start ComfyUI-0 on port 7822 exited properly from disabling.18 :19 :01.928 [Info ] Backend handler shutdown complete without saving.18 :19 :01.929 [Info ] Will shut down session handler...18 :19 :01.929 [Info ] Will save user data .18 :19 :01.932 [Info ] Session handler is shut down.18 :19 :01.936 [Info ] All core shutdowns complete.18 :19 :02.960 [Info ] Self-Start ComfyUI-0 on port 7821 exited properly from disabling.
初次安装 ，从官方渠道获取比较缓慢，优先访问 https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_base_1.0.safetensors，看到报错或者警告信息是正常状态。在国内无法访问进去，可以前往第三方网站或者社区，比如魔搭社区获取他人上传的模型数据。
sd_xl_base_1.0 模型获取 魔搭社区，获取 sd_xl_base_1.0.safetensors
如果你使用 git 获取，默认你已经部署并配置好 Git 环境，可以直接获取完整模型训练仓库 git clone https://www.modelscope.cn/AI-ModelScope/stable-diffusion-xl-base-1.0.git。当然，你也可以点击文件 sd_xl_base_1.0.safetensors 直接下载。如果急需体验，个人推荐直接下载模型文件。完整模型仓库所占空间较大，存在版本库迭代，每个作者提交同步的内容有所不同。
如果本地没有模型数据 ，需要获取别人培养的模型数据，复制到 StableSwarmUI 指定的 \StableSwarmUI-0.6.1-Beta\Models\Stable-Diffusion\OfficialStableDiffusion 目录。
传入模型数据后，进入 web UI 界面点击模型，刷新即可看到模型。个人建议，最好还是在服务器功能模块选择服务器信息功能，点击关闭服务（Shutdown Server），然后再次执行 launch-windows.bat 启动 web 应用程序。
本地启动完成后，会自动跳转到默认设置浏览器，访问 http://localhost:7801
首页展示画面，初体验同时使用到了两种模型：
sd_xl_base_1.0.safetensors chilloutmix_NiPrunedFp32Fix.safetensors
使用 sd_xl_base_1.0.safetensors 模型生成一张猫咪图像（大图在文初）：
使用 chilloutmix_NiPrunedFp32Fix.safetensors 模型生成一张猫咪图像：
Windows 平台 Python 安装包（2.7.17、3.9.4、3.10.6、3.11.5）：
链接: https://pan.baidu.com/s/1uN8ceOnPVT3GASYMQSXUvw?pwd=8888
提取码: 8888
最后，以上总结仅供参考哟，希望对你的学习和工作有所帮助。
参考资料 ：
StableSwarmUI ComfyUI 魔搭社区 (modelscope.cn)
—END—
