scenty/stable-diffusion-webui-cpu repository overview

⁠Summary

Reduced version of stable diffusion, only CPU needed, the main model file realisticVisionV60B1_v51HyperVAE.safetensors（3.97GB）is already built-in, along with multiple extension plugins, such as ControlNet, ADetailer and Deforum, which can support prompt words in multiple languages and web UI in both English and Simplified Chinese.

精简版Stable Diffusion WebUI，仅使用CPU资源，已经内置了主模型文件realisticVisionV60B1_v51HyperVAE.safetensors（3.97GB），并预装了多款扩展插件，如ControlNet、ADetailer、Deforum等，支持中文提示词和简体中文界面（在‘Settings’里切换，步骤在文末），可以满足多数日常使用场景。

【注意：直接映射模型目录/stable-diffusion-webui/models到主机，内置模型文件会被清除】

如果想通过映射目录方式管理模型文件，可以先建一个非映射的容器，把模型文件拷贝出来，再重建一个带映射目录的容器并把该模型文件放进去，要比自己从网上找资源下载快得多。

⁠Based on and many thanks to:

https://github.com/automatic1111/stable-diffusion-webui⁠

https://github.com/abdbarho/stable-diffusion-webui-docker⁠

https://github.com/Physton/sd-webui-prompt-all-in-one⁠

https://github.com/dtlnor/stable-diffusion-webui-localization-zh_CN⁠

https://github.com/Coyote-A/ultimate-upscale-for-automatic1111⁠

https://github.com/Bing-su/adetailer⁠

https://github.com/Mikubill/sd-webui-controlnet⁠

https://github.com/deforum-art/deforum-for-automatic1111-webui.git⁠

⁠How to create container?

⁠1. Linux + Docker

docker run \
--publish 0.0.0.0:7860:7860 \
-v ~/sd-outputs:/stable-diffusion-webui/outputs \
--name sd-webui \
scenty/stable-diffusion-webui-cpu:latest

• How to access it?

  Open your browser and access http://your_host_ip:7860 or http://your_docker_ip:7860, based on your network settings.

⁠2. Docker Desktop

docker run `
--publish 0.0.0.0:7860:7860 `
-v C:\sd-outputs:/stable-diffusion-webui/outputs `
--name sd-webui `
scenty/stable-diffusion-webui-cpu:latest

• How to access it?

  Open your browser and access http://localhost:7860⁠ or http://your_host_ip:7860

⁠3. WSL2.0 (Windows Subsystem for Linux 2.0)

wslc run `
--publish 7860:7860 `
-v C:\sd-outputs:/stable-diffusion-webui/outputs `
--name sd-webui `
scenty/stable-diffusion-webui-cpu:latest

• How to access it?

  Open browser and access http://localhost:7860⁠

⁠How to manage model files easily?

The image contains a default model file (realisticVisionV60B1_v51HyperVAE.safetensors) to make it easier to start and work, you can also use a mapping volume to mount the models on your host, while the original model in the container would be removed.

Taking Docker Desktop as an example:

docker run `
--publish 7860:7860 `
-v C:\sd-models:/stable-diffusion-webui/models `
-v C:\sd-outputs:/stable-diffusion-webui/outputs `
--name sd-webui `
scenty/stable-diffusion-webui-cpu:latest

You can put your own model files into the corresponding directory on your host, if the mounting directory does not contain any model file, a default model v1-5-pruned-emaonly.safetensors would be downloaded automatically. This is the directory structure of models.

├ C:\sd-models
├── adetailer
├── Codeformer
├── ControlNet
├── GFPGAN
├── Lora
├── Stable-diffusion
├──── realisticVisionV60B1_v51HyperVAE.safetensors
├── VAE
├── VAE-approx
├── deepbooru
├── hypernetworks
├── karlo

⁠NOTE: If there are any updates to the model, you must reload it from Settings or restart the container.

⁠How to switch webpages language to Chinese?

Click Settings -> User Interface -> User interface -> Localization, choose option zh_CN -> Apply settings -> Reload UI

⁠History

• [2026-09-08] Tag=2.6.0, Tag=2.6.0-mlang, Tag=latest

  Added support for Deforum

• [2026-07-23] Tag=2.5.0, Tag=2.5.0-mlang, Tag=latest

  Added support for ControlNet

• [2026-07-17] Tag=2.3.0, Tag=2.3.0-mlang, Tag=latest

  Added support for ADetailer

• [2026-07-02] Tag=2.2.1, Tag=2.2.1-mlang, Tag=latest

  Fixed the issue of installing extensions through URLs

• [2026-07-01] Tag=2.1.1-mlang, Tag=latest

  Added support for Chinese web UI