Stable Diffusion WebUI Podman

Run Stable Diffusion web browser accessible UI on your CUDA/ROCm machine without any hassle!

NOTE:

• Supports AUTOMATIC1111 and ComfyUI
• Supports CUDA and ROCm
• Supports SD 1.5 and SDXL for AUTOMATIC1111 and ComfyUI
• Supports SD 3 and 3.5 for ComfyUI
• Supports FLUX.1-dev GGUF for ComfyUI
• Requires podman-compose 1.1.0 or newer

Quick Setup

Download necessary models

podman-compose --profile download build
podman-compose --profile download up

Run AUTOMATIC1111 Stable Diffusion Web UI with CUDA support

podman-compose --profile auto-cuda build
podman-compose --profile auto-cuda up -d

You can change --profile auto-cuda to different profile:

Profile	Description
auto-cuda	AUTOMATIC1111 with CUDA support
auto-rocm	AUTOMATIC1111 with ROCm support
comfy-cuda	ComfyUI with CUDA support
comfy-rocm	ComfyUI with ROCm support

Set the container as systemd service to allow running on user login

Create AUTOMATIC1111 Stable Diffusion Web UI Podman service

Find the name of the running container

podman ps

Set the container as systemd service. E.g. for container named webui-podman_auto_1:

podman generate systemd --new webui-podman_auto_1 > ~/.config/systemd/user/automatic1111.service

Enable and run AUTOMATIC1111 Stable Diffusion Web UI Podman service

systemctl --user enable --now automatic1111.service

Stable Diffusion 3

Download models here: Stable Diffusion 3 on HuggingFace

Put SD3 models here:

/data/models/Stable-diffusion

Put SD3 text encoders here:

/data/models/CLIPEncoder

Stable Diffusion 3.5 GGUF

Download models here:

Stable Diffusion 3.5 large GGUF by city96 on HuggingFace

Stable Diffusion 3.5 medium GGUF by city96 on HuggingFace

Put SD3.5 GGUF models here:

/data/models/unet

Download SD3.5 VAE here: Stable Diffusion 3.5 VAE on HuggingFace

Put SD3.5 VAE here:

/data/models/VAE

FLUX.1 GGUF

By default, this will download the Q3_K_S quantization. But you can download different quantization for both FLUX.1-dev GGUF model and text encoder.

Download different quantization here:

FLUX.1-dev GGUF models

Text encoder

You will also need to download VAE manually because of HuggingFace login requirements.

Download VAE here: FLUX.1-dev VAE

Put FLUX.1 VAE here:

/data/models/VAE

Features

This repository provides multiple UIs for you to play around with stable diffusion:

AUTOMATIC1111

Full feature list here, Screenshots:

Text to image	Image to image	Extras

ComfyUI

Full feature list here, Screenshot:

Workflow

Contributing

Contributions are welcome! Create a discussion first of what the problem is and what you want to contribute (before you implement anything)

Disclaimer

The authors of this project are not responsible for any content generated using this interface.

This license of this software forbids you from sharing any content that violates any laws, produce any harm to a person, disseminate any personal information that would be meant for harm, spread misinformation and target vulnerable groups. For the full list of restrictions please read the license.

Thanks

Special thanks to everyone behind these awesome projects, without them, none of this would have been possible:

• AUTOMATIC1111/stable-diffusion-webui
• InvokeAI
• ComfyUI
• CompVis/stable-diffusion
• Sygil-webui
• AbdBarho/stable-diffusion-webui-docker
• city96/ComfyUI-GGUF
• city96 GGUF models
• zer0int/CLIP-GmP-ViT-L-14
• and many many more.