System and Hardware Requirements

1. x86_64/amd64 architecture
2. Ubuntu 22.04 Server Installed.
3. Nvidia Graphics Card (tested on 1060 6GB VRam).
4. Recommended is more than 200 GB available for your docker volumes.

Prerequisites

1. Install Docker on your host machine: https://docs.docker.com/engine/install/ubuntu/.
2. Install CUDA and Nvidia drivers on your host machine: https://www.cherryservers.com/blog/install-cuda-ubuntu.
   1. Tested and optimized for CUDA 11.8.
   2. Tested with Nvidia Drivers version 545.
3. Clone this repository.

Installation guide

AUTOMATIC1111

https://github.com/AUTOMATIC1111/stable-diffusion-webui

1. Go to automatic1111 subdirectory: cd automatic1111.
2. Run docker compose build automatic1111.v1.6.
3. Create a volume (100 GB free space is recommended):
   • if you want to place your data in a dedicated location: docker volume create --driver local --opt type=none --opt device=[YOUR_AUTOMATIC1111_DATA_LOCATION] --opt o=bind automatic1111_data
   • or use the default location: docker volume create automatic1111_data
4. Run docker compose up automatic1111.v1.6 -d.

Automatic1111 should be available at http://127.0.0.1:7860, docker will limit RAM usage to 8GB.

Build and run take a while... depending on your network speed, at the end you will have:

• Docker Image created.
• Docker container running with dedicated volume.
• Automatic1111 installed to the dedicated volume and accesible from your network.

Pre-installed models:

1. epiCPhotoGasm: Last Unicorn
2. Realistic Vision V6.0 B1: 6.0 B1 (VAE)
3. RunDiffusionXL: beta
4. Juggernaut XL: V7 + RunDiffusion
5. RealLife: v3.0

Pre-installed plugins:

1. deforum-for-automatic1111-webui
2. sd-webui-additional-networks
3. sd-webui-controlnet -> with models from Mikubill/sd-webui-controlnet#2039
4. sd-civitai-browser-plus
5. sd-webui-animatediff -> with models from https://github.com/continue-revolution/sd-webui-animatediff#model-zoo
6. posex
7. stable-diffusion-webui-images-browser

Customization

Environment settings

You can customize your container by setting the following env variables:

Environment Variable	Values	Default	Description
IS_LOWVRAM	true/false	true	if set to true IS_MEDVRAM will be ignored, adds: --lowvram --opt-split-attention
IS_MEDVRAM	true/false	false	adds: --medvram
USE_XFORMERS	true/false	true	adds: --xformers
USE_CUDA_118	true/false	true	installs torch optimized for CUDA 11.8

Memory settings

Default memory limit is set to 8 GB. If you want to change it you can manipulate on mem_reservation and mem_limit options from docker-compose.yml.

Kohya_SS - IN PROGRESS.

https://github.com/bmaltais/kohya_ss

FaceFusion V2

https://github.com/facefusion/facefusion

1. Go to facefusion subdirectory: cd facefusion.
2. Run docker compose build facefusion.v2.
3. Create a volume:
   • if you want to place your data in a dedicated location: docker volume create --driver local --opt type=none --opt device=[YOUR_FACEFUSION_DATA_LOCATION] --opt o=bind facefusion_v2_data
   • or use the default location: docker volume create facefusion_v2_data
4. Run docker compose up facefusion.v2 -d.

FaceFusion 2.X should be available at http://127.0.0.1:7862, docker will limit RAM usage to 8GB.

Customization

Memory settings

Default memory limit is set to 8 GB. If you want to change it you can manipulate on mem_reservation and mem_limit options from docker-compose.yml.

Usefull links

• https://github.com/AUTOMATIC1111/stable-diffusion-webui
• https://github.com/facefusion/facefusion
• https://github.com/bmaltais/kohya_ss
• https://www.cherryservers.com/blog/install-cuda-ubuntu
• https://gitlab.com/nvidia/container-images/cuda/-/blob/master/dist/11.7.1/ubuntu2204/base/Dockerfile
• https://docs.nvidia.com/deeplearning/cudnn/install-guide/index.html#package-manager-ubuntu-install
• https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
• https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=22.04&target_type=deb_network