🔥 Supports Release 1.5.1 · AUTOMATIC1111/stable-diffusion-webui - released at July 27, 2023

Stable Diffusion Web UI on EC2

A quickstart automation to deploy AUTOMATIC1111 stable-diffusion-webui with AWS EC2.

The CloudFormation template deploys an EC2 instance in your AWS environment. The user data will install and configure Stable Diffusion WebUI automatically at launch. The one-click deployment takes around 20 minutes to complete.

Overview

• A GPU-based EC2 instance

  • AMI: Deep Learning AMI GPU PyTorch 1.13.1 (Ubuntu 20.04) 20230530
  • static public IP address (Elastic IP)
  • swap volumes from instance store
  • security group allows only whitelist IP CIDR to access

• Auto install stable-diffusion-webui and run as a systemctl service

• Auto install some popular models, extensions and embeddings

How to install?

Step 1: Request AWS service quota increase

Open this link to request for service quota increase for G instances. Make sure you are requesting the correct AWS region! The limit has to be at least 4 for running one xlarge instance. You may request for a higher quota such as 8 or 16.

AWS accounts have quotas in each region that limit how many vCPUs of a particular instance type you can run at once. The quota for GPU instances are all 0 for a new AWS account. It may take a couple days for AWS to approve the quota increase.

Step 2: Create EC2 key pair

Follow this link to create a key pair so that you can connect to the instance using SSH.

Step 3: Create CloudFormation stack

1. Download template sd-webui-ec2.yaml and save it locally.

2. Log in to the AWS CloudFormation console at https://console.aws.amazon.com/cloudformation. Choose Create stack with new resources (standard), then choose Upload a template file as template source, with the file that you downloaded in the first step.

3. For stack details,

   • Stack Name: sd-webui (any name you like)

   • Ec2KeyPair: Choose a EC2 key pair

   • Ec2InstanceType: g5.xlarge or g4dn.xlarge should be good.

     • g5.xlarge - $0.526 per hour in us-east-1
     • 4 vCPU + 16 GB RAM + NVIDIA A10G GPU / 16 GB VRAM
     • available in limited regions, such as us-east-1 (N. Virginia), us-east-2 (Ohio), us-west-2 (Oregon), eu-west-1 (Ireland), eu-west-2 (London) and ap-southeast-1 (Singapore)
     • it takes around 0.5 minutes to generate an image of 512x768px with 28 steps + highres fix 2x upscaling with 20 steps
     • g4dn.xlarge - $1.006 per hour in us-east-1
       • 4 vCPU + 16 GB RAM + NVIDIA T4 GPU / 16 GB VRAM
     • available in many AWS regions
     • it takes around 1.5 minutes to generate an image of 512x768px with 28 steps + highres fix 2x upscaling with 20 steps

   • CidrBlock: Input your public address so only you can access the Web UI and ssh. Choose 0.0.0.0/0 if you don't mind everyone can access.

4. The stack status will become CREATE_COMPLETE around 10 minutes.

   Command to view installation log: tail -f /var/log/cloud-init-output.log

5. The first launch of stable-diffusion-webui takes up to 10 minutes to install its python dependencies. Then you can open the stable diffusion web UI using the link in output tab.

   Command to view WebUI log: journalctl -f -u sd-webui

Common commands in EC2

You can connect to the EC2 instance using one of the following methods.

• ssh key pair

  ssh -i <key-pair.pem> ubuntu@<ip-address>

• SSM Session Manager (from the Connect button in AWS console), then switch to ubuntu user

  sudo su ubuntu

Common commands:

• View installation log: tail -f /var/log/cloud-init-output.log

• View WebUI log: journalctl -afu sd-webui

• Start WebUI: sudo systemctl start sd-webui

• Stop WebUI: sudo systemctl start sd-webui

• Check WebUI service status: systemctl status sd-webui

• Disable WebUI start on boot: sudo systemctl disable sd-webui

• Run WebUI foreground: cd stable-diffusion-webui/; ./webui.sh --listen

Documentation

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

Installed models, extensions and embeddings

Installed Models

• ChilloutMix

Installed Loras

• JapaneseDollLikeness (v1.5)

• KoreanDollLikeness (v2.0)

• TaiwanDollLikeness (v2.0)

Installed Extensions

• Stable-Diffusion-Webui-Civitai-Helper
• sd-webui-additional-networks
• sd-webui-infinite-image-browsing
• sd-webui-roop
• sd-webui-controlnet
  • Model files for ControlNet 1.1 - canny and openpose only
• openpose-editor

Installed Textual Inversion (Embeds)

• EasyNegative
• bad_prompt
• negative_hand
• bad-hands-5

Some models that I like

You can download models easily by pasting the Civitai link in Civitai Helper -> Civitai URL in Download Model.

• SDXL 1.0
• Beautiful Realistic Asians
• DreamShaper

FAQ

Preprocessing images error when auto focal point crop enabled

If you see this error stable diffusion webui pretraining cv2.error: OpenCV(4.8.0) /io/opencv/modules/dnn/src/net_impl.cpp:279: error: (-204:Requested object was not found) Layer with requested id=-1 not found in function 'getLayerData', trying downgrade OpenCV to 4.7.0.72 as suggested here.

cd stable-diffusion-webui/
source venv/bin/activate
pip install opencv-python==4.7.0.72
deactivate
sudo systemctl restart sd-webui

How to change COMMANDLINE_ARGS

Modify /home/ubuntu/stable-diffusion-webui/webui-user.sh and restart WebUI by sudo systemctl start sd-webui. More details of command line arguments can be found at https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Command-Line-Arguments-and-Settings.

Credits

• https://koding.work/use-aws-ec2-to-build-stable-diffusion-webui/

• https://ivonblog.com/posts/stable-diffusion-webui-manuals/

• https://towardsdatascience.com/create-your-own-stable-diffusion-ui-on-aws-in-minutes-35480dfcde6a