How-To-Use-Stable-Diffusion-SDXL-Locally-And-Also-In-Google-Colab.md

[Tutorial] How To Use Stable Diffusion SDXL Locally And Also In Google Colab

On Google Colab

Click to open Colab link

Click to see where Colab generated images will be saved

Now you can set any count of images and Colab will generate as many as you set

On Windows - WIP

Prerequisites

Full tutorial for python and git installation with venv

• How To Install Python, Setup Virtual Environment VENV, Set Default Python System Path & Install Git

Download and install Python 3.10.x (3.10.6, 3.10.9, 3.10.11) and git

• 3.10.9 > https://www.python.org/ftp/python/3.10.9/python-3.10.9-amd64.exe
• git > https://git-scm.com/downloads

Automatic Installation

• This installation won't affect or your break your other installations such as Automatic1111 Web UI
• Auto Installer Script : https://www.patreon.com/posts/85678961
• Instructions of auto installer shared in the post

Manual Installation

• This installation won't affect or your break your other installations such as Automatic1111 Web UI

git clone https://github.com/FurkanGozukara/stable-diffusion-xl-demo

• Register Hugging Face and login if you don't have an account already

• https://huggingface.co/join

• https://huggingface.co/login

• Open below links and accept terms and conditions - now auto approve official weights - just type anything to the form fill sections

• https://huggingface.co/stabilityai/stable-diffusion-xl-base-0.9

• https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-0.9

• Generate your Hugging Face access token and save it somewhere

• https://huggingface.co/settings/tokens

• Enter inside cloned stable-diffusion-xl-demo folder

• Follow the below commands 1 by 1. Watch video to see how I am making

python -m venv venv

cd venv

cd scripts

activate

cd..
cd..

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

pip install -U --pre xformers

pip install -r requirements2.txt

After Installation Run As Below

• First edit app2.py and add your access_token. E.g. like below

• access_token = "hf_ovfazhEBOXmTpUsfdfdshfdydZIfMNJVKrcZ"

• The access token is only 1 time

• After that you have to save below commands depending on your VRAM as .bat file and use them to start

• Watch tutorial video again if you can't make it. All steps are shown

Low VRAM (12 GB and Below)

@echo off

set VENV_PATH=stable-diffusion-xl-demo\venv\Scripts\activate.bat

call "%VENV_PATH%"

set PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:128
set ENABLE_REFINER=false

python stable-diffusion-xl-demo\app2.py

pause

High VRAM (24 GB and above)

@echo off

set VENV_PATH=stable-diffusion-xl-demo\venv\Scripts\activate.bat

call "%VENV_PATH%"

set PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:256
set ENABLE_REFINER=true

python stable-diffusion-xl-demo\app2.py

pause