So, after all these weeks, you have decided to update StableDiffusion.
With all the extensions that you tried, you rapidly see that you have some problems of compatibility with some of them, so you decide that it could be a good idea to make a clean reinstallation of StableDiffusion.
So this is for a Windows computer:
It is important to notice that StableDiffusion doesn't use the classic windows installer procedure.
Your data is stored in a directory and StableDiffusion uses a virtual environment to have the good settings for Python for the software to run.
My first step is to locate the loras and the StableDiffusion models that you may want to save for your new installation.
Then I would destroy the directory where stablediffusion is saved.
Then I go to:
C:\Users\{username}\.conda
open the environments.txt file and check if the line related to StableDiffusion mentions a directory that still exists. Consider if the directory and/or the line in this file should be removed or not. (What did you do with the environment ? Just StableDiffusion ? Then you can probably delete).
If you used the normal procedure to install stablediffusion, you shouldn't see anything here.
Now we are ready for a new installation.
- We suppose here that you have all the python, the conda, the git, the pip... softwares that you need.
- Now create the directory where StableDiffusion should be save and open a command line window there (with CMD)
and type:
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
Once it is downloaded, go to directory where the new files are and run:
webui-user.bat
from Windows Explorer.
In the Command Line window that is open, at some point, you should see something like:
"Running on local URL: http://127.0.0.1:7860"
Open your favorite web browser and open the url that is given.
For me it is: http://127.0.0.1:7860
This is Automatic1111.
Now let's download and install SDXL for it:
You could visit this page, to have an idea of the models created by Stability AI itself:
https://huggingface.co/stabilityai
I would suggest to go here:
https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/tree/main
Download the model with a vae: sd_xl_base_1.0_0.9vae.safetensors (6.94 GB)
and paste it in this subdirectory:
(directory of Stable Diffusion)\stable-diffusion-webui\models\Stable-diffusion
Make sure to update the list of models to use the model and select the new sd_xl model.
Then set the width and the height to 1024.
And type your prompt, for example:
Fluffy and cute anthropomorphic cat dressed as Robin Hood holding a bow
If your installation is clean enough, your can use the command line "git pull" to update the interface with the latest files.