Article Directory
Artificial intelligence is really powerful. . .
I can't fall behind
Although I have not finished learning pytorch before, the hot spots are always changing. Now that large models and generative AI are so popular, I will also come to rub the heat.
Let's start by learning how to use the AIGC tool to generate a wife and a paper man
❤ 2023.6.3 ❤
References
First find a tutorial, just search on station B to find the one with the highest playback volume
Good decision is yours
→→→ [AI painting class of the first system of Station B! Learn Stable Diffusion with zero foundation, this is definitely the easiest AI painting tutorial you have ever seen | SD WebUI Nanny Level Raiders]
The name is really long. . . .
In addition, I also found a more detailed tutorial in CSDN
→→→ [Long text with ten thousand words: Stable Diffusion Nanny-Level Tutorial]
△ Installation and environment configuration
○ Install pre-installed software
· Install Python
The video is about downloading the python installation package from the official website
If you download the installation package directly, you need to download the 3.10 version of python. The video says that the new 3.11 does not support it. The author recommends downloading version 3.10.6.
Remember to check the "add python 3.10 to path" option.
If you need to install different versions of python, you can use anaconda. About installing stable diffusion with anaconda, I also found an article
→→→ [Use Anaconda to implement environment isolation and localization deployment of Stable-Diffusion]
· Install Git
download link
After installation, check whether the installation is correct
git version
OK
· Install CUDA
You can first install "GeForce Experience" to check whether the graphics card driver is up to date
First check which version of cuda the graphics card supports
nvidia-smi
download link
It's a bit big. . .
○ Clone project
Create a folder on the hard disk first, and enter this folder under the command line
If you need to specify a directory, you can set the directory address when executing the git command or enter the directory first
Open the github page of WEBui
→→→【https://github.com/AUTOMATIC1111/stable-diffusion-webui】
Copy the project name and enter it in cmd
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
Automatic installation seems a bit slow. . .
Download the code here and install it manually. . . .
Not too big. . .
good
then double click
Start installing various dependencies
stuck here
It is probably a network problem. I use the scientific method of surfing the Internet, and then use the proxifier to set up a local proxy, and then run it again. I don’t know if it is solved in this way.
But it's so big. . . By the way, haven't I installed CUDA? And why is 11.8 downloaded here.. . .
. . . . it's too slow
I decided to be lazy and add it manually with up's package
※ Autumn leaves installation package
According to up, there is also a one-click installation package here, which is very suitable for beginners. I will come down and try it.
→→→ [[AI Painting] Stable Diffusion integration package v4 released! Brand-new accelerated decompression ready-to-use explosion-proof video memory in three minutes to get started with AI painting☆Renewable☆Training☆Sinicization】
It’s a very long name again. . .
downloaded file
○ Installation dependencies
○ Unzip the compressed package
○ launch launcher
Open the compressed package file and run
one-button start
That's it. . . really convenient