https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEiXC4cOMhjQafCYJbyV7SfkiTgFKip1EeuZKAjWc6AVJEE0iWv7Bbv3FRvxzfGMwBtkN7QSyRed0bsRWKZAHab7CNnmw4BTRQw-u4E6joIBvY00Oz69k20GUBJGAy1brlz3k04jW6hCaEgum7cpUbdm-BCujkWZCfbNgTshDEJKf2PJOewjd_cqIPm3zqA/w426-h640/03759-20240819074220__20240819074211__1203829967__20240819_074220_480088.jpg

 Hello, this is Dr. Ho of Haiti.

Today we will learn about Ghibli style AI creation.

ai image generation stable diffusion



1. Download Python

You can download it from this link or by searching on Google.

https://www.python.org/downloads/release/python-310/

Be sure to download version 3.10.6.


2. Download git.

You can download it from this link or by searching on Google.

https://git-scm.com/download/win

3. Run git and install git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git in the command window.


4. And download the training AI learning model through the link.



Lora model 

Learning model link


5. Downloaded learning model file 


 Copy the downloaded learning models to the stable-diffusion-webui\models\Stable-diffusion path where Web UI is installed.



6. Download the vae-ft-mse-840000-ema-pruned.safetensors file and paste it into WebUI installation folder > models > vae.



7.Execute all of the above and run the webui-user.bat file to run Stable Diffusion Webui. The webui-user.bat file is located in the stable-diffusion-webui folder.





After running the Stable Diffusion Webui, run the Internet browser (Chrome) and enter 127.0.0.1:7860 in the address bar to access the Webui.



8. Settings

Go to the webui site, click Checkpoint Merger, and set as follows.


Select Lora in the Primary Model (A) section

Write your desired name in English in the Custom Name section.
Select check point format safetensors
Check Save as float16
In the Make in VAE section, select the vae-ft-mse-840000-ema-pruned.safetensors file downloaded in step 6.

reference 


After selecting the merged model at the Stable Diffusion checkpoint, enter (worst quality, low quality: 1.4), loli, and child in the negative prompt, enter the picture tag you want to create in English in the prompt, and then press the create button to perform AI semi-realistic simulation. You can create a picture.

If an error message related to Python appears, 
You must set environment variables or check the addition of path during installation.

And finally, in order to run the program, the graphics card specifications must be high.
RTX 2080 or so?
I have an RTX 1080 and I tried it and it was possible.