綠茶妹 發表於 23-2-20 22:48

在自己的電腦上安裝Stable Diffusion (AI生成圖片)

本帖最後由 綠茶妹 於 23-2-20 23:09 編輯



Stable Diffusion是AI生成圖片的工具,類似Midjourney。
我爬文研究了好幾套,這一套做出來可以商用,
也不會有Disco Diffusion執行很緩慢的問題。


嘿嘿嘿...終於成功了。
雖然網路上很多教學。但是最後有一個error我是爬文後才解決的。
值得記錄一下。

先貼上原始的教學文章
https://blog.jimmycstudio.com/stablediffusion-local-install/

這是我後面遇到的問題


解決方法在這裡
https://www.reddit.com/r/StableD ... tomatic1111_stable/

1- download miniconda from here https://repo.anaconda.com/miniconda/Miniconda3-latest-Windows-x86_64.exe the installation is as easy as clicking next, next, next

2- download git for windows from here https://github.com/git-for-windows/git/releases/download/v2.39.0.windows.2/Git-2.39.0.2-64-bit.exe make sure to add it to path when asked, it's just a tickbox you have to enable

3- press your windows key and search for "miniconda" and open Anaconda prompt (miniconda)

4- now type this and press inter

git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
5- now type this and press inter

cd stable-diffusion-webui
6- finally type this and press inter

python launch.py
wait for it to be complete and download v1.5 the ckpt model (google stable diffusion 1.5 model), rename it to model.ckpt and place it inside models\Stable-diffusion folder inside the stable-diffusion-webui folder and it should work from there, the v2 and v2.1 models are a downgrade so don't bother with them

let me know if this worked for you or if you need more help





綠茶妹 發表於 23-2-20 22:53

本帖最後由 綠茶妹 於 23-2-20 22:55 編輯

測試做出來的效果

Please paint a watercolor painting for me and draw a French street scene. The weather is very nice, with a few white clouds in the sky. There are three small tables on the side of the road, and many people are happily drinking coffee and chatting.





執行畫面





綠茶妹 發表於 23-2-20 23:06





Please draw a Japanese anime boy holding a Shiba Inu and walking on a road with many cherry blossoms.





Evil 發表於 23-2-21 09:33

現今的科技越來越可怕了!

miyusuea 發表於 23-2-21 19:46

{:4_153:}
頁: [1]
查看完整版本: 在自己的電腦上安裝Stable Diffusion (AI生成圖片)