Furkan Gözükara
patreon
Auto Windows Installer For Würstchen: Fast Diffusion for Image Generation
🕑 Added 2024-07-18 21:00:00 +0000 UTCPatreon exclusive posts index to find our scripts easily
Join discord to get help, chat, discuss and also tell me your discord username to get your special rank : SECourses Discord
Please also Star, Watch and Fork our Stable Diffusion & Generative AI GitHub repository and join our Reddit subreddit
19 July 2024 Update
Libraries updated to the latest version - working blazing fast
Generated images will be automatically saved inside outputs folder
--share args added
Works great with latest Torch 2.3.0 and xFrmers 0.0.26.post1
How to Use
Extract attached zip file and install and use on your desired platform.
The generated images will be saved automatically inside outputs folder
Requirements for Windows
You need to have installed Python 3.10.x, C++ tools, Cuda 11.8, FFmpeg and Git
Full tutorial to install above properly : https://youtu.be/-NjNy7afOQ0
Würstchen is explained here : https://huggingface.co/blog/wuertschen
It is super fast compared to the SDXL at 1024x1024. But quality I am yet to explore
The models will be automatically downloaded into your Hugging Face cache folder
models--warp-ai--wuerstchen , models--warp-ai--wuerstchen-prior
Total around 11 GB
The files will be downloaded into this folder : C:\Users\your_username\.cache\huggingface\hub
Comments
Furkan Gözükara
sorry for the delay. just updated the post and the installers. now works blazing fast and added few new features. it took me a lot to download models on limited mobile data :D
Việt Shin
Hello. Enter your choice (1-2): 1 Please select an option: 1. Default Precision - Higher quality higher VRAM 2. FP8 Precision - Slightly lower quality medium VRAM 3. FP4 Precision - More slightly lower quality lower VRAM Enter your choice (1-3): 2 WARNING[XFORMERS]: xFormers can't load C++/CUDA extensions. xFormers was built for: PyTorch 2.2.0+cu121 with CUDA 1201 (you have 2.2.0+cpu) Python 3.10.11 (you have 3.10.14) Please reinstall xformers (see https://github.com/facebookresearch/xformers#installing-xformers) Memory-efficient attention, SwiGLU, sparse and more won't be available. Set XFORMERS_MORE_DETAILS=1 for more details function 'cadam32bit_grad_fp32' not found
Furkan Gözükara
git and python are necessary correct. others are auto installed inside venv we make
Andrex Selivanov
Git, python, and a lot module's neded before running this installation (((( try to install on clean computer .
Furkan Gözükara
thank you so much
Rafał Ryniak
it's not that important, I can work on your other solutions, don't worry about it, you're doing a great job, thank you very much for your interest.
