The script provided in this particular repository is for users who have no GPU or have small amounts of GPU VRAM available. Please note that this process takes a very, very long time to generate images but still is better than nothing 😁
- Install
pytorch
:
pip3 install torch torchvision torchaudio
- Install
diffusers
andtransformers
:
pip3 install diffusers transformers scipy ftfy
Go to line 9 of the file main.py
and then change the value of prompt to whatever you like. Then simply run this:
python3 main.py
The final image will be saved as image.png
in the current working directory. In order to change the name, you can modify line 15 of the main.py
file. Or better, you can randomize it using libraries like uuid
.
IMPORTANT: DO NOT change line 10.
IMPORTANT: This will change soon and there will be some argparse
or sys.argv
stuff in the code. This is a fast proof of concept I wrote for a friend.
- Adding argument and flag support to the code.
- Working on the naming of the output files.