Unlike some other projects built around Stable Diffusion this Discord bot is setup only for NVIDIA GPUs and not optimized for low VRAM hardware. Consult this table before attempting to run the bot:
VRAM | Comment |
---|---|
10GB+ | Recommended - bot will run fine |
8GB-10GB | Bot will most likely run |
6GB-8GB | Bot might run - issues expected |
<6GB | Bot will probably not run, don't ask for support |
Make sure you have the CUDA Toolkit installed
Clone the repository and enter it
git clone https://github.com/harubaru/discord-stable-diffusion.git
cd discord-stable-diffusion
Run setup.bat
. If you run into any errors, try running the file as administrator
If you are on a Windows 10 system, run win10patch.bat
Modify the run.bat
file, where
--model_path
is the path to the model (make sure to replace any backslashes with double backslashes),--token=
is the token to the Discord bot--hf_token=
is your huggingface token (can be found here)
Run the run.bat
file
Run ./setup.sh
. If you run into any errors, try using sudo ./setup.sh
Modify the run.sh
file, where
--model_path
is the path to the model,--token=
is the token to the Discord bot--hf_token=
is your huggingface token (can be found here)
Run ./run.sh
To generate an image from text, use the /dream
command and include your prompt as the query. There's tons of parameters to play with so go wild!
To generate an image from another image, use the /dream
command and include the init_image
and strength
parameters. The image needs to be attached to the message.
To fill in a mask in an image, supply a prompt, the init_image
, mask_image
and strength
parameters. The mask needs to consist of black pixels in a transparent image.