stable-diffusion-finetune/README.md

Experiments with Stable Diffusion

This repository extends and adds to the original training repo for Stable Diffusion.

Currently it adds:

• Fine tuning
• Image variations
• Conversion to Huggingface Diffusers

Fine tuning

Makes it easy to fine tune Stable Diffusion on your own dataset. For example generating new Pokemon from text:

Girl with a pearl earring, Cute Obama creature, Donald Trump, Boris Johnson, Totoro, Hello Kitty

For a step by step guide see the Lambda Labs examples repo.

Image variations

For more details on the Image Variation model see the model card.

• Get access to a Linux machine with a decent NVIDIA GPU (e.g. on Lambda GPU Cloud)
• Clone this repo
• Make sure PyTorch is installed and then install other requirements: pip install -r requirements.txt
• Get model from huggingface hub lambdalabs/stable-diffusion-image-conditioned
• Put model in models/ldm/stable-diffusion-v1/sd-clip-vit-l14-img-embed_ema_only.ckpt
• Run scripts/image_variations.py or scripts/gradio_variations.py

All together:

git clone https://github.com/justinpinkney/stable-diffusion.git
cd stable-diffusion
mkdir -p models/ldm/stable-diffusion-v1
wget https://huggingface.co/lambdalabs/stable-diffusion-image-conditioned/resolve/main/sd-clip-vit-l14-img-embed_ema_only.ckpt -O models/ldm/stable-diffusion-v1/sd-clip-vit-l14-img-embed_ema_only.ckpt
pip install -r requirements.txt
python scripts/gradio_variations.py

Then you should see this:

Trained by Justin Pinkney (@Buntworthy) at Lambda