/stable-diffusion-tensorflow

Tensorflow port of the Stable Diffusion Model

Primary LanguageJupyter Notebook

Stable Diffusion in Tensorflow / Keras

A Keras / Tensorflow implementation of Stable Diffusion.

This implementation is easy to understand and have a minimal code footprint.

I have converted the weights from the original Pytorch implementation.

How to use

  1. Using the command line
python text2image.py --prompt="An astronaut riding a horse"
  1. Using the python interface
pip install --upgrade git+https://github.com/divamgupta/stable-diffusion-tensorflow

from stable_diffusion_tf.stable_diffusion import get_model, text2image
text_encoder, diffusion_model, decoder = get_model(512, 512, download_weights=True)

img = text2image("An astronaut riding a hourse" , 
	img_height=512, 
	img_width=512,  
	text_encoder=text_encoder, 
	diffusion_model=diffusion_model, 
	decoder=decoder
)

cv2.imwrite("/tmp/a.png" , img[0][... , ::-1])

Outputs

The following outputs have been generated using the Tensorflow/Keras of stable diffusion:

  1. A epic and beautiful rococo werewolf drinking coffee, in a burning coffee shop. ultra-detailed. anime, pixiv, uhd 8k cryengine, octane render

a

  1. Spider-Gwen Gwen-Stacy Skyscraper Pink White Pink-White Spiderman Photo-realistic 4K

a

  1. A vision of paradise, Unreal Engine

a

References :

  1. https://github.com/CompVis/stable-diffusion

  2. https://github.com/geohot/tinygrad/blob/master/examples/stable_diffusion.py