jags111/ukiyoe-portrait-diffusion

Ukiyo-e Portrait Diffusion

Ukiyo-e Portrait Diffusion (by avantcontra), is a custom-trained unconditional diffusion model trained on a dataset of 5,000+ 256x256 Ukiyo-e portraits using the fine-tuning openai diffusion model notebook by Alex Spirin.

Ukiyo-e Portrait Diffusion is inspired by Pixel Art Diffusion of KaliYuga.

This sample forked from Disco Diffusion 5.6.

Added "ukiyoe_portrait_diffusion_256" in Models Settings.

https://github.com/avantcontra/ukiyoe-portrait-diffusion/

Showcase:

("Floating Punk" by avantcontra.)

Dataset:

https://www.justinpinkney.com/ukiyoe-dataset/

image size: 256x256

How to use:

1. Select "ukiyoe_portrait_diffusion_256" of "diffusion_model" in Models Settings.
2. Then use DD as usual. The best image size maybe 256x256 cause of the dataset.

avantcontra: twitter, instagram, github