JoJoGAN

GAN, specifically for face generation

Currently, StyleGAN has imporved to a very efficient level.

Arguing this statement, one noticeable feature of JoJoGAN would be taking only single image for a style. Previously, StyleGAN required multiple photos for a specific style to train a certain style. But now, with a single image for a specific style, one can generate one’s own style model, which can be applied to various faces that one chooses.