StyleGAN2 Distillation for Feed-forward Image Manipulation
This provides a faster alternative for image manipulation applications, but it is incremental as it builds on existing StyleGAN2 capabilities.
The paper tackled the problem of slow latent code optimization for editing images with StyleGAN2 by distilling manipulations into a feed-forward image-to-image network, achieving comparable quality to backpropagation and state-of-the-art methods in tasks like gender swap and aging.
StyleGAN2 is a state-of-the-art network in generating realistic images. Besides, it was explicitly trained to have disentangled directions in latent space, which allows efficient image manipulation by varying latent factors. Editing existing images requires embedding a given image into the latent space of StyleGAN2. Latent code optimization via backpropagation is commonly used for qualitative embedding of real world images, although it is prohibitively slow for many applications. We propose a way to distill a particular image manipulation of StyleGAN2 into image-to-image network trained in paired way. The resulting pipeline is an alternative to existing GANs, trained on unpaired data. We provide results of human faces' transformation: gender swap, aging/rejuvenation, style transfer and image morphing. We show that the quality of generation using our method is comparable to StyleGAN2 backpropagation and current state-of-the-art methods in these particular tasks.