This repository aims to reproduce and study some fundamental embedding algorithms, particularly Image2StyleGAN: How to Embed Images Into the StyleGAN Latent Space? and Improved StyleGAN Embedding: Where are the Good Latents?. Also, it is intended to add further experimentations and review some editing operations.
This work was done during my research internship at King Abdullah University of Science and Technology (KAUST), and presented as an article poster for the Eastern European Machine Learning Summer School - EEML 2021.
- For learning purposes, you can explore the notebooks at notebooks/, where all the programming process was documented. I recommend to explore this first because you can have a better idea about my process step-by-step.
- The optimzer.py script encloses the main optimization algorithm for embeding images into the StyleGAN latent space.
- The scripts inpainting.py, super_resolution.py, colorization.py, cross_domain.py are scripts based on
optimizer.py, but with few settings to create different processing images tasks.