Abstract

While GAN images became more realistic over time, one of their main challenges is controlling their output, i.e. changing specific features such as pose, face shape and hair style in an image of a face. Style-Based Generator Architecture for GANs (StyleGAN), presents a novel model which addresses this challenge. StyleGAN yields state-of-the-art results in the generation of images.

In this work we experiment and exploit these features, within the latent space that StyleGANs provides us, to generate faces with specific characteristics, and ultimately create a tool that allows the customization of these functionalities for experimentation purposes in the ITBA Dream Laboratory. Thus, the laboratory will be able to conduct experiments using a tool that allows them to customize artificial faces, training the network with their own image dataset.