Deep Learning for Image Synthesis

Our very own Sergej Stoppel, who has been responsible for the VCF in the past years, will hold a trial lecture as a part of his PhD evaluation. All are very welcome to get a taste of how a course on machine learning in image synthesis might look like.

Abstract:
Image synthesis is the process of creating new images from some type of image description. In the recent years deep neural networks sidestepped the traditional computer graphics pipeline and instead generate images directly at the pixel level by learning from a large image collection.
This lecture will take you on a journey from the earliest neural networks to the state of the art solutions for deep image synthesis, on which we will uncover the underlying principles behind the solutions.