Angular Limits of Image Samples for Novel View Synthesis

Abstract

Novel view synthesis (NVS) is a technology that aims to generate new views of a scene from a limited set of images, enabling rendering from unseen perspectives. This thesis explores the development of NVS and tests the capabilities of NVS to create a middle view of a human face with increasingly more difficult sets of images to create the view from. A novel method is proposed to improve visual quality and efficiency, addressing challenges in reconstruction and view consistency. The results demonstrate the potential of the approach for applications in virtual reality, telepresence, and immersive media, contributing to the development of realistic scene generation.