Novel View Synthesis: a comparative analysis study

The problem of synthesizing novel views consists of generating new views of a scene using at least two reference views. The new generated views, called also novel views, correspond to virtual viewpoints where no real camera has taken snapshots. In the last few year, view synthesis has been a thriving field of research in both computer vision and computer graphics. Applications of novel view synthesis range from generating photorealistic images to the modeling of real scenes. Several approaches have been proposed in the literature for novel view synthesis. However, no comparison has been done on these approaches to evaluate their performances and qualities. This paper, aims at providing a quick overview and a comparative study of all known methods for view synthesis. The results of this study are of great interest to any work involving view synthesis. Experiments with both simulated and real images were carried out to support our conclusions.