Unified Graphics API for a Display-Agnostic Rendering System

Abstract

This research focuses on development of a unified graphics application programming interface in which several types of 3D-displays that require multi-view rendering, such as Virtual Reality/Augmented Reality Headsets, Light Field Displays, and Volumetric Displays, can exist within a heterogeneous display environment. In general, GPU architecture and programming languages are not inherently well optimized for multi-view rendering. 3D displays such as stereo, volumetric and light-field displays may require a minimum of two to several hundreds or thousands of views to be rendered per display update based on the architecture of the display. Moreover, there is not a single binding software architecture standard that can exploit the common multi-view rendering needs of multi-view 3D displays. In this thesis, I extend the current multi-view light-field rendering graphics library Object Graphics Language(ObjGL), a single common standard library developed at FoVI^3D, to support stereo rendering in virtual reality displays including Oculus Rift, and HTC Vive, as well as explore the potential challenges in the march towards a truly heterogeneous display environment.