Distributed hybrid rendering for real-time image generation

Abstract

Applications of interactive 3D graphics, including virtual reality (VR), must satisfy certain performance constraints. It is especially crucial to maintain a suitable frame rate and to respond to users as quickly as possible, disregarding the complexity of a virtual environment. In this thesis, we propose a novel approach - distributed hybrid rendering - on interactive rendering of large-scaled 3D scenes to satisfy these performance constraints. Our approach has two parts - preprocessing part and real-time part. In preprocessing, we divide virtual worlds into hierarchical cells. For each cell, we calculate cell-to-object visibility. A new visibility preprocessing method we introduce efficiently computes potentially visible objects for the given cells. The proposed visibility method deals with general 3D polygonal models and invisible objects jointly blocked by multiple objects. The approach decomposes volume visibility into a set of point visibilities, and then computes point visibility using hardware-visibility queries. We carry out experiments on various large-scale scenes, and show the performance of our algorithm. At real time, we perform a distributed hybrid rendering using preprocessed information. The distributed hybrid rendering performs hybrid rendering combining the traditional polygon-based rendering and a non-polygonal rendering, especially point-based rendering. Our approach uses rendering resources of networked multiple computers to enhance the rendering speed. Theoretically, the rendering cost of our approach does not depend on the complexity of the given 3D virtual worlds, and through experiments, we show that the actual performance of our system, in practice, does not.