Massive Model Visualization: A Practical Solution

Thumbnail Image
Date
2016-01-01
Authors
Bennett, Jeremy
Major Professor
Advisor
James Oliver
Joseph Zambreno
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Authors
Research Projects
Organizational Units
Journal Issue
Is Version Of
Versions
Series
Department
Theses & dissertations (Interdisciplinary)
Abstract

The ever-increasingly complex designs emanating from various companies are leading to a data explosion that is far outstripping the growth in computing processing power. The traditional large model visualization approaches used for rendering these data sets are quickly becoming insufficient, thus leading to a greater adoption of the new massive model visualization approaches designed to handle these arbitrarily sized data sets. Most new approaches utilize GPU occlusion queries that limit the data needed for loading and rendering to only those which can potentially contribute to the final image. By doing so, these approaches introduce disocclusion artifacts that often reduce the quality of the resulting visualization as a camera is maneuvered through the scene. The present research will demonstrate that shader based depth reprojection and OpenGL atomic writes not only increase the performance of an existing system based upon OpenGL occlusion queries, but also reduce the amount of perceived disocclusion artifacts.

Comments
Description
Keywords
Citation
Source
Copyright
Fri Jan 01 00:00:00 UTC 2016