Identifier
etd-11152015-211541
Degree
Doctor of Philosophy (PhD)
Department
Computer Science
Document Type
Dissertation
Abstract
Shadows portray helpful information in scenes. From a scientific visualization standpoint, they help to add data without unnecessary clutter. In video games they add realism and depth. In common graphics pipelines, due to the independent and parallel rendering of geometric primitives, shadows are difficult to achieve. Objects require knowledge of each other and therefore multiple renders are needed to collect the necessary data. The collection of this data comes with its own set of trade offs. Our research involves adding shadows into a lunar rendering framework developed by Dr. Robert Kooima. The NASA-collected data contains a multi-gigapixel displacement map describing the lunar topology. This map does not fit entirely into main memory and therefore out-of-core paging is utilized to achieve real-time speeds. Current shadow techniques do not attempt to generate occluder data on such a scale, and therefore we have developed a novel approach to fit this situation. By using a chain of pre-processing steps, we analyze the structure of the displacement map and calculate horizon lines at each vertex. This information is saved into several images and used to generate shadows in a single pass, maintaining real-time speeds. The algorithm is even capable of generating soft shadows without extra information or loss of speed. We compare our algorithm with common approaches in the field as well as two forms of ground truth; one from ray tracing and the other from the gigapixel lunar texture data, showing real shadows at the time it was collected.
Date
2015
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Recommended Citation
Cherry, Kevin Anthony, "Real-time Shadows for Gigapixel Displacement Maps" (2015). LSU Doctoral Dissertations. 522.
https://repository.lsu.edu/gradschool_dissertations/522
Committee Chair
Kooima, Robert
DOI
10.31390/gradschool_dissertations.522