3D Computer graphics is becoming an almost ubiquitous part of the world in which we live. being present in art. entertainment. advertising. CAD. training and education. scientific visualisation and with the growth of the internet. in e-commerce and communication. This thesis encompasses two areas of study: The design of algorithms for high quality. real-time 3D computer graphics rendering hardware and the methodology and means for achieving this. When investigating new algorithms and their implementation in hardware. it is important to have a thorough understanding of their operation. both individually and in the context of an entire architecture. It is helpful to be able to model different algorithmic variations rapidly and experiment with them interchangeably. This thesis begins with a description of software based modelling techniques for the rapid investigation of algorithms for 3D computer graphics within the context of a C++ prototyping environment. Recent tremendous increases in the rendering performance of graphics hardware have been shadowed by corresponding advancements in the accuracy of the algorithms accelerated. Significantly. these improvements have led to a decline in tolerance towards rendering artefacts. Algorithms for the effective and efficient implementation of high quality texture filtering and edge antialiasing form the focus of the algorithm research described in this thesis. Alternative algorithms for real-time texture filtering are presented in terms of their computational cost and performance. culminating in the design of a low cost implementation for higher quality anisotropic texture filtering. Algorithms for edge antialiasing are reviewed. with the emphasis placed upon area sampling solutions. A modified A-buffer algorithm is presented that uses novel techniques to provide: efficient fragment storage; support for multiple intersecting transparent surfaces; and improved filtering quality through an extendable and weighted filter support from a single highly optimised lookup table.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:310664 |
Date | January 2000 |
Creators | Ewins, Jon Peter |
Publisher | University of Sussex |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Page generated in 0.0019 seconds