Perception-Based Time Critical Rendering

To maintain the high-throughput and low-latency required by immersive graphics environments, developers employ time-critical rendering techniques to reduce graphics complexity. Because our visual system does not use all of the optical information in a scene, I assert that a perception-based approach to reducing graphics complexity is an effective strategy for developing time-critical rendering, and that this approach provides application-independence. Previous immersive systems such as vehicle simulators and virtual environment architectural walkthroughs have employed application-specific techniques to determine what graphics components of the scene may be reduced without significantly impacting the quality of the environment. But as immersive environments evolve into a general medium for human-computer interaction, application-independent techniques become important for general, time-critical rendering systems.
My approach was to exploit the characteristics and limitations of the human visual system as derived from perceptual psychology to drive the development of application-independent, complexity-reduction techniques. I contribute to the field (1) by demonstrating the utility of bridging perceptual psychology with computer engineering to reduce graphics complexity; (2) by constructing a general rendering system which transparently separates the application-specific computations from the rendering process; (3) by developing a perception-based framework for quickly establishing the relative importance of graphics objects in the visual field; (4) by developing a graphics complexity reduction technique based on scaling objects in the visual periphery; (5) by exploring and developing binocular-based techniques for reducing graphics complexity; and (6) by developing a new perception-based technique (GPIR) for reducing image memory requirements.
The new perception-based complexity-reduction techniques were measured to determine the amount of potential time savings they can produce. Since these techniques may affect scene quality, I also report the results of informal user studies to help assess the effect of degradation on a simple task: searching for objects in the immersive environment.