A programmable display-layer architecture for virtual-reality applications

Two important technical objectives of virtual-reality systems are to provide compelling visuals and effective 3D user interaction. In this respect, modern virtual reality system architectures suffer from a number of short-comings. The reduction of end-to-end latency, crosstalk and judder are especially difficult challenges, each of which negatively affects visual quality or user interaction. In order to provide higher quality visuals, complex scenes consisting of large models are often used. Rendering such a complex scene is a time-consuming process resulting in high end-to-end latency, thereby hampering user interaction. Classic virtual-reality architectures can not adequately address these challenges due to their inherent design principles. In particular, the tight coupling between input devices, the rendering loop and the display system inhibits these systems from addressing all the aforementioned challenges simultaneously. In this thesis, a virtual-reality architecture design is introduced that is based on the addition of a new logical layer: the Programmable Display Layer (PDL). The governing idea is that an extra layer is inserted between the rendering system and the display. In this way, the display can be updated at a fast rate and in a custom manner independent of the other components in the architecture, including the rendering system. To generate intermediate display updates at a fast rate, the PDL performs per-pixel depth-image warping by utilizing the application data. Image warping is the process of computing a new image by transforming individual depth-pixels from a closely matching previous image to their updated locations. The PDL architecture can be used for a range of algorithms and to solve problems that are not easily solved using classic architectures. In particular, techniques to reduce crosstalk, judder and latency are examined using algorithms implemented on top of the PDL. Concerning user interaction techniques, several six-degrees-of-freedom input methods exists, of which optical tracking is a popular option. However, optical tracking methods also introduce several constraints that depend on the camera setup, such as line-of-sight requirements, the volume of the interaction space and the achieved tracking accuracy. These constraints generally cause a decline in the effectiveness of user interaction. To investigate the effectiveness of optical tracking methods, an optical tracker simulation framework has been developed, including a novel optical tracker to test this framework. In this way, different optical tracking algorithms can be simulated and quantitatively evaluated under a wide range of conditions. A common approach in virtual reality is to implement an algorithm and then to evaluate the efficacy of that algorithm by either subjective, qualitative metrics or quantitative user experiments, after which an updated version of the algorithm may be implemented and the cycle repeated. A different approach is followed here. Throughout this thesis, an attempt is made to automatically detect and quantify errors using completely objective and automated quantitative methods and to subsequently attempt to resolve these errors dynamically