Department of Computer Science & Engineering

High-resolution video playback (> 1 megapixel) has become a commodity in homes (Blu-Ray players, internet streaming) and movie theaters (digital HD technology). Immersive virtual reality systems can display tens of millions of pixels today, for instance CAVE-like environments driven by 4k projectors. However, when video is displayed in virtual environments (VEs), where the video screen is part of the virtual world, the resolution of the video is fairly low, and so is its frame rate, typically much lower than standard TV. Allowing high-resolution video playback in VEs can add more realism to the virtual world (e.g., a virtual movie theater), and it can enable a new class of applications which were not possible before (e.g., virtual video surveillance centers). In this paper, we propose an algorithm based on mipmapped video frames, where each image of the video stream is stored at multiple levels of resolutions, to interactively play high-resolution video in VEs. In addition, we propose an approach to maintain a constant video playback rate, as well as optimizations for the algorithm, such as a memory management mechanism and predictive prefetching of data. Finally, we analyze the playback of three different types of high-resolution video clips in an immersive VE.

Pre-2018 CSE ID: CS2009-0952