A Hybrid LOD-Sprite Technique for Interactive Rendering of Large Datasets

We present a new rendering technique, termed LOD-sprite rendering, which uses a combination of a level-of-detail (LOD) representation of the scene together with reusing image sprites (previously rendered images). Our primary application is an acceleration technique for virtual environment navigation. The LOD-sprite technique renders an initial frame using a full-resolution model of the scene geometry. It renders subsequent frames with a much lower-resolution model of the scene geometry and texture-maps each polygon with the image sprite from the initial full-resolution frame. As it renders these subsequent frames the technique measures the error associated with each low-resolution polygon, and uses this to decide when to re-render the scene using the fullresolution model. The LOD-sprite technique can be efficiently implemented in texture-mapping graphics hardware. The LOD-sprite technique is thus a combination of two currently very active thrusts in computer graphics: level-of-detail representations and image-based modeling and rendering (IBMR) techniques. The LOD-sprite technique is different from most previous IBMR techniques in that they typically model the texture-map as a quadrilateral, as opposed to a lower-resolution scene model. This scene model, even if only composed of a few polygons, greatly increases the range of novel views that can be interpolated before unacceptable distortions arise. Also unlike previous LOD techniques, the LOD-sprite algorithm dynamically updates the image sprite every several frames. The LOD-sprite technique can be implemented with any LOD decomposition.