Abstract
We propose an architecture for lightweight visualization of high-quality 3D objects based on data compression, data streaming, virtual texturing and WebGL. Our method retains visual fidelity of the original scene, improves loading time and maintains real-time rendering speed. We assume that the user is restricted to low-performance GPUs and slow Internet connections (1 megabit per second or lower). For geometry compression, we use entropy-encoding techniques that achieve up to 95% storage savings. Textures are stored as sets of tiles which feeds the virtual texturing engine. With use of the Crunch library, tiles are compressed with results similar to JPEG but with much faster transcoding to DXT on the GPU. The initial 27.7MB dataset takes an average of 5 minutes to load. Our approach, takes less than 5 seconds on average. A wide range of applications benefit from our architecture such as e-commerce, cultural heritage, virtual worlds, videogames, and scientific visualization, among others.