VRAIS '97 Proceedings of the 1997 Virtual Reality Annual International Symposium (VRAIS '97)
| Hi-index | 0.00 |
This paper presents a method to generate a stream ofdata which continuously refines a polygonal model fromthe coarsest representation to the original high-fidelitymodel. Unlike conventional level of detail algorithmsthat use only a small number of discrete approximationsfor each object, our method allows a smooth, continuousrefinement of the object over time while the stream istransmitted. The approach is based on a hierarchicalclustering algorithm which produces the modelrepresentation of steadily increasing detail and hasseveral advantages over conventional models with few,discrete levels of detail: The model can be transmittedand decoded incrementally, yielding a validapproximation at any stage of the process. Theincremental encoding is extremely compact, so modelsare smaller than in their original form. Any desired levelof detail can be selected during rendering at interactivespeed. It is also possible to have variable detailresolution within a single model, which is useful forrendering models with large extent.