Incremental encoding of polygonal models

The paper presents a method for generating a stream of data which continuously refines a polygonal model from the coarsest representation to the original high-fidelity model. Unlike conventional level of detail algorithms that use only a small number of discrete approximations for each object, the method allows a smooth, continuous refinement of the object over time while the stream is transmitted. The approach is based on a hierarchical clustering algorithm which produces the model representation of steadily increasing detail and has several advantages over conventional models with few discrete levels of detail: the model can be transmitted and decoded incrementally, yielding a valid approximation at any stage of the process. The incremental encoding is extremely compact, so models are smaller than in their original form. Any desired level of detail can be selected during rendering at interactive speed. It is also possible to have variable detail resolution within a single model, which is useful for rendering large models