Real time visualization of large data using clustered projections

Abstract

Modern graphics systems rely on many techniques to achieve interactive frame-rates for rendering large and complex scenes. Impostoring is a popular technique where rendering of repetitive geometry is optimized by leveraging pre-calculated information for each type of repeated geometry. This paper discusses a view independent impostor-algorithm that stores the sampled projections of a given surface in a clustered format. The implications of such a format on the rendering performance and storage are discussed along with some of its advantages like selective adjustment of level of detail and shadowing. © 2014 Springer International Publishing.