Methods and arrangements for compressing image based rendering data using multiple reference frame prediction techniques that support just-in-time rendering of an image

摘要

Methods and arrangements are provided for compressing image-based rendering (IBR) data, storing and transporting the compressed IBR data, and subsequently providing selective just-in-time (JIT) rendering of an image using only a portion of the compressed IBR data. An array of images is formatted according to a multiple reference frame (MRF) structure that includes anchor frames and predicted frames. Each predicted frame references at least one anchor frame. Each anchor frame is independently encoded. Each predicted frame is encoded using motion compensation and a predicted residue associated with the referenced anchor frames. A resulting MRF bitstream includes encoded anchor frame data, encoded predicted frame data and indexing information that allows for selective random access to individual macroblocks within the encoded anchor frame data and the encoded predicted frame data. Thus, a new view of an image may be rendered by simply accessing only that macroblock information that needed to render the new view. Once decoded, macroblock information can be selectively stored in a cache and accessed again and again, to speed up the rendering of subsequent views.