After decompressing the BL and EL streams, the decoder needs to reconstruct the left and right views at full resolution. One example embodiment of such a method, to be referred to as “the difference” method, is depicted in FIG. 13A. The vertical or horizontal frequencies that are missing in the base layer can be constructed as a pixel-wise difference of the enhancement layer (1262) and the carrier signal 1302. Carrier signal 1302 can be reconstructed using decoded BL signal 1252 and the decoder RPU (e.g., 1255) using processing that matches the processing of the encoder Carrier RPU (1220). An example of such a process is depicted in FIG. 15. Residue signal 1317 is then up-sampled (e.g., in 1320) and merged by pixel-wise addition with the up-sampled frame-compatible (FC) reconstructed base layer (FC-L and FC-R) to reconstruct the full resolution (FR) left and right views (FR-LE and FR-RE). In one embodiment, to reduce complexity, one may reuse the output from codec RPU 1255 because the codec RPU and carrier RPU tend to share the same filters. In the case when one partition and one filter are employed in the codec RPU, the two RPUs apply exactly the same processing.
As depicted in FIG. 13A, legacy receivers may still decode a pair of half-resolution, frame-compatible, views (FC-L and FC-R), by performing the proper up-sampling on the BL signal 1252.
