<p>Various implementations are described. Several implementations relate to depth coding. According to a general aspect, prediction mode information is accessed for reconstructing a target block of data from a picture. The prediction mode information indicates that the target block of data was predicted using a target-block prediction mode that was determined based on a neighbor-block prediction mode of a neighbor block. The neighbor block is a spatial neighbor of the target block in the picture. The neighbor-block prediction mode of the neighbor block is determined. The target-block prediction mode of the target block is determined based on the neighbor-block prediction mode. The target block is reconstructed by predicting the target block using the determined target-block prediction mode.</p>