The embodiments use prediction weights in connection with predictive encoding and decoding of a depth map in the form of a frame (10) with pixels (11) having pixel values representing distances between a camera (40) and objects. The prediction weight is calculated based on depth clipping plane parameter values associated with the frame (10) and with a reference frame (20, 30) based on which pixels of the frame (10) are encoded and decoded. Using depth clipping plane parameter values when calculating prediction weights, improves coding efficiency of depth map coding and decoding.