We propose a method of interframe prediction in depth map coding that uses pixel-wise 3D motion estimated from encoded textures and depth maps. By using the 3D motion, an approximation of the depth map frame to be encoded is generated and used as a reference frame of block-wise motion compensation.