A three dimensional [3D] image signal is processed for rendering 3D image data (33) on a specific 3D display, e.g. an auto-stereoscopic display. A first depth map (34) and saliency of the 3D image data are determined. A display depth sub-range (35) of a usable depth range (36) of the 3D display is determined and provides a higher 3D image quality for a viewer than the 3D image quality across the usable depth range. A depth mapping function is determined in dependence of the saliency data. The depth mapping function maps the first depth map to a second depth map for generating views for the 3D display. Advantageously the salient range of depth values is mapped towards the display depth sub-range.