An apparatus for determining a depth map for an image comprises an image unit (105) which provides an image with an associated depth map comprising depth values for at least some pixels of the image. A probability unit (107) determines a probability map for the image comprising probability values indicative of a probability that pixels belong to a text image object. A depth unit (109) generates a modified depth map where the modified depth values are determined as weighted combinations of the input values and a text image object depth value corresponding to a preferred depth for text. The weighting is dependent on the probability value for the pixels. The approach provides a softer depth modification for text objects resulting in reduced artefacts and degradations e.g. when performing view shifting using depth maps.