Achieving a configuration that reduces the artificiality to give a strange feeling about the viewpoint of the user displayed on the display unit appearing different from the actual viewpoint. Photographed images from a plurality of different viewpoints are input to generate a plurality of virtual viewpoint images, and then, the plurality of virtual viewpoint images is combined to generate a combined image to be output on a display unit. The virtual viewpoint image generation unit generates a plurality of user viewpoint-corresponding virtual viewpoint images each corresponding to each of viewpoints of each of a plurality of viewing users viewing the display unit, while the image combining unit extracts a portion from each of the plurality of user viewpoint-corresponding virtual viewpoint images in accordance with a relative position between the viewing user and the display unit, and combines the extracted image to generate a combined image. The combined image is generated by extracting a display region image located at a front position of the viewing user at the viewpoint corresponding to the virtual viewpoint image from among the user viewpoint-corresponding virtual viewpoint images corresponding to individual viewing users.