First, a stereoscopic vision reference distances is individually determined for each virtual object disposed in a virtual space, based on a depth position of each of the virtual objects. Next, each of the virtual objects is rendered and combined based on one pair of virtual cameras such that a relative position, with respect to the screen of a stereoscopic display device, of a virtual object, viewed by a user through a screen of the stereoscopic display device, matches a relative position of the virtual object with respect to a stereoscopic vision reference surface for the virtual object in the virtual space.