A value for expressing distance from a virtual viewpoint is set to every predetermined compositional unit of a first image; a second image is generated; a coefficient is defined corresponding to the value for expressing the distance set to every predetermined compositional unit; and the first image and the second image are synthesized based on the coefficient defined for every predetermined compositional unit. This successfully creates an image with a natural perspective by which the compositional unit is blurred according to the distance thereof from the virtual viewpoint, while causing only a small load to CPU.