A stereoscopic image is generated by applying three-dimensional processing to two or more images, which are obtained through photographing from different positions. The stereoscopic image is stereoscopically displayed, and an instruction to start changing a parallax level of the stereoscopic image is received. In response to the instruction to start changing the parallax level, stereoscopic display of the stereoscopic image is switched into two-dimensional display of the two or more images overlapped with each other. An instruction to change the parallax level is received. In response to the instruction to change the parallax level, the two or more images are two-dimensionally displayed with the parallax level of the images being changed.