Three-dimensional processing is applied to more than one images taken from different points of view to generate a stereoscopic image, and the stereoscopic image is stereoscopically displayed. An instruction to control a stereoscopic effect of the stereoscopically displayed stereoscopic image is received, and the three-dimensional processing is applied to the more than one images to achieve a stereoscopic effect according to the instruction. Then, control values of the stereoscopic effect according to the instruction are recorded in a recording unit with the control values being associated with the more than one images. At this time, if there are the control values of the stereoscopic effect recorded in the recording unit, difference values between the control values of the stereoscopic effect according to the instruction and the recorded control values of the stereoscopic effect are recorded in the recording unit.