Multiple marks are attached to a subject. A camera shift direction is set to the right direction. Before capturing Nth (N is a natural number equal to or larger than 2) L and R images, (N−1)th L and R images are read from a memory card, and feature points common to the (N−1)th L and R images are extracted. At least three feature points located in a right edge portion of each of the (N−1)th L and R images are selected as guide points. The right edge portion containing all three guide points is cut out from each of the (N−1)th L and R images to generate L and R guide images which guide a next camera position. The L and R guide images are combined into the left edge portions of the L and R images (through images), respectively. The combined images are displayed on an LCD.