An image-taking apparatus generates an output image from a photographed image obtained by photographing and displays this output image on a display screen of a display section. The image-taking apparatus includes a recognition processing section for recognizing a specific subject (facial region of a person, or a facial component such as eyes) included in the photographed image. Referring to results of this recognition, the image-taking apparatus evaluates the dimension of the specific subject in each photographed image. If the dimension evaluated is relatively large, a relatively small magnification ratio to be employed upon the generation of the output image from the photographed image is provided. If the dimension evaluated is relatively small, a relatively large magnification ratio to be employed upon the generation of the output image from the photographed image is provided. As a result, the dimension of the specific subject in the output image is kept substantially constant.