A light ray output from a subject is separated into left and right light rays by mirrors in an area in which a diverging light ray output from a point of the subject is converted into a parallel light ray. When a subject is looked at from the right side or the left side, the light rays are reflected by various mirrors. The light rays reflected by the mirrors are made incident on image forming lenses. These light rays form an image on a light receiving surface of an imaging device. In this way, a high-quality 3D image having an appropriate parallax level and a small chromatic aberration is captured.