In the advent of a new age of natural user interaction with 3D gesture recognition cameras (e.g. Kinect, Leap Motion, Intel Real Sense, etc.), it is expected to display a floating image on top of a flat display to directly manipulate rendered objects and to utilize 3D interface with a full function. To present floating images with a conventional 2D display, researchers have studied diverse, passive re-imaging techniques to converge the displayed image in a midair. For instance, Repro3D [Yoshida et al. 2011] combined a pepper's ghost technique with a retro-reflective multi-array-projection to augment a floating, motion-parallax image with real scenes at the back of a half-mirror. Smoot et al. [Smoot et al. 2011] demonstrated a volumetric scanning mid-air display by a rim driven varifocal beam-splitter and a high-frame rate LED display. HaptoMIRAGE [Ueda et al. 2014] applied three large fresnel lens in a lateral direction to construct a wide-angle-view autostereoscopic display with a haptic feedback. Monnai [Monnai et al. 2014], Kim [Kim et al. 2014] and Maekawa [Maekawa et al. 2006] employed micro corner mirror arrays to form a clear real image with a wide range of focal points.