The optical image on a sensor formed by a conventional lens mostly determines the quality of final image. Computational Photography is a concept to change the game of taking photograph by offering attractive functions such as blur reduction or posterior focus adjustment. In this framework, optical device is assumed as an encoder of sampled incoming rays, and the final image is reconstructed by decoding of the captured intensity distribution on the sensor. In this article I will introduce several basic concepts and achievements in this field by reproducing my talk in ISSCC2012.