Object-based tilt-shift photography

Abstract

Tilt-Shift photography captures specific focused range by tilt-shift lens. The tilt-shift lens compensate the perspective problems by tilt and change the depth of field by shift. Using the tilt-shift lens, the viewers can pay their attention to the specific objects naturally. In this paper, we develop an automatic method to produce tilt-shift images. An energy minimization approach is proposed by considering the object property and boundary smoothness based on the Markov Random Field (MRF) framework. In this way, the weaknesses of tilt-shift lens, such as high-cost, huge and heavy can be improved. Our results are compared with other software approaches and show the effectiveness of our algorithm.