Stereoscopizing cel animations

摘要: While hand-drawn cel animation is a world-wide popular form of art and entertainment, introducing stereoscopic effect into it remains difficult costly, due to the lack physical clues. In this paper, we propose method synthesize convincing animations from ordinary 2D inputs, without labor-intensive manual depth assignment nor 3D geometry reconstruction. It mainly automatic need producing lengthy sequences, but with option allowing users adjust or constrain all intermediate results. The system fits nicely existing production flow animation. By utilizing T-junction cue available in cartoons, first infer initial, not reliable, ordering regions. One our major contributions resolve temporal inconsistency by formulating as graph-cut problem. However, resultant insufficient for generating effect, cannot be directly used its discontinuous nature. We further through an optimization process formulated constraints. This second contribution. optimized result spatiotemporally smooth synthesizing effect. Our has been evaluated on wide range obtained cases.