Fast computation of seamless video loops

Abstract: Short looping videos concisely capture the dynamism of natural scenes. Creating seamless loops usually involves maximizing spatiotemporal consistency and applying Poisson blending. We take an end-to-end view of the problem and present new techniques that jointly improve loop quality while also significantly reducing processing time. A key idea is to relax the consistency constraints to anticipate the subsequent blending, thereby enabling looping of low-frequency content like moving clouds and changing illumination. We also analyze the input video to remove an undesired bias toward short loops. The quality gains are demonstrated visually and confirmed quantitatively using a new gradient-domain consistency metric. We improve system performance by classifying potentially loopable pixels, masking the 2D graph cut, pruning graph-cut labels based on dominant periods, and optimizing on a coarse grid while retaining finer detail. Together these techniques reduce computation times from tens of minutes to nearly real-time.