Creating waterfall animation on a single image

Abstract

A static image always becomes more eye-catching with an animation. In this paper, we present a system for adding a waterfall animation to a single image by extracting a flow animation from a video sequence. To the best of our knowledge, this is the first attempt by researchers to create a waterfall animation on a still waterfall image. Such work poses challenges in many areas, including color consistency, texture consistency, flow velocity, block matching, and block effects. The proposed method integrates optical flow, line integral convolution, color transfer, graph-cut, and multi-resolution splining techniques to mimic a real waterfall on a single image. It uses a segmentation process to separate the necessary foreground and the unnecessary background. Then, flow analysis is performed on the target image and source video. Finally, flow similarity and a synthesis process are applied to form the animation. Experiments generated 8 animation results that prove the feasibility of the proposed method. The limitations and potential impact of this research are also discussed in our experimental results.