OmnimatteRF: Robust Omnimatte with 3D Background Modeling
This work addresses video matting for applications in casual and professional video production, representing an incremental improvement over prior methods.
The paper tackles the problem of video matting by proposing OmnimatteRF, which combines dynamic 2D foreground layers with a 3D background model to improve scene reconstruction in real-world videos, resulting in better quality reconstructions as demonstrated in experiments.
Video matting has broad applications, from adding interesting effects to casually captured movies to assisting video production professionals. Matting with associated effects such as shadows and reflections has also attracted increasing research activity, and methods like Omnimatte have been proposed to separate dynamic foreground objects of interest into their own layers. However, prior works represent video backgrounds as 2D image layers, limiting their capacity to express more complicated scenes, thus hindering application to real-world videos. In this paper, we propose a novel video matting method, OmnimatteRF, that combines dynamic 2D foreground layers and a 3D background model. The 2D layers preserve the details of the subjects, while the 3D background robustly reconstructs scenes in real-world videos. Extensive experiments demonstrate that our method reconstructs scenes with better quality on various videos.