The "Vertigo Effect" on Your Smartphone: Dolly Zoom via Single Shot View Synthesis

Yangwen Liang, Rohit Ranade, Shuangquan Wang, Dongwoon Bai, Jungwon Lee; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2020, pp. 344-345

Abstract


Dolly zoom is a technique where the camera is moved either forwards or backwards from the subject under focus while simultaneously adjusting the field of view in order to maintain the size of the subject in the frame. This results in perspective effect so that the subject in focus appears stationary while the background field of view changes. The effect is frequently used in films and requires skill, practice and equipment. This paper presents a novel technique to model the effect given a single shot capture from a single camera. The proposed synthesis pipeline based on camera geometry simulates the effect by producing a sequence of synthesized views. The technique is also extended to allow simultaneous captures from multiple cameras as inputs and can be easily extended to video sequence captures. Our pipeline consists of efficient image warping along with depth-aware image inpainting making it suitable for smartphone applications. The proposed method opens up new avenues for view synthesis applications in modern smartphones.

Related Material


[pdf] [supp]
[bibtex]
@InProceedings{Liang_2020_CVPR_Workshops,
author = {Liang, Yangwen and Ranade, Rohit and Wang, Shuangquan and Bai, Dongwoon and Lee, Jungwon},
title = {The "Vertigo Effect" on Your Smartphone: Dolly Zoom via Single Shot View Synthesis},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2020}
}