Computational Cinematography and Photography

Computational Cinematography and Photography
The advance in imaging technology enables people to capture more images and videos. However, most of them are captured by amateur users, and are of low quality, due to various reasons, such as the lack of experience, skill, and professional equipment. My research provides automatic or interactive tools for people to improve image and video quality.
	3D Video Stabilization: A 3D virtual cinematography system that uses structure from motion to estimate the 3D camera motion and a set of 3D scene points, plans a new stabilized camera path, and generates the final video by novel view synthesis. This system is featured by the content-preserving warp that transforms each frame to follow the new camera motion while avoiding visual distortion. This feature makes our system work well with scene motion in a video. Content-Preserving Warps for 3D Video Stabilization Feng Liu, Michael Gleicher, Hailin Jin and Aseem Agarwala ACM SIGGRAPH 2009. PDF Website Youtube
	Re-cinematography: The first solution for achieving an idealized camera path from a shaky video in 2D. Re-cinematography: Improving the Camera Dynamics of Casual Video Michael Gleicher and Feng Liu ACM Multimedia 2007, pp.27-36. (best paper runner-up) PDF Website Re-cinematography: Improving the Camerawork of Casual Video Extended version of MM07 paper on ACM Transactions on Multimedia Computing, Communications, and Applications. Vol. 5, Issue 1, Oct. 2008:1-28. PDF
	Shadow removal: an approach for removing shadows in a photograph while requiring only rough user input and creating texture consistency between the shadow and lit areas. Texture-Consistent Shadow Removal Feng Liu and Michael Gleicher ECCV 2008 PDF Website
	Image and video super-resolution: an algorithm for image and video super resolution that aims to maximize the visual quality. Visual-Quality Optimizing Super Resolution Feng Liu, Jinjun Wang, Shenghuo Zhu, Michael Gleicher and Yihong Gong. Journal of Computer Graphics Forum. Vol. 28, Issue 1, 2009: 127-140. PDF