Given two video sequences of different scenes, the problem of seamlessly transferring a 3D moving object from one sequence to the other is a complex task. In this paper, we present a method to extract the alpha matte of a moving 3D object from a source video, and then correctly augment the object into another target video. Our framework builds upon techniques in natural image and video matting, composition, and image-based rendering. Natural image matting is usually composed of: foreground and background color estimating and alpha estimating. Our proposed technique uses local and global color information to estimate the accurate alpha values and extends this approach to extract a moving object from a video sequence. From a video sequence consisting of a single moving object over a stationary background, we combine motion statistics, color and contrast cues to extract a foreground (moving) object efficiently with no interaction from the user. During the process of compositing the moving object (from source video) on the target video, our approach composes the object accurately and ensures that the implanted moving object does not collide with other objects (static or moving) already present. An intuitive user interface (UI) tool is designed and implemented to provide flexible control and facilitate 3D composition for the user. © 2008 ACM.