The core to many 3-D related applications is establishing reliable dense correspondences among the multiple cameras used in a given imaging system. This paper proposes a structured light scanning technique for directly solving the difficult multiframe correspondence problem across any number of cameras without any searching or calibration. The technique consists of temporally encoding positional information of projector space and capturing the projected result with one or more fixed cameras. In contrast to traditional image-based approaches, the proposed technique does not rely on consistent textures across frames. Furthermore, it automatically determines visibility across all cameras in the system and scales linearly in computation with the number of cameras. The proposed technique forms the core of a complete framework to synthesize new views of a given 3-D scene at interactive rates. An efficient interpolation method using barycentric coordinates renders new views at interactive rates from the computed correspondences. The light projector is shown to serve as an additional viewpoint for interpolation, thereby expanding the range of synthesizable views. The paper also creates an intuitive user interface for specifying the desired virtual view in the absence of any 3-D geometry. Experimental results using real- world data demonstrate the effectiveness of the proposed framework for creating interactive 3-D media. 25 Pages