As many technologies related to the broadcasting have been developed, the demand for multimedia services which can provide more immersive and realistic sense has rapidly increased. The multi-view video is considered as the strongest candidate to satisfy this demand. It is a set of image sequences of the same scene captured by more than three cameras. 3DTV uses this multi-view video as an input. More than one viewpoint displayed through a stereo monitor or a multi-view display can provide 3D sense to users using the disparity between the adjacent viewpoints. If the number of viewpoint is increased, we can watch more immersive and realistic video. To increase the number of viewpoint, we synthesis an intermediate viewpoint from adjacent view images based on depth information of the scene. Depth information is measured by an active depth sensor, or estimated by using passive method such as stereo matching. Our research on 3DTV covers capturing multi-view video/depth, preprocessing, depth estimation, and intermediate view generation.