A method, system and computer program provide a mechanism for smoothing the transition back from a virtual (computer generated) scene to a related video stream. An event such as a user input or timeout is received triggering a return to display of the video stream from a virtual scene related to content of the video stream. A number of time points and/or camera angles are either presented to the user or are automatically searched for the best match. The list may be presented in order according to an automatically detected matching criteria. The virtual scene may a scene constructed locally within a computer or digital video recorder (DVR) and the matching performed locally based on angle and time information provided from a content provider such as a server, or the virtual scene generation and matching may be performed at a remote location such as the content server.