A user input indicating a request to preview a media item at a particular point in time in the media item is received from a client device and it is determined whether a length of a scene covering the particular point in time corresponds to a single preview or multiple previews. Responsive to the length of the scene corresponding to the single preview, the single preview is provided for display on the client device. Responsive to the length of the scene corresponding to the multiple previews, a portion of the scene covering the particular point in time is identified, and a preview associated with the identified portion of the scene is provided for display on the client device.