Real-time participation within a media presentation is disclosed. A participant overlay may be generated by preprocessing a media presentation to extract a time-variant metadata stream associated with one or more interactive regions depicted in the media presentation. The participant overlay includes simulated physics, which may include boundaries of the interactive regions defined in accordance with content in said media presentation. A participant's interaction with said media presentation through the interactive regions may be captured and combined with the media presentation for display on a display. The interaction may comprise a user action upon one or more elements in the media presentation, a user action expressed in the media presentation through an avatar, or a user action in said media presentation in connection with a game element.