Method to generate frames on demand starts with a system receiving a request for a media content item from a client device. The request includes a media content identification and a main user identification. The system transmits to the client device a playlist including a first set of media content item segments. While the first set of media content item segments is being displayed on the client device, the system renders a second set of media content item segments using the media content identification and the main user identification. Rendering the second set of media content item segments can include rendering a main user avatar based on the main user identification and incorporating the main user avatar into the second set of media content item segments. The system then updates the playlist to include the second set of media content item segments. Other embodiments are disclosed herein.