For most of film history, the audience’s job was to sit still. A camera found a face, held it, and asked you to feel something. The close-up remains the most intimate device cinema ever invented. What it could not do was let you answer.
Interactive film has spent decades trying to solve the other half of that problem. In 1967, Radúz Činčera’s Kinoautomat let a theater vote on what happened next. More recently, Black Mirror: Bandersnatch turned the remote into a set of forks in the plot. In each case the viewer remained a selector. You chose a path. You did not walk into the room and speak to the person standing in it.
Something quieter has been happening among people who never thought of themselves as part of that experiment. They opened an app to talk to a character — often first as an AI companion, an invented person with a voice and a temperament — because a single face is something you can already hold a conversation with. That habit looks, from a distance, like a sideline to cinema. It is closer to the missing piece. The path toward interactive video that can actually reach an audience at consumer scale does not begin with a generated city. It begins with a character you can interrupt mid-sentence.
A fully realized world, generated as you move through it, is still too slow and too costly to stream to millions. A face, by contrast, already works at consumer scale. Real-time avatars are the wedge into real-time video: once a person will sit with a character and talk, the rest of the frame can be generated, piece by piece, without asking the audience to wait for a future that is not ready yet. The sequence is simple, and it is the one Prelulu has been building toward. First the face. Then the cast. Then the scene. Then the world.
The first step is a live conversation. On a video call, an AI character answers in its own voice, with expression, on an avatar model built for that purpose. You can cut in the way you would cut into a real conversation. Many people arrive here looking for company. That is not a lesser use of the form. It is how a mass audience first learns to treat a figure on a screen as someone who can hear them. From there, the relationship expands. Users create their own characters and assemble a cast, the way a director might, except the cast can look back. And then the story itself stops being something you only read or watch. Key beats become scenes you step into — not as a branching menu, but as a room you can move through and a person you can talk to face to face.
You can already see where that leads. In early November, Prelulu shared a collaboration preview with Hy-Nomas, Hunyuan’s interactive world-generation model, built around a short interactive microdrama. The demonstration is brief, and that is part of its force. Sienna answers a question about a gala, face to face, the conversation visible beneath her live portrait. The view travels through the party to Evelyn, who answers in character and tells you to lower your voice before the guests notice. Walter is holding a folder; asked about the paperwork, he hints that the real books are somewhere else in the room. Movement through a scene, and conversation with the people inside it, happen in the same continuous gesture. The explorable views are generated by Hy-Nomas; the conversations carry voice, lip-sync, and a reply.
The playable story behind that preview is Before Nine, an interactive revenge drama. You take the role of Nora Quinn at her husband’s gala. Grant intends to divorce her in front of the guests. He does not know that she owns more of his company than he does. Around her stand an assistant, a mother, a controller, a private-equity buyer, a fixer — each with a different motive, each available as someone you can approach and speak with. The night is designed to last about ten minutes and opens onto seven endings. What matters for cinema is not the plot summary. It is the change in the viewer’s job. You are no longer choosing door A or door B from the couch. You are in the room.
This is the evolution that looks, in hindsight, almost inevitable. Entertainment has been moving for years from watching toward participating. Roleplay and story scenes are no longer a sketch of that idea; they are live. People still begin, in large numbers, with a single character and the private feeling of companionship, because that is the scale at which the technology already works. From that face, a cast gathers. From the cast, a scene. From the scene, eventually, a world. The last of those is not something you can ship to everyone tomorrow. You can, however, see it — in a conversation that can be interrupted, in a gala you can walk across, in a character who answers when you ask about the folder in her hands.
Cinema taught us how to look at a face until it felt close enough to touch. The next move is to let that face hear you, and to build, at consumer scale, the kind of story that can answer back. The future of interactive video will not arrive all at once as a finished world. It is arriving the way movies themselves arrived: first through a close-up that made someone on a screen feel present. This time, you can talk to them.
