Every viewer a different cut: the rise of generative interactive film

Interactive film is an old dream with a short, uneven history. Laserdisc arcade games, choose-your-path DVDs and streaming specials all asked the same question: what if the audience could steer? And they all ran into the same wall. Every branch had to be written, cast, lit, shot and edited. Choice multiplied cost, so choice stayed small.

Generative models break that arithmetic. When scenes can be composed rather than captured, a story no longer has to decide in advance which branches are worth paying for. That makes interactive film possible at a depth that was never affordable before, and it raises a harder question: what does a film become when it is different for every viewer?

From branches to rivers

Classic interactive film is a tree. You reach a fork, you choose, you follow a pre-made path to the next fork. Viewers learn quickly that choices are few and endings are fixed, and they start to play the structure rather than the story.

Generative film behaves more like a river delta. The source is fixed: the world, the characters, the central dramatic question. The channels are not. Small choices shift the current, and the narrative finds its own way to the sea. Two viewers can share the same premise, the same cast and the same ending beats, and still remember genuinely different films.

A branching film offers you a menu. A generative film offers you a role.

The grammar we need

Cinema took decades to develop its grammar: the cut, the close-up, the reverse shot. Generative interactive film needs its own, and it is being invented now. A few principles we work to:

• Authored spine, generated flesh. The story’s structure, themes and key turning points are written by people. What happens between them can be composed live, in character, in response to the viewer.
• Continuity is sacred. Faces, places, costumes and tone must hold across every generated moment. Nothing breaks immersion faster than a character who looks different after a choice.
• Choice must carry consequence. If the viewer’s input could be ignored without changing anything, it should not be asked for. Meaningful choice is rarer, and better.
• Pacing beats possibility. Infinite options can stall a story. The best generative scenes still know where they are going and when to cut.

Text first, then image, then motion

The medium is arriving in layers. Text-led interactive stories, illustrated as they unfold, are already practical and already delightful; this is the space Montage works in today, where each scene is written live and illustrated in a consistent style. Image sequences with voice are close behind. Fully generated, real-time, continuous video that responds to the viewer is advancing quickly in research, but remains expensive and brittle at feature length.

That layering is a gift. It means creators can develop the craft of generative storytelling now, on mediums that work, and carry that craft forward as the visuals catch up.

What it means for creators

The fear is that generative film removes the filmmaker. Our experience is the opposite. The more a story can change, the more it needs a strong author: someone who knows what the story is about, who the characters really are, and which moments are non-negotiable. Generation scales the performance. Authorship gives it a reason to exist.

The future of film is not one cut for everyone. It is one vision, met differently by everyone who steps inside it.