The Question That Starts Everything
In animation you learn quickly that the idea comes before everything else. Before the rig, before the layout, before anyone asks how long it will take. You protect the idea in its early form because once production logic arrives, the fragile things get cut first.
AI filmmaking is the same. Before a single prompt is written, before a character is designed, before the story is structured — there is a question. It costs nothing to ask. It takes seconds. And it contains everything that follows.
What if?
The three films on this site each started with one:
- What if a research station had been abandoned for twenty years — and a scientist's daughter was the first person back? That question, before any tool was opened, contained the story, the mood, the conflict, and the ending of The Last Analog. Everything else was answering it.
- What if a product film was completely sincere — but the product didn't exist? The Summit Rest System started there. The €247 ultralight sitting pad, the technical drawings, the voiceover describing thermoformed comfort geometry — all of it came from holding that question long enough to answer it properly.
- What if nothing happens — and that's the whole film? In Between is a film about Tuesday. No journey, no transformation. Just a day. That question was the film.
Protect the Question
"What if?" has to be protected from "Can I?" for long enough to actually generate ideas. The moment "Can I?" enters the room, "What if?" sits down.
They are not enemies. They are sequential. "Can I?" is essential — it is where craft lives, where the technical work earns its place. But it comes second. Always second.
The failure mode is letting "Can I?" arrive too early. A good idea dismissed before it's fully formed. A direction abandoned because the first technical objection surfaced before the creative possibility was understood. The question deserves time to breathe before it faces scrutiny.
Ask "What if?" freely. Answer it completely. Then — and only then — ask "Can I?"
It Works at Every Scale
The question doesn't only apply at the story level. It applies at every level of the work simultaneously — and the answers at each level feed the others.
Story level: What if the whole film takes place in one room? What if we never see the antagonist's face? What if the ending is the beginning?
Scene level: What if these two characters never make eye contact? What if this scene has no dialogue at all? What if the most important thing in the scene is an object, not a person?
Shot level: What if the camera is on the floor? What if the only light is what's already in the room? What if we cut before the action completes?
Prompt level: What if the haze is twice as thick as usual? What if the character is looking the wrong way? What if this insert shot is the whole shot?
A shot-level "What if?" can rewrite a scene. A scene-level "What if?" can reframe the story. Ask the question at every level before locking anything.
AI Changes the Stakes
In animation — or any medium with real production cost — "What if?" is expensive to answer. Building a set, lighting it, shooting it, animating it: every answer carries a price tag that makes the question feel dangerous. You get cautious. You stop asking the wilder ones.
In AI filmmaking, the iteration cost of a "What if?" is almost nothing. Ask it. Generate it. Look at it. The answer is in front of you in seconds. This changes the relationship with the question entirely. You can ask more of them. Further out. Wilder. Earlier. Before committing to anything.
The tool doesn't judge the question — it just answers it.
The discipline that matters in AI filmmaking is not restraint in asking "What if?" — it is knowing which answers to keep.
Write Them Down
Keep a running document. Not a shot list, not a production document — just a place where "What if?" questions live without obligation. Some will become shots. Some will become scenes. Some will sit there for three projects and then suddenly become the thing that makes a fourth project work.
The question that felt technically impossible today is the one the tools will be able to answer next year.
"What if?" is the only question with no wrong answer. Every other question in production has constraints. This one doesn't. Guard it.