There is a moment in every project where the client says yes. The question that decides the whole production is: yes to what, exactly?

In traditional filmmaking that question took decades to answer, and the answer became pre-production. In AI filmmaking, where generation feels instant, teams are quietly skipping the answer again. They show a storyboard, get an approval, generate the film, and then discover the client approved a set of pictures, not a film. The film has rhythm, and nobody ever showed them rhythm.

After enough of those discoveries, our studio adopted a hard rule: storyboards ship with an animatic. Always. One without the other lets the client approve something they never saw.

What an AI storyboard actually proves

The storyboard proves framing. That is its entire jurisdiction, and within it, it is irreplaceable.

A board answers the spatial questions: what is in the frame, where the subject sits, how close the camera is, where the light comes from, what the world looks like. In AI video pre-production this matters even more than on a live set, because the frames on the board are not sketches of what a camera might capture. They are often the literal anchor images, the start frames and references, that will be fed to the engine. Approving the board is approving the pixels the film will be born from.

That makes the modern AI storyboard stronger than its paper ancestor: it is not a promise of the image, it frequently is the image. When a client signs off on frame three, frame three exists.

But here is what the board cannot prove, no matter how finished it looks: it cannot prove time. A board has no duration. Every viewer reads it at their own pace, imagines their own cut, hears their own music. Twelve people approving the same storyboard have approved twelve different films.

What the animatic proves that the board cannot

The animatic proves rhythm. It is the board placed on a timeline: the frames cut to real durations, in order, usually with temp sound or a music reference underneath.

Suddenly the invisible decisions become visible. That contemplative opening the client loved as a still turns out to hold for six seconds, and now they feel the six seconds. The product reveal lands on the music, or does not. The three quick beats in the middle read as energy, or as panic. Duration is a creative decision in every single shot, and the animatic is the only document where that decision can be seen, felt, and disputed before it costs generation money.

In AI production this is not a luxury pass. Engines bill by the second with minimum durations per clip, and pacing decisions determine how many shots exist, how long each runs, and which ones can fuse into a single longer take. The animatic is where those calls get made deliberately. A rhythm problem caught in the animatic costs a timeline edit; caught after generation, it costs a regeneration cycle and a bruised schedule.

Board plus animatic: the full contract of AI video pre-production

The board says: this is what you will see. The animatic says: this is how it will feel to watch.

Together the two documents form a complete approval. A client who has signed both has genuinely approved the film. A client who has signed only the board has approved a gallery.

Our working sequence in the lab and on client work:

  • Direct the key frames first. These stills carry the world, the light, the brand. They are built with intention because many of them become the actual anchors for generation.
  • Assemble the board from those frames and resolve every framing dispute here, where a change is a redrawn still.
  • Cut the animatic from the approved frames, with real durations and a sound reference. Resolve every rhythm dispute here, where a change is an edit.
  • Only then generate. The engine receives anchors that have survived two rounds of honest scrutiny.

Notice what this sequence refuses: it refuses to let generation be the first time anyone experiences the film in time.

The anti-shortcut argument

The seduction of AI video is speed, and the seduction whispers: why board at all, just generate and react. We have tested that road too, and it is slower. Reacting to finished generations means paying full generation cost for every draft of a decision that a still or a timeline could have drafted for cents. Iteration is cheapest at the lowest fidelity that can carry the decision: framing decisions live in stills, rhythm decisions live in the animatic, and only surface decisions, motion quality, texture, the life of the shot, genuinely require generation.

That is the Sentimagem discipline applied to process, direction over generation: settle the film as frames before you spend on motion, settle it as rhythm before you spend on rendering, and deliver the finished piece knowing the client already met it twice.

The board proves framing. The animatic proves rhythm. A film is both, so an approval must be both. Everything else is asking your client to sign a contract with a missing page.

In AI filmmaking the storyboard proves framing and the animatic proves rhythm; a film is both, so a real client approval must be both.