How The Asking Was Made

The film is set in a drowned future where the living technology that once ran the planet has gone quiet. A boy carries the last working seed to the guardian who switched them off. The premise line was rewritten several times before any picture existed, because one verb decided whether a viewer would file the story under magic or under technology.

The narration started as a voice speaking to the audience. It felt like exposition, so we gave it a reason to exist inside the story: it became a record left by the character who shut the seeds down, which she finishes a century later when the boy arrives. The narration and the ending now pay each other off.

Before writing shots, we took two feature trailers we admired and measured them: where the cuts fall, where the dialogue sits, how the intensity rises and dips. Both shared the same shape, an early bump, two deliberate dips, then one long climb to a late peak. Our first rough cut had calm motion throughout. The numbers showed us the gap more clearly than our memory of those trailers did.

Every character was designed and fixed in reference images before production, along with the main locations, so the same faces and the same flooded city hold from the first frame to the last.

Short shots leave little room for words. In one five-second hand-off, a powerful man gives the seed to a courier, and the line had to explain both that the seed obeys him and that it will not stay with him. The dialogue went through several rounds until it was one short sentence each, and the prop did the rest: the seed flickers uneasily in his hand and settles in the courier's.

We showed the cut to people who knew nothing about it and wrote down what they thought had happened. Then we treated every misread as a fault in the film, not in the viewer.

• An older woman speaking straight to the lens was read as the boy's grandmother. We remade the scene as a conversation, over the shoulder, with the boy given a line of his own.
• A shot of two children on a beach, meant as anonymous witnesses, was read as the two leads. We replaced the foreground with a group of villagers of different heights and ages, so it could not be mistaken for anyone in particular.
• The boy's wish came before a storm, so the storm read as the wish failing. Moving the storm in front of the wish fixed it without making a single new shot.
• A courier appeared several times just after the boy's first scene and was taken for the boy as an adult. Cutting the courier's extra appearances solved it.
• Several generated clips came back with voices saying things nobody wrote. Every line of audio in every clip is now checked before it reaches the edit.

Most fixes were edits, reorders or replacements from footage we already had. Only a few needed new generation.

The first grade matched each shot to a reference on its own, and shots in the same place ended up looking different. We regraded by scene instead, removed the vignette, and pulled back the glow on the brightest shots. The film started to read as one piece.

Narration was placed by listening to where words actually end, not by trusting automatic timings, and the whole mix was checked line by line once assembled to prove nothing had been clipped.

The trailer opens with a rapid burst of cuts and spends its best image in the first seconds, before a viewer knows what they are looking at. That trick works for a film audiences already know. An unknown film needs context before speed. And the final export sat quieter than platform level while touching full scale at its peaks. Both lessons are now part of how we finish every film.

Watch it on its work page.