StarZero

Your cast, your lines, shot for vertical.

Write the scene with exact dialogue and give the agent reference images for the characters and the setting. It generates the scene in continued segments, checks each take, and assembles one 9:16 episode.

A pig this good. From a script and three reference images, generated in three segments.

Keep the cast for the next episode.

A series needs the same faces, places and voices every time. Brief each new episode against the same references and voice notes, and the agent carries them through.

Cast and setting
Reference images for each character and location, attached to every generation.
Dialogue
Exact lines, with each voice described once. For a voice that must never change, lay generated voiceover with one fixed voice over the picture.
Length
3 to 10 seconds per generation, continued up to 40, then cut together.
Finish
Captions, music, sound effects and graphics on the same 9:16 timeline.

Each segment starts where the last one ended.

The agent drew the key poses as stills from the three reference images, then generated each segment from one still to the next. It watched and listened to every segment before going on, and the cuts land on matching frames.

  1. The capuchin farmer on his stool beside the pig, the reporter's microphone in frame

    Opening frame

    Segment 1 opens here: “Why does your pig have a wooden leg?”

  2. The farmer turns to the pig with both hands open

    The gesture

    Drawn from the references. Segment 1 ends on it, and segments 2 and 3 start from it.

  3. The farmer points at the pig with a straight face

    The point

    Segment 3 ends here, on the punchline: “A pig this good, you don’t eat all at once.”

This is the third cut of the scene. See the first two and the notes that changed them.

What happens in episode one?

Bring the premise and the cast. Planning a whole series? Talk it through with us first.