Studio workflow / v1

From one sharp premise to one finished short.

The goal is not to automate taste away. It is to remove enough friction that the best idea gets a chance to become a finished piece.

1. Write the viewer’s first question

Start with a premise that creates an immediate gap: Why are the astronauts doing a trust fall? or What did the internet’s first meeting feel like? If the first sentence does not create curiosity, more visual polish will not rescue it.

2. Lock the visual brief

Choose the format before generating: vertical 9:16, one focal subject, a readable silhouette, and a lighting direction. Keep the brief short enough to repeat. We prefer a cinematic light recipe over a pile of adjectives.

3. Generate for editability

Make the frame serve the cut. Leave breathing room for captions, avoid tiny story details, and make the subject legible at phone size. Save the prompt and the chosen frame so the result can be explained later.

4. Add a human-paced voice

Write narration that sounds spoken, not like a museum placard. A 20-second short usually needs one hook, two or three beats, and a final turn. Silence is part of the edit.

5. Assemble and inspect on a phone-sized canvas

Build the vertical MP4, then inspect it at the size a viewer will actually see. Check the first two seconds, caption contrast, voice clarity, and whether the ending earns a replay.

Current limitation: generated imagery can fail on continuity, hands, text, and exact historical detail. We document those failures instead of presenting the pipeline as push-button filmmaking.

6. Archive the recipe

Keep the title, caption, prompt, source assets, audio, and final export together. The archive is what turns a one-off experiment into a studio practice.

Browse the archive →