Story to Screen

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Story to Screen

Inspiration

Every film starts as a rough idea in someone's head, but turning that idea into something a production team can actually work from — structure, characters, shots — takes time and experience most first-time storytellers don't have. We wanted an agent that acts like a junior script supervisor: take a raw premise and hand back a real starting blueprint.

What it does

You type a short story premise. The agent works through it in three chained steps: it breaks the story into a three-act structure, identifies the core cast of characters with quick descriptions, and generates a shot-by-shot storyboard — each shot tagged with a mood/tone to guide visuals or music. The whole pipeline runs from a single input and returns a complete, structured production starting point.

How we built it

The app is a single-page HTML/CSS/JS interface calling the Gemini API directly from the browser. We prompt Gemini to reason through the story in stages — structure, then characters, then shot list — and return everything as structured JSON, which we render into a storyboard-style interface designed to feel like a director's notebook rather than a generic AI output screen.

Challenges we ran into

Getting the model to stay consistent across all three outputs (acts, cast, shots) so they actually reflect the same story, rather than drifting into generic tropes, took some prompt iteration. We also wanted the output structured enough to render cleanly without needing a backend.

Accomplishments that we're proud of

A working, no-backend, agentic pipeline that goes from a blank page to a usable film blueprint in one click — and an interface that actually feels like part of the filmmaking process, not just a chatbot.

What we learned

Small, well-scoped agent pipelines (three clear steps, one clear output shape) are far more reliable and demo-friendly than trying to make one prompt do everything.

What's next for Story to Screen

Letting users regenerate individual sections (just the shot list, just the cast), export the blueprint as a PDF "leave-behind," and add reference images per shot using Google's image generation tools.

Built With

• ai
• api
• css
• gemini
• google
• html
• javascript
• studio