AnimPrompt AADS V25.6
Contest demo Open generator

JUDGE WALKTHROUGH01 / 03

From scene idea
to performance direction.

AnimPrompt turns an animation scene brief into a renderer-ready prompt. Google Gemini reasons about the direction; application code assembles the final prompt and inserts protected dialogue exactly once.

THE FASTEST TEST

Run the preloaded Digi shot.

No setup needed. The brief and exact dialogue are already filled in.

01

Open the generator

The live form is the home page. Its starting brief asks Digi to press a lab-console switch, react through a full-body chain, and settle into a living hold.

Go to the preloaded Digi example
READY TO RUN
02

Keep the defaults, or tune a shot

Optional controls let you change visual, directing, and performance styles; action, emotional, and performance scales; duration; aspect ratio; mood; and target engine.

Default setup: Semi-Stylized · Naturalistic · Natural · Moderate action · High emotion · High performance · 10 seconds · 9:16.

OPTIONAL
03

Initialize sequence

Submit the form with Initialize sequence. The app sends the brief and selected controls for generation, then assembles the returned direction into a production prompt.

GEMINI → AADS
04

Review, then take it with you

Read the generated prompt in the Production prompt panel. With a result available, use Copy or Download to take the text out of the demo.

OUTPUT

WHAT THE DEMO DOES

Direction with clear boundaries.

One connected model

Google Gemini is the reasoning model used for generation. The target-engine menu shapes the renderer-facing prompt; it does not switch the connected model.

Google Gemini Video is a target

Selecting Google Gemini Video describes the intended renderer in the output format. It is not a separate Gemini connection or a video-rendering action in this demo.

Dialogue is assembled in code

The model is instructed not to include protected dialogue. Application code adds the supplied line once after Gemini returns its direction.

Holds keep their subject

The direction distinguishes a living character hold from a locked camera, stable object contact, or fixed environment—so one hold does not imply all motion stops.

LOOK AT THE MOTION

Compare direction with animation references.

Open sample animations