The Formula
Cinematography + Subject + Action + Context + Style and ambiance (Google's published structure for its video models).
- Cinematography
- Shot size, angle and camera movement, stated first.
- Subject
- The main character or focal point.
- Action
- What the subject does in this one continuous moment.
- Context
- The environment and background elements.
- Style and ambiance
- The look, mood and lighting, plus the sound: ambience, effects and any spoken line.
The Rules
Open with the camera
Start with shot size, angle and movement, then the subject. That is the first slot in Google's structure.
Always describe the sound
Sound is always on and cannot be switched off. Name the ambience and the sound effects tied to visible actions, or the model fills the track itself.
Say "no dialogue" when nobody speaks
Without it, the model may invent a line to go with the always-on audio.
Quote the line and name the speaker
Give the speaker and delivery, then the exact words in quotation marks, for example: She says quietly: "Not tonight."
Keep to one continuous moment
Write one take with no cuts. Clips run 3 to 10 seconds, so one action and one line is plenty.
Phrase everything positively
This model takes no negative prompt and no seed, so describe what you want rather than what to avoid.
Frame for 16:9 or 9:16
Those are the only two aspect ratios, so compose the action for wide or vertical framing.
A Template You Can Copy
[Shot size, angle and camera movement] of [subject] [action] in [context]. [Style, light and ambiance]. [Speaker] says [delivery]: "[exact words]" (or: no dialogue). [Ambience and sound effects tied to visible actions].
A Real Example
The shot: Night, rain-streaked diner window. A woman in a red coat sits alone in the last booth, turning a ring on the table. Slow push in from the window. Mara says: "You kept it." (6 seconds, 16:9)
A slow push-in shot moves through a rain-streaked diner window toward a woman in a red coat sitting alone in the last booth at night. She slowly turns a ring between her fingers on the table, eyes lowered, as warm amber light from the diner interior contrasts with the cold, rain-blurred night outside. As the camera nears, she lifts her eyes and says quietly: "You kept it." Soft rain ticks against the glass, a low diner hum and distant clink of cups fill the quiet space, with a faint, melancholic ambient tone underneath.
It opens with the cinematography (a slow push-in through the window), then the subject, her action, the context and the warm-against-cold light. The line is quoted with the speaker's delivery, and the sound track is spelled out (rain on glass, diner hum, a faint ambient tone) because this model always generates audio.
Dialogue
There is no special syntax: name the speaker and delivery, then put the exact words in quotation marks. Because sound is always generated, write "no dialogue" for shots where nobody should speak, and keep the line short enough to finish inside a 3 to 10 second clip.
She says quietly: "Not tonight."
Camera Moves
| Move | How to write it for Gemini Omni 1.1 Flash |
|---|---|
| Slow push in | A slow push-in shot moves through the rain-streaked window toward her. |
| Static medium shot | Medium shot, static camera, a tired worker rubbing his temples at his desk. |
| Tracking | Low-angle tracking shot follows the runner along the wet street. |
| Handheld close-up | Handheld close-up on his hands as he counts the coins. |
| Slow pan | Wide shot, the camera slowly pans across the empty platform to the last passenger. |
Common Mistakes
| Mistake | Fix |
|---|---|
| Saying nothing about sound. | Sound is always generated. Describe ambience and effects, and write "no dialogue" if nobody speaks. |
| Writing a negative prompt such as "no blur". | This model takes no negative prompt; describe the result you want instead. |
| Describing cuts or several shots. | Keep to one continuous moment per clip. |
| Planning a 1:1 or 4:3 shot. | Only 16:9 and 9:16 are available; frame for one of them. |
| A long speech in a short clip. | Clips top out at 10 seconds; keep to one short line. |
Length, Resolution and Price
Gemini Omni 1.1 Flash makes 3 to 10 second clips in 16:9 or 9:16, with sound always on, spoken dialogue and an end frame. It costs $0.03 per second at 360p, $0.10 at 720p, $0.15 at 1080p and $0.30 at 4K, and takes no seed or negative prompt (checked 2026-10-22). Every model side by side is on the model spec sheet.