Veo 3 prompt guide: how to write prompts that work (with examples)

Veo 3 rewards specific prompts. The same idea written vaguely and written well can look like two different models. This guide gives you a formula that works, the words that make the biggest difference, and examples you can copy straight into the text to video generator.

The Veo 3 prompt formula

Write your prompt in this order. You do not need every part every time, but the order helps the model understand what matters most.

1. Subject: who or what is on screen, with two or three visual details.
2. Action: one clear thing that happens.
3. Setting: where, and when (time of day, weather).
4. Camera: shot size and movement.
5. Light and style: the look, such as golden hour, neon, documentary, 35mm film.
6. Audio: dialogue in quotes, sound effects, ambience, music.

Example:

Adding dialogue and sound

Veo 3 generates audio with the video, which is its biggest advantage. To use it well:

• Put speech in quotes and say who says it: The man turns to the camera and says, "You have to try this."
• Keep lines short. An 8-second clip fits about one or two short sentences.
• Describe the voice if it matters: calm, excited, whispering, British accent.
• Name the ambience and effects: rain on a window, crowd noise, waves, keyboard typing.
• Ask for no music if you will add your own later.

Camera words that work

Use one camera move per clip. Combining several moves in 8 seconds usually looks messy.

Lighting and style words

• Natural: golden hour, overcast daylight, soft window light
• Moody: neon reflections, low-key lighting, candlelight
• Commercial: bright studio lighting, clean white background
• Film looks: 35mm film grain, anamorphic lens flare, documentary style
• Social: vertical phone footage, UGC style, natural and unpolished

12 example prompts

UGC and ads

1. Vertical phone video. A woman in her 20s in a sunny kitchen holds a green smoothie to the camera and says, "Three weeks of this and I'm never going back." Handheld, natural light, casual and authentic.
2. Close-up of hands opening a matte black skincare box on a marble counter, slow push-in, soft studio light, satisfying unboxing sounds.
3. A man jogging at sunrise along a beach, tracking shot from the side, golden light, waves and steady breathing, he glances at his smartwatch and smiles.

Cinematic

4. Wide shot of a lone astronaut walking across red desert dunes, slow pull-back, dust in the wind, deep rumbling ambience, epic and quiet.
5. Rain-soaked Tokyo alley at night, neon signs reflecting in puddles, a woman in a red coat walks toward the camera, tracking shot, cinematic 35mm look, rain and distant traffic.
6. Close-up of an old fisherman's weathered face in a storm, low-angle, cold blue light, crashing waves, he says quietly, "We're going back out."

Social and fun

7. A golden retriever wearing sunglasses rides a skateboard down a sunny boardwalk, tracking shot, upbeat summer vibe, wheels rolling on wood.
8. Static shot of a tiny cat chef stirring a pot in a miniature kitchen, warm light, cozy, bubbling sounds.
9. A grandmother looks at the camera and says, "Back in my day we didn't need Wi-Fi," then winks, living room, handheld, natural light.

Product and B-roll

10. Macro shot of coffee beans falling in slow motion onto a wooden table, warm side light, crisp sound of beans landing.
11. A sleek white sneaker rotating on a floating platform, slow orbit, clean studio lighting, soft whoosh.
12. Aerial drone shot over a turquoise lagoon at sunrise, slow forward movement, gentle wind and birds.

Common mistakes

• Too many actions. "She wakes up, makes coffee, drives to work and gives a talk" is four clips, not one.
• Conflicting styles. "Anime, photorealistic, black and white neon" confuses the model.
• Long dialogue. A paragraph of speech will not fit in 8 seconds.
• No camera direction. Without it you get a generic medium shot.

Veo 3

Google's most advanced AI video model with cinematic quality

Try Veo 3

Save credits while you iterate

Draft with Veo 3 Fast at 720p and 4 or 6 seconds until the prompt is right, then render the final with Veo 3 at 1080p. If the shot needs a specific person or product, generate a still with text to image and animate it with image to video instead.

Frequently asked questions