Veo 3 Prompt Generator: Ready-to-Paste Prompts for Veo 3.1

Veo 3.1 gives you 8 seconds and a soundtrack. Most wasted Veo credits come from prompts written for a silent model: no audio direction, and three actions squeezed into a clip with room for one. This Veo 3 prompt generator page gives you Google’s recommended formula, nine prompts to paste right now, and the dialogue and sound syntax Veo listens to.

What Veo 3.1 Is Good At Right Now (and Where It Stops)

As of September 2026, from Google’s documentation and pricing pages:

• Clip length: 4, 6, or 8 seconds per generation. Longer pieces are built by extending or stitching clips.
• Resolution: 720p or 1080p, with 4K output available on the Gemini API at a higher per-second price.
• Aspect ratio: 16:9 or 9:16. Vertical works for Shorts and Reels, including Ingredients to Video since Google’s January 2026 update.
• Native audio: dialogue, sound effects, and ambient noise generated with the picture, in sync. This is still Veo’s biggest advantage.
• Inputs: text, reference images (Ingredients to Video), and first-and-last-frame transitions.
• Price on the Gemini API: $0.40 per second for Veo 3.1, $0.10 per second for Veo 3.1 Fast at 720p, and $0.05 per second for Veo 3.1 Lite at 720p.
• Watermark: every clip carries Google’s SynthID marker.

The limits matter for how you write. Eight seconds holds one action and one payoff, or four quick timestamped beats. Dialogue has to fit in a breath. If you need 15 to 30 seconds in a single pass, the Kling, Seedance and Hailuo pages cover models that go longer.

The Veo 3.1 Prompt Formula

Google’s Veo 3.1 prompting guide puts the camera first, then the subject, then the action, then the setting, then style. Audio goes on its own labeled lines.

[Cinematography: shot size, angle, camera move]
+ [Subject: who or what, with 2-3 visual details]
+ [Action: one clear action with a payoff moment]
+ [Context: place, time of day, light source]
+ [Style & ambiance: film stock, grade, texture]
Dialogue: Speaker says, "One short line."
SFX: one or two specific sounds tied to the action
Ambient noise: the room or street underneath it all

Two rules matter most. Put every spoken word in quotes and attribute it to a speaker. And label your sounds, so Veo does not fill the gap with generic background music.

9 Copy-Ready Veo 3.1 Prompts

Doorbell camera: package thief raccoon

Doorbell camera footage, fisheye lens, high angle looking down at a front porch at night, infrared black-and-white image. A fat raccoon waddles up the steps, grabs a small cardboard package with both paws, then freezes and stares straight into the camera when the porch light clicks on. It drags the box off the porch anyway. Slight fisheye distortion, compression artifacts, low frame rate feel. SFX: porch light click, cardboard scraping on wood. Ambient noise: crickets, a distant car passing.

Why it works: the freeze-and-stare beat is a single payoff that fits inside 8 seconds, and labeled sounds keep Veo from scoring it with music. More in this style on the doorbell camera prompts page.

CCTV: the cat who knows

Static CCTV security camera mounted in the top corner of a small kitchen, wide angle, grainy daytime footage with washed-out colors. A ginger cat walks along the counter, stops next to a full glass of water, looks back at the camera, and slowly pushes the glass off the edge. The glass shatters on the tile floor and the cat does not flinch. A woman's voice off-camera says, "Mango, I saw that." Static surveillance framing, slight lens distortion. SFX: glass shattering. Ambient noise: fridge hum.

Why it works: an off-camera voice gives you dialogue without a visible mouth, so there is no lip-sync to get wrong. See the CCTV prompts page for more setups.

Handheld phone fail: black ice vs trash day

Handheld phone video, vertical, slightly shaky, filmed by a friend standing on a snowy driveway on an overcast morning. A man in shorts and a puffer jacket drags a trash bin toward the street, steps onto a patch of black ice, and slides slowly with the bin until both tip over into a snowbank. The friend filming laughs and says, "Every single year." Phone auto-exposure adjusting, flat winter light. SFX: plastic bin wheels scraping on ice, a soft thud into snow.

Why it works: a slow slide keeps the motion readable, and the cameraman’s line lands after the fall, the way real fail clips are timed.

Street interview with two speakers

Handheld street interview on a busy Mumbai sidewalk at golden hour, medium close-up. A young man holds a small microphone toward an elderly shopkeeper in a white kurta standing in front of his tea stall. The interviewer asks, "What's the secret to good chai?" The shopkeeper smiles, taps his chest and says, "Patience. And too much sugar." Shallow depth of field, traffic blurred behind them. Ambient noise: scooter horns, street chatter.

Why it works: two short quoted lines with clear speaker attribution is exactly the pattern Google documents for multi-person dialogue. Browse street interview prompts for more formats.

Cinematic: lighthouse in a storm

Slow dolly-in, low angle. A lighthouse keeper in a yellow oilskin coat climbs the last steps of a spiral iron staircase at night, lantern in hand. He reaches the top, shoulders open a heavy door, and wind blasts rain across his face. Cold blue moonlight through the windows, warm lantern glow on his beard. Shot on 35mm film, heavy grain, desaturated teal grade. SFX: door hinge groan, rain hitting glass. Ambient noise: howling wind, waves breaking far below.

Why it works: camera-first ordering follows Google’s formula, and one continuous action fills the clip without a cut.

Product ad with voiceover

Macro close-up, slow orbit around a matte black wireless earbud case resting on wet black stone. A single drop of water falls onto the lid and splashes outward in slow motion as the lid snaps open, revealing earbuds with a soft white LED glow. Hard rim light from behind, deep shadows, glossy highlights on the water. A calm male voiceover says, "Built for the rain." SFX: a water drop, a magnetic lid click.

Why it works: a voiceover needs no on-screen speaker, and the lid click lands on the reveal.

Four beats with timestamps

[00:00-00:02] Close-up of a street vendor's hands cracking two eggs onto a hot griddle at night. SFX: loud sizzle.
[00:02-00:04] Medium shot, he flips a paratha into the air with a spatula and catches it on the griddle.
[00:04-00:06] Over-the-shoulder shot, he folds the paratha around the egg and wraps it in newspaper.
[00:06-00:08] Wide shot of the crowded night market, steam rising under strings of bare bulbs. Ambient noise: crowd chatter, pans clanging.

Why it works: Veo’s timestamp syntax scripts four cuts inside one 8-second clip, ideal for food and process content.

Ingredients to Video: mascot plus product

Using the reference images of the orange cartoon fox mascot and the blue water bottle: the fox skateboards down a sunny wooden boardwalk, kicks the board up into its paw, stops, and takes a long drink from the blue bottle. Vertical 9:16, tracking shot at the fox's eye level, bright saturated 3D animation style. The fox says, "Okay. Now I'm ready." SFX: skateboard wheels rolling over wooden planks.

Why it works: the references hold the mascot and product steady, so the text only covers action, camera and one line.

First and last frame: outfit change

The camera holds a static wide shot of a bright bedroom. The woman from the first frame, sitting on her bed in pajamas with a coffee mug, stands up, spins once, and mid-spin the outfit blurs into the street clothes from the last frame as she lands facing the door. Warm morning window light. SFX: a quick whoosh on the spin. Ambient noise: soft room tone, birds outside.

Why it works: first-and-last-frame mode builds the transition, so the prompt only describes the connecting motion.

Veo-Specific Tips That Save Credits

• Say exclusions as descriptions. Google’s guide recommends “a desolate landscape with no buildings or roads” over a bare “no man-made structures.” The negative prompts guide has more examples.
• Set aspect ratio in the settings. Writing “vertical” helps composition, but the 9:16 toggle is what actually changes the frame.
• Draft on Fast or Lite. At $0.05 to $0.10 per second, you can test a dozen prompt variations for the price of two or three Standard clips, then rerun the winner on Veo 3.1.
• Use timestamps only when you need cuts. A single continuous action reads fine as plain prose.
• Name the camera move, once. “Slow dolly-in” beats “cinematic camera movement.” The camera movement cheatsheet lists terms that translate well.

Where Gemini Omni Flash Fits

Google also has a newer video model, Gemini Omni Flash, in preview on the Gemini API. Google describes it as fast video generation, editing, keyframe interpolation and extension with native audio. It is a separate line, not Veo 4, built around refining a clip with follow-up instructions. The prompts on this page carry over as-is, since the camera-first formula and labeled audio lines are plain text.

From One Idea to 10 Veo Prompts With LzyPrompt

Writing 50 prompts like these for a month of Shorts takes a weekend. That is the part LzyPrompt handles.

Type the idea in plain words, like “raccoon steals a package, caught on doorbell cam.” Pick a Camera Style (doorbell camera, CCTV, handheld phone video, dashcam POV and more) or leave Auto-detect on, and optionally set Shot Type, Lighting and Aesthetic. You get 10 variations, each with a scene title, a full prompt with setting, lighting, action beats, imperfections and sound cues, and a ready-to-post caption.

LzyPrompt has no Veo integration or model picker. You copy a prompt and paste it into Veo yourself. Before you hit generate, do three quick Veo edits: move the sound cues onto SFX: and Ambient noise: lines, add one quoted line if you want speech, and cut any beat that will not fit in 8 seconds. If you are batching through the Gemini API, export all 10 to CSV and feed the file to your script. You can try LzyPrompt free for 7 days.