Veo 3 Prompts: How to Structure Them, with Examples

Google's Veo 3 made one thing popular: prompts written like tiny screenplays, with a shot, an action, a camera move and even a line of dialogue or sound. This guide breaks down how people structure Veo 3 prompts, with examples you can adapt. Most of the technique transfers to other tools too, including generatorvideo.ai. One honest difference up front: both generate sound, but spoken dialogue is Veo's strength, so on our side keep sound cues to ambience, effects and music. If you want to test a prompt cheaply, try it here: 3 free videos, then $1 each.

The anatomy of a Veo 3 prompt

A common structure people use with Veo 3 has five to seven parts, usually in this order:

• Shot type: close-up, wide shot, over-the-shoulder, aerial.
• Subject: who or what, with two or three visual details.
• Action: what happens during the clip, in plain verbs.
• Setting: location and time of day.
• Camera movement: dolly in, slow pan, handheld, static.
• Style and light: film look, color, lens feel.
• Audio: dialogue in quotes (Veo), ambient sounds, music mood.

Veo clips run around 8 seconds, so one scene with one beat of action is the sweet spot.

Example Veo 3 prompts, annotated

• "Close-up of a weathered sailor at the helm in a storm, rain lashing his face, he shouts 'Hold on!', handheld camera, cold blue light. Audio: howling wind, creaking wood." - Dialogue in quotes plus separate audio cues.
• "Wide aerial shot of a lone red car on a desert highway at sunset, camera slowly descends toward the road, warm cinematic grade. Audio: distant engine hum." - Camera move described as a path.
• "Medium shot of a street food vendor in Bangkok flipping noodles in a wok, flames flaring, night market bokeh. Audio: sizzling, crowd chatter." - Sound that matches visible action.
• "Over-the-shoulder shot of a girl reading a letter by candlelight, she whispers 'He's coming back', slow push-in, warm 1940s film look." - Period style named explicitly.
• "Static shot of a fox cub peeking out of a snowy burrow, blinking at the camera, soft overcast light. Audio: quiet wind, crunching snow."

Adapting Veo 3 prompts for generatorvideo.ai

generatorvideo.ai generates synchronized sound too, but it isn't built for spoken lines. Keep the ambience and effects cues, and move dialogue into visuals:

• Remove dialogue in quotes - the model may try to render text or odd mouth movement. Keep cues like "howling wind" or "sizzling" as plain description.
• Replace a spoken line with a visible reaction: "she smiles and wipes a tear".
• Show sound sources visually: "flames flare", "waves hit rocks", so the generated sound has something to sync to.
• Keep the shot, subject, action, camera and light structure exactly as is.

Our clips run 3-30 seconds (free videos: 3, 5 or 8 s), so an 8-second Veo-style scene maps well. Switch sound on or off per clip, and add voiceover afterwards in CapCut or Premiere.

Camera and style phrases that transfer well

• "slow dolly in", "orbit around the subject", "crane shot rising", "handheld, slight shake"
• "shallow depth of field", "anamorphic lens flare", "35mm film grain"
• "golden hour", "overcast soft light", "neon night", "candlelight"
• "1970s documentary look", "glossy commercial", "stop-motion style"

A fuller list is in our camera movement prompts guide, and how to write text to video prompts covers structure in more depth.

Where Veo 3 fits, and when to use something else

Veo 3 is a strong choice when you need synchronized sound or dialogue in the clip itself. It's accessed through Google's Flow and Gemini apps with Google AI plans, or via the Gemini API and Vertex AI; our Veo 3 pricing page explains the credits as of September 2026. If you only need visuals - B-roll, loops, social clips you'll score yourself - a pay-per-clip generator can be the cheaper route. On generatorvideo.ai each clip is a flat $1 with no subscription; 20 test clips from a $20 top-up is a practical way to refine a prompt before spending premium credits. Test a prompt now.