What are the best Veo 3 prompts?

The best Veo 3 prompts for short-form combine five elements: specific subject, one clear motion, explicit camera framing, 9:16 format specification, and at least one audio cue. Prompts between 3-6 sentences work best. Google DeepMind's prompt guide confirms the model responds best to detailed, intentional descriptions covering subject, context, action, and style.

How do you prompt for vertical video?

Include "9:16 vertical format" or "vertical composition" directly in the prompt. Design the scene for a tall frame — single subjects, close-ups, overhead angles. YouTube's Shorts guidelines recommend keeping key elements in the center 75% of frame to avoid UI overlays — the same principle applies to AI-generated clips.

Can Veo 3 prompts make viral videos?

No prompt guarantees virality. A prompt generates 5-8 seconds of raw footage. What makes something perform is the hook, pacing, text overlay, audio choice, and posting context. The AI clip is one ingredient, not the whole recipe. Your bottleneck isn't footage — it's shipping speed and testing volume.

Do prompts replace editing?

Not close. A Veo 3 clip is raw material. You still need hooks, captions, music, CTA, and platform-native pacing in an editor. The prompt gets you from zero to a usable visual faster than shooting. But posting an unedited AI clip is like posting raw camera footage — same result.

Best Veo 3 Prompts for Shorts and Reels

Hey, Maya is here. I spent my first two weeks with Veo 3 writing prompts that looked great on paper and produced clips I couldn't use. Sweeping dolly shots. Moody shallow depth of field. Golden hour everything. Every prompt guide I found was optimized for cinematic widescreen — slow, atmospheric, wide. Beautiful to watch, completely wrong for TikTok.

Here's what nobody tells you about Veo 3 prompts: writing them for short-form platforms is a different discipline than writing them for film-style content. TikTok, Reels, and Shorts don't reward cinematic beauty. They reward hooks, motion in the first second, and a feel that belongs on the platform. Most prompt guides skip all three.

This piece breaks down how to structure Veo 3 prompts for vertical short-form — real examples, real mistakes, and the adjustments that actually made my output usable.

Why Veo 3 Prompts Need Short-Form Structure

Veo 3 generates clips up to 8 seconds with synchronized audio — dialogue, sound effects, and ambient noise in a single pass. With the Veo 3.1 update, it now supports native 9:16 vertical video, resolutions up to 4K, and improved character consistency through reference images. Google's official Veo page confirms the model handles both 16:9 and 9:16 formats, runs at 24fps, and processes audio at 48kHz stereo.

That's the spec sheet. Here's the problem: most prompts people write describe scenes built for a horizontal monitor. A wide establishing shot of a city skyline. Two people walking side by side. A car driving across a desert. These compositions fall apart in a vertical feed.

Short-form has different rules. The first 1.5 seconds decide whether someone swipes. The frame is tall and narrow. Pacing should feel like content that lives on the platform — not a trailer ported from YouTube. If your prompt doesn't account for these constraints, the output might look impressive on your desktop and die in a feed. Looking like a movie isn't an advantage. Looking platform-native is.

What Makes a Good Veo 3 Prompt

Google Cloud's prompting guide for Veo 3.1 recommends covering subject, action, style, camera, and audio in every prompt. For short-form, I add two more: platform format and hook timing. Here's how each element works when you're prompting for Shorts and Reels.

Cinematic Neon City Night: Effective Veo 3 Prompts Example

Subject

Be specific. "A person" gives you something generic. "A woman's hand placing a small amber perfume bottle on a marble countertop" gives you a clip you can use. For product content, the product should be the subject, not a person holding it. Close-ups and detail shots generate more reliably than full-body scenes in vertical.

Motion

Veo 3 handles simple, directional motion well — a hand entering frame, liquid pouring, steam rising, fabric falling. Complex choreography still breaks. For short-form, you rarely need complex motion. One clear motion per clip is enough. That pour, that reveal, that slow drop — that's your visual beat.

Camera

Specify camera explicitly. "Close-up," "medium shot," "overhead shot" are more reliable than leaving the model to guess. For vertical, skip wide establishing shots — they waste frame space on a phone. Tight framing works better in 9:16. Keep camera movement to one axis: a slow tilt up, a gentle push in, a static lock. Don't stack multiple moves in 8 seconds.

Platform Format

The element most people skip. Include "9:16 vertical format" directly in your prompt. Even with native vertical support, specifying format reinforces the AI's composition decisions. Without it, you sometimes get output that technically fits 9:16 but feels like a cropped horizontal shot — subjects pushed to edges, dead space at top and bottom.

Audio or Pacing Intent

Veo 3 generates audio natively — you can specify dialogue in quotes, describe sound effects, and set ambient tone. For short-form, audio prompting is more about pacing than production value. An impact sound in the first second creates a natural hook. Silence followed by a sudden sound creates tension. One or two audio cues per prompt is enough. Platform audio — trending sounds, voiceover, music — will usually replace the AI audio anyway, but the pacing it creates in the visual still matters.

Abstract Texture Photography Achieved with Veo 3 Prompts

Prompt Examples by Use Case

Close-up, 9:16 vertical, a hand slowly places a matte black skincare bottle onto a wet stone surface, water droplets visible, soft natural daylight from the left, camera slowly pushes in, ambient sound of water dripping on stone.

This gives you a 5-8 second clip with clear product focus, subtle motion, platform-ready framing. Generate 3-5 variations with different surfaces and lighting, then pick the strongest. Five variations beats one "perfect" video every time.

Faceless Loop

Overhead shot, 9:16 vertical, a notebook open to a handwritten to-do list being checked off by a hand holding a black pen, warm desk lamp lighting, the last checkbox is ticked and the pen is set down, soft ambient paper rustling sound.

Overhead and close-up angles eliminate the need for consistent faces — still a weak spot in AI video. For faceless content, these angles are your safest bet.

Story Hook

Medium shot, 9:16 vertical, a woman in a yellow raincoat at the edge of a rooftop looking down at a city street, wind blowing her hair, she turns toward camera with a surprised expression, SFX: distant city traffic then sudden silence.

The "turn toward camera" creates a natural hook — viewers want to know what happens next. The audio shift from noise to silence is a pacing trick that works even if you swap the audio later. Pair with a text overlay in editing and you've got a scroll-stopper.

Trend-Style Visual

POV shot, 9:16 vertical, a hand reaches out and opens a mystery gift box, revealing a golden object that catches the light, camera leans in slightly, warm soft lighting, satisfying unwrapping sound effects.

POV and unboxing formats consistently rank among the most replicated structures on TikTok. The TikTok Creative Center surfaces these in its Top Ads section regularly. Designing prompts around proven platform structures gives you output that actually fits.

Dynamic Aerial Tracking Shots Using Action Veo 3 Prompts

Common Prompt Mistakes

Asking for Cinematic Shots When the Platform Is TikTok

"Sweeping crane shot over a misty valley at golden hour" — gorgeous prompt, dead on Reels. Wide shots with slow camera movement don't hold attention on a phone. You need tight, immediate, subject-forward compositions. Save crane shots for YouTube long-form.

Ignoring Hook Timing

If nothing visually interesting happens in the first 2 seconds of your clip, the rest doesn't matter. Your prompt needs to front-load the visual payoff. Put the action or reveal at the start of the description, not the end. Veo 3 tends to follow the temporal order of your prompt — first described, first rendered.

Forgetting Vertical Composition

Even with 9:16 specified, if your prompt describes a naturally horizontal scene — a panorama, a car crossing a desert, two people side by side — the model struggles to compose it vertically. Design scenes for a tall frame from the start. Single subjects, vertical objects, close-ups, and overhead angles suit 9:16 naturally.

Photorealistic Portrait Generation Using Veo 3 Prompts

Conclusion

The gap between "impressive AI demo" and "content that works on TikTok" is a prompting problem. Veo 3 can produce platform-ready clips, but only if you prompt for the platform — not for a film reel.

Tight framing. One clear motion. Vertical composition. Front-loaded payoff. Generate variations, drop the best clip into your editor, add the layers that make it perform. First version doesn't need to be perfect — just shippable. Post it first. Let the data tell you what to optimize.