Veo Prompts: Cinematic Examples for Google Veo 3.1
A useful Veo prompt gives one subject, one action, a camera direction and a sound cue. The copyable examples below show how to combine those parts in a short scene. Check the model and settings available in your generation service before using a version-specific control.
What Google Veo Pays Attention to in a Prompt
Google's prompt guidance emphasizes framing, movement, style, lighting, location, action and dialogue. Put one visual event at the center of a short clip. Instead of “cinematic woman in a city,” write what she does, where the camera is, and what sound belongs to the shot. Specificity gives the model a clearer target, but generated dialogue, hands and motion still need review.
Subject, Action, Camera, Style, Audio: The Veo Formula
Copyable formula: “A [shot size] frames [subject] in [setting]. [One action] happens as the camera [one move or static]. Lighting is [source and quality]; style is [visual treatment]. Audio: [ambient sound/effect]. [Speaker] says: ‘[short line].’”
Dialogue: “Medium shot of an original florist closing her shop at dusk. She ties one bundle of flowers; the camera stays still. Warm window light. Audio: quiet street and paper rustle. Florist says: ‘These are for tomorrow.’”
Ambience: “Wide shot of a rain-soaked train platform at dawn. One empty carriage arrives slowly. The camera tracks right. Cool blue light. Audio: rain on the roof, distant brakes, no speech.”
Product: “Close shot of a plain ceramic travel cup on a kitchen counter. A hand lifts it once; camera pushes in gently. Soft morning light. Audio: lid click and kettle hiss. No brand claims or on-screen text.” These are illustrative prompts, not actual outputs.
Annotated Prompts: Spoken Dialogue, Ambient Sound and Product Shots
In the florist example above, “Medium shot” sets the framing; “ties one bundle” gives the action; “camera stays still” limits movement; “quiet street and paper rustle” sets ambience; and the named florist plus a quoted line assigns speech. Change one part when a draft fails. If a word is mispronounced, shorten or revise the line and listen again. These annotations explain the prompt, not a tested output or a guarantee of native audio.
Check Model Settings Before Generating in CrePal
Availability note, checked November 4, 2026: Google's Gemini API documentation reports that the veo-3.1-generate-preview, veo-3.1-fast-generate-preview and veo-3.1-lite-generate-preview endpoints were shut down on October 22. The older examples and variant table remaining on that page are not a current API availability guide. This notice concerns those endpoints; it does not establish which Veo versions CrePal currently offers.
Before using a tier-specific prompt, inspect the current model selector and duration, reference-image and audio controls in the service you use. If that service currently offers both Fast and Lite, test the same short shot in each and compare the actual controls, output, credit cost and time shown there. Choose the option that meets the shot requirements; the tier name alone does not establish current quality, speed or availability. CrePal publicly names Google Veo but does not establish that every Google tier is selectable in every project. For a longer story, plan separate shots with consistent subject anchors and review each clip before assembly.
Keep the Prompt Stable When Comparing Available Variants
If your current service offers more than one Veo variant, test with the same short scene: “A plain ceramic cup on a kitchen counter; a hand lifts it once; fixed camera; soft morning light; lid click and kettle hiss.” Record the selected model name, duration, aspect ratio, audio setting and any reference assets. Then judge the same four outcomes in each clip: the cup stays the same shape, the hand performs one lift, the sound matches the action, and the scene ends cleanly.
Do not infer current availability from an old example snippet. Google's Gemini API documentation says the named Veo 3.1 Preview endpoints were shut down on October 22, 2026, while other access routes may differ. The model selector in the service used for production is the decision point. If only one version is offered, this becomes a single-clip acceptance check rather than a comparison. Save the prompt and settings together with the chosen clip.
If the action drifts, use the prompt-to-video troubleshooting guide to change one control at a time. For a multi-shot alternative with reference assets and sound roles, compare the scene plan with the Seedance 2.0 prompt guide, without assuming identical controls are available.
Check the Clip Against Its One Main Job
| Intended result | Keep in the prompt | Review the output | If the check fails |
|---|---|---|---|
| A spoken line | One named speaker, a short quoted line and a visible face | Correct speaker, complete words and mouth timing | Shorten the line or separate speakers into shots |
| A camera move | One direction, pace and stable subject | Straight edges, subject identity and final framing | Remove secondary moves or reduce the motion |
| A product action | One product, one action and a restrained sound cue | Shape, hand interaction, label and sound timing | Simplify the action; reject a changed product |
Record the model and settings actually offered by the service beside each draft. If the selected route lacks native audio or a reference control, adapt the plan instead of treating this table as proof that the control exists.
CrePal Product Sources and How to Use Them
CrePal's video creator page describes model routing by shot and a reviewable scene plan. For Veo, check the current route in the product before treating a model name or endpoint in this guide as available; Google documentation linked above governs the named API variants and retirement dates. The shot templates below are starting instructions, not a verified CrePal/Veo output gallery.
Google DeepMind's Veo prompt guide separates framing and motion, style, lighting, character, location, action and dialogue. Use those as fields to diagnose one weak shot, rather than adding every field to every prompt. The guide is about Veo 3 prompting; it is not evidence that a specific CrePal route exposes a particular Veo version today.
Related official reading: CrePal's Veo motion article. CrePal's Veo motion article describes testing pans, push-ins and other camera moves, and reports that faster moves can create visible artifacts. For the Veo examples here, specify one restrained camera movement and inspect edge stability; check the current model selector because the article's tested interface and variants may differ from today's route. The article reports its author's tests; this page has not reproduced them or documented its own Veo output.
Evidence and Review Notes
Evidence note added 5 November 2026. The prompts in this guide are editorial examples. The linked CrePal pages document product workflows or illustrative use cases; they do not verify an output from these exact prompts. No prompt-output pair is published here.
To document a result for this topic: Record the exact available model label, prompt, settings and clip; review speech, camera path and product actions. Keep the exact prompt, model/version, settings, output URL, revisions and human review decision together.
Test your prompt in CrePal
Start with one example from this guide, then compare the result with your intended subject, style and details.
Site and support
CrePal's Terms identify SIGNOO PTE. LTD. as the service provider. The prompt examples here are instructions to try, not recorded generation tests.
Frequently Asked Questions
Put the spoken line in quotation marks and say who speaks and how, such as a calm narrator.
Clip duration depends on the model and service currently selected. Check its available setting before writing the shot; plan multiple scenes when the story exceeds one clip.
Google documented reference-image options for certain Veo 3.1 variants. Check the current model and interface for reference inputs; do not rely on the retired Preview endpoints as an availability guide.