How to use Veo 3 for ads: packshots, Flow and frames

What is Veo 3, and how do creators use it for ads?

Veo is Google's video model, and it generates sound with every clip. Most creators reach it through Google Flow, where Google's "Ingredients to Video" keeps characters and objects consistent from reference images. "Frames to Video" generates the motion between a start image and an end image (Google).

In the tutorials we reviewed, Veo is mostly used for product ads: packshots, pours, reveals and short commercials. UGC-style talking heads are rarer here than in our Seedance guide.

Facts from Google, checked 12 Oct 2026 (Gemini API docs, Google Cloud guide):

• Models: the current ones are Veo 3.1, Veo 3.1 Fast and Veo 3.1 Lite. Veo 3 and Veo 3 Fast are listed as deprecated in the API.
• Clip length: 4, 6 or 8 seconds, at 720p, 1080p or 4k. 1080p and 4k need 8 seconds.
• Aspect ratio: 16:9 or 9:16 only.
• Reference images: up to three, all of one person, character or product (not on Lite).
• Start and end frames: you can set a first frame and a last frame.
• Extension: a clip can be extended 7 seconds at a time, up to 20 times.
• Audio is always on, and every video carries a SynthID watermark. English is the fully supported language.

Which tools do creators pair with Veo 3?

Product ads made with Veo start with a still: a real packshot or a generated image. Based on 9 public tutorials posted between 12 Jan 2026 and 11 Oct 2026, reviewed by Anglr.

How do you make an ad with Veo 3, step by step?

The tutorials agree on a still-first, product-first order. Where they differ is how much of the ad Veo makes in one go: a single generation from a storyboard, or one clip per shot.

1. Get a clean packshot with a readable label. Muhammad Wahaj treats label accuracy as the first job. Mayur jadav shoots the product on a phone and removes the background. Abu kalam khandaker maps the final label artwork onto jar renders before any video.
2. Plan the shots as a storyboard. Muhammad Wahaj builds a ten-scene storyboard from hook to hero shot, Sami Hasnain makes storyboard frames for composition and lighting, and Janvi Shah generates images scene by scene and keeps only the ones that fit. Our AI storyboards guide covers this step for every model.
3. Have an LLM turn the plan into Veo prompts. Muhammad Wahaj and Mayur jadav have ChatGPT write prompts that keep the product the same in every shot, and Janvi Shah gives Gemini her selected frames plus a shot-by-shot breakdown of the scenes.
4. Upload the product as a reference. Muhammad Wahaj, Mayur jadav and Sami Hasnain all work in Google Flow, with the product image uploaded so the logo and label hold. Google's limit is three reference images of one subject per clip.
5. Animate stills you have approved. Abu kalam khandaker and @casualhyped run image-to-video from finished stills. Manar Elmasry sets both a start and an end frame in Flow, so Veo only has to invent the motion between two stills.
6. Cut, type and score it in an editor. Abu kalam khandaker cuts every few seconds and adds typography and an end-card call to action. Manar Elmasry adds a separate voiceover and music, and Sami Hasnain and Mayur jadav assemble the clips into the final ad.

Not every creator finds Veo the right fit. George Stock picks Veo for a cinematic look and other models for creator-style footage. Ian Ramírez Rivera reports faces and storefronts drifting mid-shot with Veo 3 and moved a client ad to Seedance 2.5. Test your own product on two models before you commit a campaign to one.

How should you write a Veo 3 prompt?

Use Google Cloud's order: cinematography, subject, action, context, then style and ambiance.

• Sound: put dialogue in quotation marks after the speaker, start sound effects with SFX: and describe the background with Ambient noise:.
• Negatives: describe what you want instead of what you don't.
• Settings, not prompt text: duration and aspect ratio are set in Flow or the API.
• Timestamps: Google's own guide shows a timestamped prompt with four two-second beats inside one 8-second clip. Unlike Seedance's guide, it treats timing as something you can prompt.

For a product reference, keep the prompt about what changes: the camera move, the action around the product, the light. Let the reference image carry the label and shape. Mayur jadav names the hard part: "The most challenging part was ensuring the AI used the EXACT product packaging without redesigning the label."

Templates by ad format are on Veo 3 prompts, and the AI ad prompt builder writes a single shot in Google's order. To estimate a batch of 8-second takes, use the AI video cost calculator and check plans on Veo 3 pricing.

Common problems and fixes

What creators report

These are the authors' own claims, not tested by Anglr.

• Muhammad Wahaj reports a 10-second vertical beverage spot in 23 minutes from one product photo.
• @casualhyped reports two hook clips made with Nano Banana Pro and Veo 3 reaching 23M and 7.1M views.
• Ian Ramírez Rivera reports more drift from Veo 3 than from Seedance 2.5 on his client project.

A clean product spot is a strong start, but brands buy ads that sell. If your AI product ads already perform, Anglr pays AI creators on performance to remake winning scripts: see what we look for. For the full ad-style workflow, read AI product videos or AI commercials, or browse AI ad guides.