How to create an AI product video ad in 4 shots

To create an AI-assisted product video ad, define one audience, one verified message and one next action. Build a short shot plan around the real product, then prepare the visuals and narration needed for that plan. Review the complete edit against your product information before distributing it.

Upload your product photos to Product Video: they become a structured video ad you can review before publishing.

Our example uses a fictional desk organizer called “Desk Pocket” to demonstrate the planning process.

Start with the action, not the visual effect

Decide what a viewer should do after watching: inspect the product details, choose a size, explore a collection or read an offer. Select one action that the destination page actually supports. Asking people to visit a page with no relevant information makes even a clear ad less useful.

For our example, the action is “View the organizer’s details.” The audience is someone looking for a simple way to keep everyday stationery together. The message is about the intended use, not a promise to increase productivity or eliminate clutter permanently.

Write that message in one sentence before choosing camera movement, music or visual style. When a proposed scene does not help communicate the sentence, leave it out of the first version.

Prepare an approved product sheet

List the product name, actual appearance, dimensions, materials, included items and authorized reference images. Add the destination URL and any approved offer terms. Mark missing information as missing rather than asking AI to fill it in.

Our fictional brief specifies a single organizer with two compartments and a plain exterior. Pens shown in a scene are demonstration props, not included accessories. The material, delivery details and sale terms are deliberately unspecified, so they must not appear as invented facts in the ad.

Separate product facts from creative choices. You may explore a neutral background or warmer lighting, but a different number of compartments changes the product. Keep a short “must remain accurate” list beside the visual brief and check every accepted asset against it.

Choose what needs to be real

Use actual product photographs or footage when the audience needs to inspect details. Use an illustration to explain an idea or set a scene, clearly enough that viewers are not misled about what they are seeing.

If accurate product appearance is difficult to preserve in generated motion, keep the product image static and use a simpler edit. Animation is optional; showing the correct item is not. A restrained presentation may serve the message better than a visually elaborate but inaccurate clip. To animate a product photo without distorting it, follow the method to animate a photo with AI.

This is especially important for text, logos, proportions and included components. A pleasing image can still be unsuitable for an ad if it changes what the customer expects to receive.

A 4-shot structure for a product video ad

Give each part a clear job.

Part	Communication job	Visual plan	Copy or narration	Acceptance check
Opening	Introduce the everyday situation	A desk with a few stationery items	“Keep your everyday essentials together.”	No exaggerated problem or invented result
Product	Identify the item	Authorized product image or footage	“Meet Desk Pocket.”	Appearance matches the approved reference
Use	Explain the intended use	Real placement of pens in the organizer	“A place for the items you reach for.”	Props are not confused with included items
Action	Give the next step	Product view and a separately added end card	“View the product details.”	Destination and wording match

Read the sequence without music. You should still understand what is being shown and why. If the story only makes sense because of a dramatic soundtrack, simplify the visual or verbal explanation.

Prepare the ad in Deepnia

Open Product Video and upload your product photos: it turns them into a structured video ad. For a single shot, use Deepnia Video with Gemini Omni, which accepts up to 7 reference images.

Start with one scene that matters to the message. Review its product details before producing more assets. If you are still organizing shots, the text-to-video planning guide explains how to separate an outline into manageable visual instructions.

Keep the visual request specific

Describe one action or composition at a time. Avoid combining multiple scene changes, a product transformation, detailed text and complex camera movement in the same request. Gemini Omni includes sound; Kling 3.0 chains several shots in one clip.

Example prompt for an illustrative opening:

A quiet home-office desk beside a window, with a few everyday stationery items arranged naturally. Soft daylight, stable framing and a neutral background. No people, logos, readable text or product-performance demonstration. Leave clear space for a title added during editing.

For shots where the organizer must be exact, add your product photos as reference images.

Write narration that adds information

Say what the viewer needs to understand rather than describing every visible movement. Keep the product name and next action easy to hear. If a scene already shows an item being placed in a compartment, narration can explain the intended use instead of repeating the action.

Read the script aloud before preparing audio. Shorten sentences that are awkward to speak and verify names, numbers and offer wording. The natural voiceover guide covers preparation and listening checks; then generate the voice in Deepnia Audio.

Review the accepted narration alongside the visual sequence. A correct sentence paired with an unrelated image can still create an unintended impression. Keep music low enough that it does not conceal essential information.

For a tighter advertising script, follow the short-ad voiceover guide. It focuses on one message, a measured recording and a clear closing action rather than simply reading copy faster.

Add readable text during the edit

Before adding another layer, check that the soundtrack leaves space for the offer and closing instruction. Use the background-music brief and listening checks if you need music; for an original track, use Suno V5 in Deepnia Audio.

Prepare product names, offer terms and end-card wording as separate editable elements. Do not rely on incidental text inside a generated image for information that must be exact. Check the final design at the size people are likely to view it.

For a finished edit with voice, captions and music, Storytelling builds the scenes from your script and photos, then assembles them into one film.

Before adapting to another placement, check its current format requirements and whether essential content remains visible. A crop can remove the product, a qualification or the next action even when the original composition worked well.

Review the offer and destination together

Watch the complete export and compare it with the approved product sheet. Check appearance, captions, narration, included items and the final action. Confirm that any displayed offer is available on the destination and that the page explains the necessary details.

Keep permissions and applicable platform requirements in your review, including for AI-created assets. Avoid unsupported testimonials, certifications or performance claims, and correct ambiguous scenes before distribution.

Save the approved version with its references and copy. When testing another creative direction, identify what changed so you can interpret any observations later.

Finish one ad before building a larger kit

Your first deliverable is one coherent ad with accurate product information and a clear next action. Once that is approved, you can adapt the material into other assets using the product launch kit guide. To carry the same message into other formats, follow the guide to adapting a campaign into images, video and audio.

Complete your product sheet, then upload your photos to Product Video to get your first ad.