# Image-to-Video Prompts for Product Ads: 12 Examples — Cut Marketing

> Twelve image-to-video prompts we actually shipped, from turntable spins to pours and macro slides, with the six-part motion formula, model picks and fix lines.

# Image-to-Video Prompts for Product Ads: 12 Examples

Twelve image-to-video prompts we actually shipped, from turntable spins to pours and macro slides, with the six-part motion formula, model picks and fix lines.

Published 2026-11-10 Last updated 2026-11-16 11 min read [Cut Marketing Team](/editorial)

![AI-generated printed photo of a white sneaker on an easel beside a phone on a camera slider showing the sneaker mid-turn, with a lime tape arc on the desk](/samples/blog/image-to-video-prompts-product-ads.webp)

To write an image-to-video prompt for a product ad, describe only what changes after the first frame: one subject action, one camera move, any change in light, the sound, what must not appear, and a line telling the model to hold the product’s shape. The image already defines the scene, so re-describing it wastes the prompt and invites drift. The twelve prompts below are ones we actually shipped in our own showcase in November 2026, with the model each was rendered on.

**In short:**

-   The still is the scene; the prompt is the motion. Write verbs, not adjectives.
-   One action and one camera move per clip. A second action is a second clip.
-   End with an exclusion line and a stability line. Every prompt we shipped has both.
-   When physics or logos go wrong, fix it with one explicit sentence, not a longer prompt.
-   Crop the first frame to your publishing ratio before you upload it.

## What an image-to-video prompt is

Image-to-video is a generation mode in which a model takes one still image as the first frame and animates it for a few seconds according to a written prompt. For product ads the still is usually a product photo or a generated product scene, and the prompt is a short description of motion. In Cut Marketing (Cutma) this is how [Product Video](/tools/create-video-ad) works when you give it a reference image.

## The six-part motion formula

| Part | What to write | Example from a shipped prompt |
| --- | --- | --- |
| 1\. Subject action | One physical event, in order | “Beer streams from the brass tap into the tulip glass, the creamy foam rises to the rim and settles into a perfect dome” |
| 2\. Camera | One move, or explicitly none | “the camera holds steady with a gentle push-in” |
| 3\. Light and atmosphere | What shifts in the air or on surfaces | “reflections glide across the lenses and glints travel along the titanium rims” |
| 4\. Audio | Sounds and music, for models with native audio | “Native audio: crisp beer pour and fizz, low taproom chatter, mellow indie music.” |
| 5\. Exclusions | What must not appear | “No text, no captions, no logos appear.” |
| 6\. Stability | Hold the product | “The subject keeps its exact shape, colour and proportions throughout; smooth, stable, cinematic motion with no warping.” |

Parts 5 and 6 are the same two sentences at the end of every prompt below except prompt 5, where the stability line is rewritten for the chest. Copy them as they are.

## 12 prompts we shipped

Each prompt started from a 2K still. The stills and finished clips are in the [showcase](/showcase); the model named is the one the published clip was rendered on. For brevity the shared ending (parts 5 and 6) is written as **\[ending\]**.

### Rigid products: the camera or a turntable moves

**1\. Turntable rotation (eyeglasses, Kling 4.0)**

> The eyeglasses rotate slowly on the plinth, reflections glide across the lenses and glints travel along the titanium rims, the background stays softly blurred. Native audio: quiet boutique ambience, a soft elegant piano melody. \[ending\]

**2\. Push-in with a light change (EV charger, Kling 4.0)**

> The lime LED ring around the charge port pulses gently like a heartbeat, soft light rolls across the pearl paint, slow push-in toward the port. Native audio: a soft electric charge hum, a calm futuristic synth pad. \[ending\]

**3\. Lateral macro slide (bouclé sofa, Kling 4.0)**

> A slow lateral macro slide along the curved bouclé arm, dappled leaf shadows sway gently across the looped yarn, sunlight warms. Native audio: quiet room tone, faint rustling leaves outside, soft jazz piano. \[ending\]

**4\. Tracking beside fast motion (tyre, Kling 4.0)**

> The wheel spins fast with motion-blurred spokes and a glittering spray of water keeps flying off the tread through the backlight the whole time. The camera stays locked low and tight on the wheel with the same framing as the first frame, travelling alongside at the same speed; the background stays a soft dark blur of round city-light bokeh with no signs, no neon lettering and no buildings coming into focus. The wheel centre stays a plain dark hub with lug nuts and the sidewall stays blank; no emblem, badge, symbol or lettering ever appears. Native audio: tire hiss on wet road, a deep engine hum, cinematic bass hit. \[ending\]

**5\. The reveal (game treasure chest, Kling 4.0)**

> The ornate chest lid swings fully open on its hinges, and from inside the chest a fountain of golden light, glowing green gems and gold coins erupts upward out of the chest, rises high, then showers back down around it and spills onto the stone ledge; light flares across the stone walls and the camera pushes in slightly. Nothing falls from above; everything bursts up from inside the chest. Native audio: heavy creak of the lid, a magical whoosh, coin and gem jingles, triumphant game fanfare. No text, no captions, no logos appear. The chest keeps its exact shape, colour and ornament throughout; smooth, stable, cinematic motion with no warping.

### Soft things: the subject moves

**6\. The pour (beer, Seedance 2.5)**

> Beer streams from the brass tap into the tulip glass, the creamy foam rises to the rim and settles into a perfect dome, bubbles race upward, the camera holds steady with a gentle push-in. Native audio: crisp beer pour and fizz, low taproom chatter, mellow indie music. \[ending\]

**7\. The pull-apart (burger, Seedance 2.5)**

> The right half of the smash burger slowly pulls away from the left half, molten cheese stretches into long glossy strands and sags gently, steam curls upward, the camera creeps in slowly with a subtle macro push. Native audio: soft sizzle, a gentle crunch, warm low lounge music. \[ending\]

**8\. Steam and a sprinkle (air fryer, Seedance 2.5)**

> Steam billows up from the crispy fries in the basket, a pinch of salt flakes falls and bounces on the fries, slow push-in. Native audio: crispy crackle, soft kitchen ambience, cheerful light pop music. \[ending\]

**9\. Hands, top-down (suitcase, Seedance 2.5)**

> Top-down view: the hands lower the folded sweater into the suitcase, smooth it flat, then tuck a rolled shirt beside it, natural tidy movements, sunlight soft and steady. Native audio: soft fabric rustle, zipper tick, light cheerful acoustic music. \[ending\]

**10\. Slow motion on hair (salon blow-dry, Seedance 2.5)**

> Slow-motion: the round brush glides down the hair while the dryer blows, silky strands lift, flow and fall with a glossy shine, the camera holds a smooth close-up. Native audio: soft dryer whoosh, salon ambience, smooth R&B instrumental. \[ending\]

### Places: the atmosphere moves

**11\. Ambient scene (hot-spring bath, Gemini Omni Flash 1.2)**

> Steam drifts and swirls across the hot spring, gentle ripples carry lantern reflections, a maple leaf floats down onto the water, a slow forward dolly toward the valley. Native audio: trickling spring water, evening insects, a distant temple bell, soft koto ambience. \[ending\]

**12\. Parallax through a window (train journey, Veo 3.1 Fast)**

> The landscape glides past outside the window with natural parallax, the near fields streak by while distant mountains move slowly, warm sunlight flickers across the table and the latte surface trembles gently. Native audio: rhythmic train on rails, soft cabin hum, mellow acoustic music. \[ending\]

**\[ending\]** = “No text, no captions, no logos appear. The subject keeps its exact shape, colour and proportions throughout; smooth, stable, cinematic motion with no warping.”

## Camera moves that behave

| Move | Wording that works | Use it for |
| --- | --- | --- |
| Push-in | “slow push-in toward the \[detail\]” | Hero shots, a single feature |
| Turntable | “rotates slowly on the plinth, the background stays softly blurred” | Rigid products seen from all sides |
| Lateral slide | “a slow lateral macro slide along the \[surface\]” | Texture and material |
| Locked-off | “the camera holds steady” | Pours, steam, anything where the subject moves |
| Tracking | “the camera stays locked low and tight with the same framing as the first frame, travelling alongside at the same speed” | Wheels, moving vehicles |
| Top-down | “top-down view: the hands \[action\]” | Packing, assembling, plating |
| Forward dolly | “a slow forward dolly toward the \[far point\]” | Rooms, venues, landscapes |

What to avoid: fast orbits, whip pans, zooms across the whole scene and more than one camera instruction per clip. The more the viewpoint changes, the more of the product the model has to invent.

## Choosing the model and the draft cost

| Shot | Model | Draft setting | Credits |
| --- | --- | --- | --- |
| Rigid product, camera or turntable moves | [Kling 4.0](/models/kling-4.0) | 720p, 5 s | 140 |
| Food, liquid, steam, hands, hair, fabric | [Seedance 2.5](/models/seedance-2.5) | 480p, 5 s | 100 |
| Place and atmosphere | [Gemini Omni Flash 1.2](/models/gemini-omni-flash-1.2) | 720p, 4 s | 48 |
| Simple scenic move | [Veo 3.1 Fast](/models/veo-3.1-fast) | 720p or 1080p, 8 s | 32 |
| Many variants of one idea | [Wan 3.0](/models/wan-3.0) | 720p, 5 s | 70 |
| Quick, low-cost first drafts | [Seedance 2.0 Fast](/models/seedance-2.0-fast) | 480p, 5 s | 40 |

One credit lists at $0.01, and plans start at $9 a month. The reasoning behind each pairing is in [Seedance 2.5 vs Kling 4.0 vs Veo 3.1 Fast](/blog/seedance-vs-kling-vs-veo-product-ads), and the full budget math is in [how much an AI video ad costs](/blog/ai-video-ad-cost-2026).

## Step by step

1.  **Prepare the first frame.** Use a sharp still with the whole product visible. If the clip is for Reels, Shorts or TikTok, crop it to 9:16 first: with Seedance and Wan the output follows the first frame’s shape. Still-image templates are in [AI product photography prompts](/blog/ai-product-photography-prompts).
2.  **Decide the one thing that moves.** Product, hand, liquid, light or camera. Pick one as the lead.
3.  **Write the six parts** in the order of the formula. Keep the action sentence in the order things happen.
4.  **Generate a short draft** on the cheapest setting that suits the shot.
5.  **Review at normal speed, then frame by frame** at the start, middle and end. Check shape, label, hands and the background. The checklist is in [how to review an AI product video](/blog/review-ai-product-videos).
6.  **Fix with one sentence.** If something specific went wrong, name it and say what should happen instead. Then regenerate.
7.  **Add captions and the call to action afterwards**, never in the prompt. Finish the Ad in the studio adds auto captions in 40 styles and your exact hook, price and CTA as real text, plus AI narration from your script and free CC0 music, and exports 9:16, 1:1 and 4:5 at once. Any outside editor works too.

## Fix lines for common failures

| What went wrong | Sentence to add |
| --- | --- |
| Objects fall or fly from the wrong direction | “Nothing falls from above; everything bursts up from inside the \[container\].” |
| A logo or lettering appears mid-clip | “The \[surface\] stays blank; no emblem, badge, symbol or lettering ever appears.” |
| The background sharpens into signs or buildings | “The background stays a soft dark blur with no signs and no buildings coming into focus.” |
| The camera wanders off the product | “The camera stays locked on the \[product\] with the same framing as the first frame.” |
| The product bends or melts during motion | “The \[product\] keeps its exact shape, colour and ornament throughout.” |
| Too much happens at once | Delete actions until one remains. |

Notice that prompts 4 and 5 are the longest of the twelve. Simple shots need short prompts; the long ones are long because each extra sentence corrects one specific failure.

## Limits

-   Clips are 4 to 15 seconds depending on the model, from a single first frame. A sequence of scenes is several clips joined afterwards; Finish the Ad in the studio joins and trims them.
-   Text on packaging is not reliably preserved through motion. Keep the label large and still, or keep it out of the moving part of the frame.
-   An audio line shapes the sound on models with native audio, but exact spoken dialogue is not guaranteed. Carry the message in captions, or add AI narration from your script in Finish the Ad.
-   A prompt cannot make a claim true. What the clip shows the product doing must be something it does.

## FAQ

### How do I write an image-to-video prompt for a product?

Describe what changes after the first frame, not what the image already shows. Use six parts: one subject action, one camera move, any light or atmosphere change, an audio line, an exclusion line for text and logos, and a stability line that says the product keeps its exact shape, colour and proportions.

### What should I not put in an image-to-video prompt?

Do not re-describe the whole image, stack several actions, or ask for text to appear. The image already defines the scene, extra actions make the product drift, and generated lettering is unreliable. One action and one camera move per clip works best.

### Which camera moves work best for product videos?

Simple ones: a slow push-in, a turntable rotation on a plinth, a lateral macro slide, a locked-off shot where only the subject moves, and a top-down view of hands. Fast orbits, whip pans and multi-step moves raise the chance that the product changes shape.

### Why does my product change shape in AI video?

Usually because the prompt asks for too much motion, the first frame hides part of the product, or nothing tells the model to hold the shape. Use a clear still, one slow move, and end with a line such as ‘the subject keeps its exact shape, colour and proportions throughout’.

### What image should I use as the first frame?

A sharp still in the aspect ratio you will publish, with the whole product visible and no text you need to stay readable. With Seedance and Wan the output follows the first frame’s shape, so crop to 9:16 before uploading for Reels, Shorts or TikTok.

### How much does an image-to-video clip cost?

In Cut Marketing clips start at 32 credits ($0.32) for an 8-second Veo 3.1 Fast clip at 720p or 1080p. Other from-prices for one clip: Seedance 2.0 Fast costs 40 credits (5 seconds, 480p), Gemini Omni Flash 1.2 costs 48 credits (4 seconds, 720p), Wan 3.0 costs 70 credits (5 seconds, 720p), Seedance 2.5 costs 100 credits (5 seconds, 480p) and Kling 4.0 costs 140 credits (5 seconds, 720p). One credit lists at $0.01, and plans start at $9 a month.

## Animate one product still

Sign in with Google and [open Product Video in the studio](https://app.cutmarketing.app/studio/create-video-ad?ui_locale=en), upload a product still as the reference and paste one of the prompts above with your own subject. No prompt yet? The free [AI video prompt generator](/free-tools/ai-video-prompt-generator) writes three for the studio’s video models (2 free runs a day with a free account). An 8-second 1080p Veo 3.1 Fast clip is 32 credits ($0.32), and the studio shows the credit cost before you generate; video needs more than the 30 trial credits, so see [pricing](/pricing) first. The [image-to-video generator](/image-to-video-generator) page shows what the tool produces.

## Keep reading

[Seedance 2.5 vs Kling 4.0 vs Veo 3.1 Fast for Ads](/blog/seedance-vs-kling-vs-veo-product-ads): Which AI video model fits which product ad shot: Seedance 2.5, Kling 4.0 and Veo 3.1 Fast compared on length, resolution, price per second and our own 40-clip shoot.

[How to review an AI product video before export](/blog/review-ai-product-videos): Use an actual bottle study to check product shape, camera movement, exact ad copy and approval. A practical review before handing off an AI product video.

[Best AI Ad Generators in 2026: 9 Tools for Product Ads](/blog/best-ai-ad-generators-2026): The best AI ad generators for product ads in 2026, compared: what 9 tools make, input to output, prices checked 11 November 2026, and which one fits your goal.

## Make the next ad in minutes.

Start with a product image or a short brief. The studio shows the credit cost before you generate.

[Open studio](https://app.cutmarketing.app/?ui_locale=en)
