# AI product photography prompts: 8 ready-to-use prompts for every listing image

> AI product photography prompts work best when each one targets a single listing image: a hero shot on white, an infographic, dimensions, a lifestyle scene, a person with the product, a seasonal background. We shot a travel mug on a kitchen counter and turned that one photo into seven listing images with GPT Image 2.5, Nano Banana and Seedream — the mug stayed the same in every shot. Full prompts and results below.

Published 2026-10-03 · Updated 2026-10-03
URL: https://ai2face.com/blog/promty-dlya-kartochek-tovara

A product photography prompt works when it has three things: the product photo as a reference, an instruction not to change the product, and a description of one specific listing image. We tested this on a single travel mug: we photographed it on a kitchen counter and ran it through eight prompts, from the hero shot to a video cover. This guide covers image prompts only; titles and descriptions are a separate job for text models. How to plan the whole image set is covered in the guide to [AI product images for ecommerce](/blog/kartochki-tovarov-nejroset).

## 1. How to write an AI product photography prompt: five building blocks

Every prompt below is built from five blocks. Skipping any of them is a common reason for a failed image.

| Block | What to write | Example |
|---|---|---|
| Reference and lock | what to take from the photo and what not to change | “keep the mug exactly as in the photo: shape, color, lid” |
| Image type | which listing image you need | “hero shot on a white background”, “dimensions slide” |
| Scene and light | background, props, lighting | “light oak desk, soft window light” |
| Text | every label in quotes, plus its position | “a headline at the top: ‘Dimensions’” |
| Constraints | aspect ratio and what to exclude | “vertical 3:4, no other text, no logos” |

The model developers recommend the same. OpenAI's GPT Image prompting guide advises putting required text in quotes, describing its placement, and asking to “change only X” while listing what to preserve. For [Nano Banana](/models/nano-banana), Google offers a product-shot template: product, surface, lighting setup, camera angle, detail in focus.

Pick the model for the job:

| Image | Model | Why |
|---|---|---|
| Hero shot, lifestyle, close-ups | [Nano Banana 2](/models/nano-banana-2) | fast and keeps the product accurate from a reference |
| Hard background swaps, a person with the product | [Nano Banana Pro](/models/nano-banana-pro) | cuts out the product more reliably and renders people better |
| Infographics, dimensions, any text | [GPT Image 2.5](/models/gpt-image) | error-free Cyrillic text in our test |
| Seasonal and themed backgrounds | [Seedream](/models/seedream) | changes the setting cleanly around the product |

## 2. Prompt for the main product photo

The main image is the product on a white or light solid background with nothing else in frame. Yandex Market's seller help puts it plainly: a white or transparent background is best, as long as it doesn't get in the way.

```
Cut out the travel mug from this photo and place it alone on a clean white
studio background. Remove the kitchen, kettle, towel and everything else
completely. The mug centered, upright, three-quarter view, filling about 75%
of the frame height, soft even studio light, a soft natural shadow under
the mug. Keep the mug exactly as in the photo: proportions, matte sage green
color, bamboo lid, grey slider, steel rim. No other objects, no text,
no logos. Vertical 3:4
```

![On the left, a travel mug on a kitchen counter next to a kettle and a towel; on the right, the same mug on a clean white background](https://ai2face.com/_astro/before-after.D3xrZIQc_Z1rUPRP.jpg "Before: a kitchen photo. After: a hero shot on white. Nano Banana Pro. We generated the source photo for this test too — in real work, upload a real photo of your product")

The word “cut out” and an explicit list of what to remove do the job. On a first attempt with a softer “make a studio photo,” Nano Banana 2 left the kitchen in place and stretched the mug, so we rejected that image.

## 3. Prompt for a product infographic

An infographic is the product plus 3–4 benefits of 2–4 words each. Anything longer won't read on a phone screen. Put each label in quotes and ask for it to appear exactly once.

```
Turn this product photo into a marketplace infographic slide. Keep the travel
mug exactly as in the photo. The mug on the left on a soft light sage green
background with a subtle shadow. At the top a large headline
«Термокружка «Шалфей»». On the right three short benefits, each with a thin
line icon: «Держит тепло до 6 часов», «Не протекает в сумке», «Крышка из
бамбука». Clean modern layout, dark green text, all Russian text exactly as
written, each label appears once, no other text, no logos. Vertical 3:4
```

### Prompt for a dimensions slide

```
Turn this photo into a product dimensions slide. Keep the mug exactly as in
the photo, centered on a clean white background. Thin dark grey dimension
lines with arrows: a vertical one on the right labeled «18 см», a horizontal
one under the mug labeled «8,5 см». A headline at the top «Размеры», a small
line under the mug «Объём 450 мл». All Russian text exactly as written,
no other text, no logos. Vertical 3:4
```

![Two travel mug slides: an infographic with the headline “Термокружка «Шалфей»” and three benefits with icons, and a “Размеры” slide with 18 cm and 8.5 cm dimension lines](https://ai2face.com/_astro/infographic-set.KRYo3oye_Z2pHifA.jpg "Infographic and dimensions — GPT Image 2.5: every Russian label is error-free and the icons match the meaning. The numbers in the prompt are placeholders — take real ones from your product")

The model doesn't know or check numbers: it will print whatever you give it. Take heat retention, capacity and dimensions from the manufacturer's specs.

### Prompt for a close-up detail shot

A close-up shows what gets lost in a wide shot: material, mechanism, texture.

```
Macro photo of this travel mug's bamboo lid from above at 45°. The grey
slider is open, showing the drinking hole, the steel rim and the top of the
matte sage green body. Bamboo grain visible. Soft studio light, light grey
background, all details exactly as in the reference, no text, no logos.
Vertical 3:4
```

![Close-up of the travel mug's bamboo lid with the grey slider open and the drinking hole visible](https://ai2face.com/_astro/detail.C-JYFYQN_ZVtYqC.jpg "Detail shot — Nano Banana 2. The model invented the lid mechanism on its own: the slider is closed in the source photo. Check spots like this against the real product")

## 4. Prompt for a lifestyle product photo

A lifestyle shot answers the buyer's question “how will this look at my place?” Describe the place, the time of day and 2–3 props — more props pull attention away from the product.

```
Place this travel mug on a light oak desk in a bright Scandinavian home
office in the morning. Next to it a closed paper notebook, a pen and a small
potted plant, soft window light, shallow depth of field. No laptop, no
screens, no brand logos. Keep the shape, matte sage green color, bamboo lid,
grey slider and steel rim exactly the same. Photorealistic lifestyle photo,
no text. Vertical 3:4
```

![A travel mug on a light oak desk by a window next to a kraft notebook, a pen and a potted succulent](https://ai2face.com/_astro/interior.BwPFU18I_ZyJaXO.jpg "Lifestyle — Nano Banana 2. The mug matches the source and there are no third-party logos in frame")

The line “no laptop, no brand logos” came from a failed image: the first version had a laptop with a readable manufacturer logo. Yandex Market rejects photos with unrelated logos, and someone else's brand in your listing is a legal risk anyway.

## 5. Prompt for a product on a model

Clothing, accessories and anything held in the hand need a person to show scale and use. [Nano Banana Pro](/models/nano-banana-pro) renders people better: hands and faces look more natural.

```
A young woman in a beige trench coat walks down an autumn city street in the
morning, holding this travel mug at chest height. The mug is clearly visible
and in sharp focus, her face natural and relaxed. Keep the shape, matte sage
green color, bamboo lid, grey slider and steel rim exactly as in the
reference. Daylight, photorealistic lifestyle photo, no text, no logos.
Vertical 3:4
```

![A young woman in a beige trench coat and scarf walking down an autumn street with a green travel mug in her hand](https://ai2face.com/_astro/model.BqtE5Yr1_ZtTeX5.jpg "Person with the product — Nano Banana Pro. The fingers and grip look natural and the mug is in focus")

For clothing the setup is the same, but the reference is a photo of the garment, and the prompt describes the fit, the model's height and what to show: length, cut, fastening. If you need the same face across every image, the AI2Face character builder is easier: it saves the character and drops it into any scene. More on shooting people in the [AI photoshoot](/blog/ii-fotosessiya) guide.

## 6. Prompt for a product background

Swapping the background is the cheapest way to refresh a listing for a season or a sale: same product, new setting.

```
Replace the background: put this travel mug on a white wooden surface in a
cozy winter holiday setting — fir branches, a few pine cones, the edge of a
knitted sweater sleeve, warm fairy lights softly blurred in the background.
Keep the mug exactly the same: shape, matte sage green color, bamboo lid,
grey slider, steel rim. Photorealistic product photo, no text, no logos.
Vertical 3:4
```

![A travel mug on a white wooden surface among fir branches, pine cones and a knitted sweater, with blurred fairy lights behind](https://ai2face.com/_astro/seasonal.DEtg2uFx_2n1Ifw.jpg "Seasonal background — Seedream 5.0 Pro. The mug is unchanged; only the setting is new")

To make a set for another season, change only the setting line: “autumn leaves and a plaid blanket,” “a summer picnic on the grass,” “a black background with neon lighting” for a sale.

## 7. Prompt for a product video cover

A short clip starts from the finished hero shot: the photo is the first frame and the prompt describes the motion. That's a job for video models — [Kling](/models/kling), [Veo 3.1](/models/veo-3-1) or [Seedance](/models/seedance).

```
The first frame is this photo. The camera slowly orbits the travel mug by
half a turn, soft studio light glides over the matte body, light steam rises
from the open hole in the lid. The mug doesn't change shape or color. White
background, no text, vertical 3:4
```

Move the camera and the light, not the product: when the object spins on its own, models distort its shape more often.

## 8. Prompts by product category and for Amazon, Etsy and Shopify

The formula from section 1 is the same for any product; what changes is what the buyer needs to see.

| Category | What to add to the prompt | Key images |
|---|---|---|
| Clothing and shoes | fit, length, fabric close-up, model height | on a model, details, dimensions |
| Beauty | product texture, a swatch, packaging in a bathroom | details, lifestyle, ingredients |
| Electronics | ports, what's in the box, in the hand for scale | details, contents, in hand |
| Home and kitchen | table setting, an interior scene, material | lifestyle, dimensions, details |
| Kids' products | a calm palette, a play scenario, safety | in use, benefits |
| Food | ingredients, a cross-section, a serving on a plate | details, ingredients, serving |

What changes between Amazon, Etsy, Shopify stores and other platforms is the aspect ratio, image size and the rules for text on images. Every marketplace publishes them in its seller help, so check there and add the right ratio to every prompt. For reference, here's what Russia's largest marketplaces specify — we checked their official help pages:

| | Ozon | Yandex Market | Wildberries |
|---|---|---|---|
| Aspect ratio | vertical 3:4 in all categories | images are fitted to 3×4 on the storefront | see seller help |
| Resolution | clothing, shoes, accessories — from 900 × 1200 px | from 300 × 300 px | see seller help |
| File | up to 10 MB, JPEG, PNG, HEIC, WEBP | up to 10 MB, JPG, PNG, HEIC, WEBP | see seller help |
| Background | — | white or transparent is best | — |

Yandex Market also lists what must not appear on product images: price tags, discounts, store or marketplace logos, watermarks, contact details, links and delivery info. On-image text can't say “best,” “perfect,” “original” or make comparisons like “alternative to.” For an infographic, that means stating benefits as facts (“Keeps drinks hot for up to 6 hours”) rather than claims (“The best travel mug”) — a good habit on any platform.

## 9. Common mistakes in product photo prompts

- **Generating the product from scratch.** The model draws a similar but different item. Always use a real photo as the reference.
- **Not locking the product.** Without a line like “keep exactly: shape, color, details,” models “improve” the product — they shift the shade, proportions or hardware.
- **Too much text.** The model can render a paragraph, but buyers won't read it. Keep it to 3–4 lines of 2–4 words.
- **Made-up numbers.** The model will print any dimensions. Check every number against the specs.
- **Other brands' logos in frame.** Write “no brand logos” and check tech, clothing and packaging in the background.
- **Inconsistent style.** Background, palette and font should repeat from prompt to prompt — otherwise the listing looks assembled from different stores.
- **Too many edits at once.** If an image is almost there, change one thing: “move the headline down,” “make the background lighter.” The model keeps the rest. More on step-by-step edits in the guide to [Nano Banana prompts](/blog/nano-banana-pro-prompts).

## Bottom line

A working product photography prompt is the product photo, a “keep exactly” line and a description of one image with its text in quotes. With that approach, one kitchen snapshot became seven listing images, and the mug matched the source in all of them. Give text and infographics to GPT Image 2.5, scenes and people to Nano Banana 2 and Pro, seasonal backgrounds to Seedream. Check every image: the model fills in details that weren't in the photo, like the lid mechanism in our test. In AI2Face all of these models sit in one generator, the cost is shown before you generate, purchased credits don't expire, and new accounts get starter credits for the first images.

## FAQ

### Should I write product photo prompts in English?

English or any other language works — the models understand both. Write the on-image text in the language your buyers read and put it in quotes so the model reproduces it verbatim.

### Which AI model follows product photography prompts best?

For images with text, GPT Image 2.5: in our test every Russian-language label came out without errors. For lifestyle shots and people, Nano Banana 2 and Nano Banana Pro; for seasonal backgrounds, Seedream. All of them are available in the AI2Face generator.

### Can I make a listing image from a prompt without a product photo?

You can, but you shouldn't. The model will draw a similar but different product, and the buyer won't receive what they saw. Upload a real photo as a reference and state in the prompt that the product must not change.

### Do marketplaces allow AI-generated product images?

We didn't find an outright ban in the help pages of Ozon and Yandex Market, and Yandex Market accepts quality photos and realistic illustrations. The key condition everywhere is that the image shows the same product, color and contents the buyer will get — check your marketplace's own rules before uploading.

### Do these prompts work for Amazon, Etsy and Shopify?

Yes. The prompts aren't tied to a platform; only the aspect ratio changes. Set the ratio your marketplace uses and check image size, file weight and text rules in its seller help.

## Sources

- https://developers.openai.com/api/docs/guides/image-prompting
- https://ai.google.dev/gemini-api/docs/image-generation
- https://docs.ozon.ru/global/products/requirements/media/image-requirements/
- https://yandex.ru/support/marketplace/ru/assortment/create/main-fields/images