Skip to main content
Prompt Engineering

AI Product Photo Prompts for E-commerce: Copy-Paste Templates

Copy-paste AI product photo prompts for studio shots, backdrop swaps and selfie-style ads, with the real prompts and results behind five examples.

LT

Lovino Team

September 29, 2026•7 min read
AI Product Photo Prompts for E-commerce: Copy-Paste Templates

There are two very different jobs hiding behind "AI product photos". One is inventing a product picture from text, which is good for ad concepts and mood boards. The other is editing a photo of your real product, which is the only way to keep your actual shape, colour and label. This page gives you prompts for both, and shows where each one goes wrong.

Product photo prompts at a glance

JobRouteCost on LovinoMain risk
Concept or mood-board shotText prompt to imageZ-Image Turbo, 1 creditThe product is invented, not yours
Your real product on a new backdropEdit your photo with an instructionFLUX Kontext Fast, 2 creditsColour and label drift
Selfie-style ad stillText prompt to imageZ-Image Turbo, 1 creditThe person is not a real customer

Route one: generate a studio shot from text

The hero image above came from this prompt. It is worth reading slowly, because each clause has one job.

Prompt: Studio product photo of plain sand-coloured headphones with no logo on a plaster plinth, soft editorial light, 85mm lens, main subject in the upper two-thirds of the frame, calm darker lower third left clear for text

Diagram splitting a real product prompt into product and surface, light, lens, framing and space for text
Diagram splitting a real product prompt into product and surface, light, lens, framing and space for text

The last two clauses are the useful trick: they ask for the subject in the upper two-thirds and a calm lower third "left clear for text", so a price or headline can sit there later. All the prompts in this section say "plain" or "unbranded" on purpose. A text prompt cannot know your product, so it draws a generic one. Here are two more, both made with Z-Image Turbo.

Unbranded amber perfume bottle on wet black stone with water droplets and dramatic side light, made with Z-Image Turbo
Unbranded amber perfume bottle on wet black stone with water droplets and dramatic side light, made with Z-Image Turbo

Prompt: No visible text, letters, words, numbers, logos or signage anywhere in the image. Studio product photo of an unbranded amber perfume bottle on wet black stone with water droplets, dramatic side light, 100mm lens, main subject in the upper two-thirds of the frame, calm darker lower third left clear for text

Plain white running shoe floating above a splash of water on a pale background, made with Z-Image Turbo
Plain white running shoe floating above a splash of water on a pale background, made with Z-Image Turbo

Prompt: Studio product photo of a plain white running shoe with no logo floating above a splash of water, clean pale background, high-speed flash, main subject in the upper two-thirds of the frame, calm darker lower third left clear for text

Route two: edit your own product photo

This is the route for a real SKU. The starting point is one AI-made photo of a sneaker on a cafe table. The middle panel used FLUX Kontext Fast, the right panel Nano Banana 2 Lite.

One AI-made photo of a sneaker on a cafe table, then two edits of it: a white studio backdrop with FLUX Kontext Fast and a white marble surface with Nano Banana 2 Lite
One AI-made photo of a sneaker on a cafe table, then two edits of it: a white studio backdrop with FLUX Kontext Fast and a white marble surface with Nano Banana 2 Lite

White studio: Replace the background with a clean seamless white studio backdrop and soft studio shadow. Keep the sneaker exactly the same.

Marble: Place this exact product in a white marble surface with soft natural window light, as a high-quality commercial product photo. Keep the product's shape, colours, proportions and any text already printed on it exactly as they are. Do not add any other text, letters, logos or signage.

Compare the two results before you copy either prompt. The white studio version removed the table and the cup and kept the sneaker's shape and cream details. The marble version kept the coffee cup, and it moved the laces and heel tab from cream towards grey, even though the prompt said to keep the colours. That is the check to run on every edit: put the result next to the original and look at colour, proportions and any printed text.

Selfie-style ad stills

Ad creative that looks like a phone photo of a person holding the product is a common format. This one is a prompt for the still.

Selfie-style photo of a young woman holding up an unbranded skincare serum bottle, made with Z-Image Turbo
Selfie-style photo of a young woman holding up an unbranded skincare serum bottle, made with Z-Image Turbo

Prompt: Smartphone selfie-style photo of a young woman in her bright bedroom holding up a small unbranded skincare serum bottle to the camera, main subject in the upper two-thirds of the frame, calm darker lower third left clear for text

The woman is generated, not a customer. Do not present a generated person as a real reviewer or as a testimonial, and follow your platform's rules on disclosing AI-made ads.

Template prompts to adapt

These are patterns, not tested prompts. Swap the words in [brackets] and check each result.

  • Studio: Studio product photo of a plain [product] with no logo on a [surface], soft editorial light, 85mm lens, main subject in the upper two-thirds of the frame, calm darker lower third left clear for text.
  • Water: Studio product photo of a plain [product] floating above a splash of water, clean pale background, high-speed flash.
  • Edit, backdrop: Replace the background with a clean seamless white studio backdrop and soft studio shadow. Keep the [product] exactly the same.
  • Edit, surface: Place this exact product on [surface] with soft natural window light, as a high-quality commercial product photo. Keep the product's shape, colours, proportions and any text already printed on it exactly as they are.
  • Selfie ad: Smartphone selfie-style photo of a [person] in a [room] holding up a small unbranded [product] to the camera.

Test one tricky product before you batch

Pick the hardest product you sell: something reflective, transparent or covered in printed text. Run one edit prompt on it and compare the result with the original for shape, colour and every letter. If it fails, change the prompt or the model and test again. Only batch the catalogue once the hard case passes. It is the same test we recommend for AI fashion photography, where fabric detail is the equivalent of a printed label.

Save one prompt per surface, not one per product. A good marble clause or white-studio clause can be reused across the catalogue, while the product name in the sentence changes.

Before you publish

  • Compare with the original. Colour, shape, stitching and printed text.
  • Read every label. Generated text can be wrong, and a changed logo can be a legal problem.
  • Check marketplace rules. Marketplaces publish their own image requirements, and some ask for plain backgrounds.
  • Keep the original file. You will want it when a listing needs an exact match.

What it costs

The Product Photo tool costs 2 credits per image, and the free plan's 3 daily credits (signup required) cover one run. The Background Remover costs 1 credit and returns a cut-out you can place on your own backdrop. Free-plan downloads are watermarked; paid plans start at $9.99 a month for 50 credits, or a pack at $8 adds 50 credits ($0.16 each).

Keep going

For the wider workflow read AI product photography for e-commerce and AI product photos for Shopify. If the product needs to appear in a video ad, see how to make UGC product ads with AI, and the e-commerce solutions page covers catalogue-scale work. Text on packaging is easier with GPT Image 2 prompts.

LT

Written by Lovino Team

Lovino's editorial team documents practical, reproducible workflows for AI image and video creation.

Follow on Instagram

AI-assisted media is identified in context. Product workflows are tested by the Lovino team; outcomes vary by prompt, model, and source material.

Read our editorial and testing policy

Ready to try it yourself?