Guide · 2026

AI Product Photography: The Practical 2026 Guide

You photographed your product on a table by the window. It looks fine. Then you put it next to a competitor's listing and it looks like a snapshot next to an advertisement. That gap is not your camera. It's lighting, background, framing, and consistency — the four things a studio charges for.

AI product photography closes most of that gap. Not all of it, and not for every product. This guide is about where it works, where it doesn't, and how to make the images you generate actually get found.

What "AI product photography" means in practice

There are three different things sold under the same name, and they behave differently:

Background replacement. The tool cuts your product out and places it on a new background — white, a marble surface, a garden. Your product pixels are preserved. This is the safest category, and it's what marketplaces are happiest with.

Scene generation. The tool builds an environment around your product: surfaces, props, light direction, depth of field. Your product still comes from your photo, but everything around it is generated.

Full synthesis. The model generates the product too, from a description or a rough reference. This is where listings get suspended, because the item in the picture is not the item in the box.

For anything you sell, the first two are the categories you want. The rule underneath every marketplace policy is the same: the image must represent the actual product the buyer receives. Keep the product real and you stay inside the lines.

Start with one clean source photo

AI removes the studio, not the input. A weak source photo produces weak output no matter which tool you use.

Diffuse light beats bright light. An overcast window or a white sheet over a lamp. Hard sun creates blown highlights that no tool can recover.
Shoot straight on, at product height. Shooting down at an object makes it look smaller than it is.
Fill the frame, but leave room. You want the product large and uncropped, with a little margin so the tool can reframe.
Clean the product first. Fingerprints on glass and dust on black plastic survive every generation step.
Shoot reflective items against something dark and plain. Watches, jewellery, and chrome mirror the room — including you.

The marketplace rules that actually get enforced

Each channel wants something different from the same product, which is why sellers end up with folders of nearly-identical files.

Amazon (main image). Pure white background — RGB 255, 255, 255. The product should fill at least 85% of the frame. No text, logos, borders, watermarks, or graphics. One straight-on view of a single item, no collages. Images need to be at least 1,000px on the longest side for zoom to work, and larger files give a better zoom experience. Necklaces are the known exception to the "nothing cropped" rule — a long chain may run to the frame edge. Secondary images are where lifestyle scenes belong.

Etsy. No white-background requirement at all. Etsy's thumbnails are square and its buyers respond to context — hands, texture, scale. Shoot big: Etsy allows zoom and buyers use it.

Shopify and your own store. Nobody enforces anything, which is the trap. Consistency is what makes an independent store look like a brand: same crop ratio, same background family, same light direction across every product.

Instagram, TikTok, Pinterest. Vertical or square, and the product needs to survive being viewed at thumb size on a phone in daylight.

A practical workflow: generate one clean white-background shot for marketplaces, then two or three scene variations for social and your own store, all from the same source photo. Three shapes — square, portrait, landscape — cover almost every placement.

Marketplace image requirements at a glance

Requirement	Amazon (main image)	Etsy	Your store / Shopify
Background	Pure white, RGB 255,255,255	No requirement — context is welcome	Your choice — keep it consistent
Product fills	At least 85% of the frame	No rule	Consistent framing across the catalogue
Minimum size	1,000px on the longest side for zoom (1,600px+ is better)	2,000px on the shortest side for crisp zoom (3,000×2,250 recommended)	High-resolution; match your theme
Props, text, logos	Not allowed on the main image	Allowed	Allowed
Number of photos	Multiple; the first is the main image	Up to 10 (5+ recommended)	As many as your theme supports
First image shape	Square-ish, product centred	Landscape or square (Etsy auto-crops a square thumbnail)	One consistent ratio
Worth knowing	Necklaces may be cropped to the edge; apparel rules vary by sub-category	Buyers rely on the zoom, so shoot large	Consistency is what makes a store look like a brand

Verified from Amazon and Etsy seller documentation, September 2026. Marketplace rules change — check your category's current guide before you shoot.

Getting the images found, not just made

Name the file like a human would search for it. brass-desk-lamp-matte-black.jpg, not IMG_4471.jpg.

Write alt text that describes the picture, not the keyword. "Matte black brass desk lamp on an oak desk beside a notebook" tells a search engine and a screen reader the same true thing.

Keep images on pages that have real text around them. An image with no context is an image with no ranking signal.

Use structured data honestly. Product schema for product pages, Article and FAQPage where they match what's visible on the page. A FAQPage block describing questions a visitor cannot see on that page is a rule violation, not a shortcut.

Serve modern formats and real dimensions. WebP or AVIF, correct width and height attributes, lazy loading below the fold.

Where AI answers get their picture of you

ChatGPT, Perplexity, Gemini and Claude increasingly sit between a seller and a buying decision. Three things decide whether your product shows up there:

Your pages must be readable without JavaScript. Most AI crawlers don't execute it.
You have to be described somewhere other than your own site — directories, listicles, forum answers, review sites.
Your claims have to be checkable. A specific, modest, verifiable claim survives being quoted; a vague superlative gets dropped.

When to hire a human photographer anyway

Honest answer: when the product's material is the product. Jewellery where the stone's fire is the selling point, fabric where the weave matters, food where steam and freshness carry the sale. AI is excellent at context and consistency. A good photographer is still better at making one specific object look like the best version of itself.

The realistic split for most small sellers: a photographer for the hero shot of your best seller, AI for the other forty listings that would otherwise stay on a window sill.

Where Shotvana fits

Shotvana takes one photo and builds the scene around it while keeping your product as it is. You type a word — garden, marble, dark luxe — or pick one of six styles, optionally add a model, and get a finished image in about a minute, in square, portrait, or landscape. There's a free trial of five photos, packs from $9, and monthly plans from $14. It exports PNGs up to 1536px, which clears Amazon's zoom threshold; if you need larger files for print, a photographer or an upscaler is still the right tool.

FAQ

Are AI-generated product images allowed on Amazon and Etsy?

Yes, as long as the image accurately represents the actual product. The line both marketplaces enforce is misrepresentation — a generated background is fine, a generated product is not.

Will AI photos hurt my ranking?

Marketplaces rank on conversion and compliance, not on how an image was made. A clean, compliant, honest image that gets clicks performs like any other.

What size should I export?

At least 1,000px on the longest side for Amazon zoom; larger is better. For Etsy and your own store, bigger files reward the buyers who zoom in.

Do I still need a photographer?

For hero shots of material-driven products, usually yes. For the long tail of listings, AI is faster and far cheaper.

How many images does one product need?

One compliant main image, plus three to five supporting shots: scale, detail, in-use context, and packaging if it matters.