Amazon's product photo requirements for 2026 fit in one sentence. Pure white background, product filling at least 85% of the image, at least 1,000 pixels on the longest side for zoom, and a real photograph of the real product as the main image. If your main image does not comply, Amazon can pull your listing from search results until the image is replaced.
The requirements themselves have not changed dramatically. What did change in 2026 is who is looking at your photos. Amazon puts an AI assistant between the customer and your listing, shows images it generates itself in the search bar, and since July asks for a label on AI-generated people in your images.
This article lists the technical requirements, shows how to use the nine image positions, and where the difference between photography, CGI and AI imagery actually matters.
Amazon Product Photo Requirements: The Technical Specifications
Amazon's product photo requirements are concrete. The main image, the one customers see first in search results and on the product page, is held to stricter rules than the additional images. Amazon publishes them in its own seller guide and in the image guide in Seller Central.
| Requirement | Specification | Notes |
|---|---|---|
| Main image background | RGB 255, 255, 255 (pure white) | Not off-white, not light gray, the value is exact |
| Main image product fill | At least 85% of the image | Product large and centred |
| Main image content | Professional photograph of the actual product | No illustration, mockup, text, logo or watermark |
| Minimum size | 500 pixels on the longest side | Amazon rejects anything smaller |
| Zoom | From 1,000 pixels on the longest side | Seller Central recommends 1600 pixels or more |
| Maximum size | 10,000 pixels on the longest side | Practical working range: 2000 to 3000 pixels |
| Colour mode | sRGB (CMYK is accepted) | CMYK is for print, colours shift on screens |
| Number of images | Up to nine per product | Amazon recommends at least six |
| File format | JPEG, TIFF, PNG or non-animated GIF | Amazon prefers JPEG |
Sources for the table: Amazon's photo guide for sellers lists 85% fill and 500 to 10,000 pixels on the longest side, plus pure white, the file formats and the advice of at least six images. Amazon's guide to product listings gives the maximum of nine photos. The rule that the main image is a photograph of the actual product, and the 1600-pixel recommendation for zoom, are in the image guide in Seller Central.
Size is where most sellers trip up. Amazon itself says it prefers images larger than 1,000 pixels so customers can zoom in. In categories where material, texture or finish drive the purchase, such as clothing, home goods and jewellery, a weak zoom costs conversion directly.
The practical working range is 2000 to 3000 pixels: enough for zoom, without needlessly heavy files. If you also sell on bol or through Google Shopping, deliver square at a minimum of 1500 pixels. Why that is the only size that works on all three channels at once, and why your photographed white is almost never 255, is covered in why your white is not 255.
What Happens If Your Main Image Does Not Meet the Requirements?
A main image that does not comply can lead Amazon to suppress your listing. The product then disappears from search results until you upload an image that does comply. You will find such a listing in Seller Central under suppressed listings, with the reason attached. Amazon mentions this possibility itself in the image guide in Seller Central.
An earlier version of this article said Amazon had started enforcing more strictly in 2026 and was suspending listings without warning. We could not find that at Amazon itself. Amazon does not publish how often or how automatically it checks images. The stories about a new, stricter round of checks come almost entirely from parties selling a tool or service to fix it. So we removed that claim.
What we can say: in the two documented cases of a suppressed listing we found during our research into white backgrounds, it happened when images were replaced on a listing that was already running, not at the first upload. That is exactly the moment you touch files while revenue depends on them. So check a new set before you swap it in, not after.
And one genuinely new rule arrived in 2026. Since 22 July 2026 Amazon has asked sellers to flag images and videos containing a photorealistic person generated entirely by AI with the keyword contains-synthetic-performer in the file's metadata. Amazon then shows customers a notice where applicable. The trigger is a New York law that since June 2026 requires disclosure of such synthetic performers in advertising (CNBC, July 2026). Real people edited with AI and images without people are not covered.
The lesson for sellers: the technical requirements are not an optional quality upgrade but the price of entry. Anyone on Amazon with non-compliant images carries an operational risk the moment they change something on the listing.
How to Use the Nine Image Positions Strategically
Amazon gives you up to nine image positions and itself recommends at least six. Many sellers fill three or four, often with the same white shot from a slightly different angle. That is a misunderstanding of what product photos do in the purchase process.
A customer on your product page works through a fixed set of questions. Good images answer them in the right order, not at random and not repetitively. How much that matters is measured by Baymard Institute: 42% of users try to work out the size of a product from the product images, and 37% of the sites studied offer no image at all that shows scale.
The structure varies by category. Below is the information hierarchy we use in the listing design for K&G Goods and Evite & Sammit: main image, context, explanation, dimensions.
Position 1, main image: white background, product filling at least 85% of the image, sharp and bright, and a photograph of the actual product. This is the image shown in search results. The only question it has to answer: "Is this the product I am looking for?"
Position 2, second view: customers want to know what the product looks like from every side. A back view, side view or top-down view gives three-dimensional insight.
Position 3, scale shot: how big is the product? A photo with a hand, a person or a familiar object next to it removes that uncertainty. According to Baymard this can also be a computer-generated model or object, as long as the proportions are right.
Position 4, detail shot: a close-up of the strongest or most distinctive detail. Stitching on clothing, finish on furniture, material on electronics. This is the position that defends a premium price.
Positions 5 and 6, use and information: position 5 shows the product in use, in a recognisable setting. Position 6 summarises dimensions, weight and compatibility in a readable infographic. The strict white-background rules do not apply to positions 2 to 9.
Position 7, variants: if your product comes in several colours, sizes or versions, show the full range here.
Positions 8 and 9, brand and trust: position 8 is mood: the product in an attractive setting, for customers who already understand the function. Position 9 builds trust: certifications, warranty or a specific quality claim you can back up.
For the structure of a complete Amazon listing, also see our product listing design service.
Real Photo, CGI or AI Image: When Does It Matter for Conversion?
Short answer: for the main image the production method formally does matter, because Amazon asks for a photograph of the actual product there. For the additional images what counts is whether the image is honest and convincing, and there context and consistency beat technique.
71% of Dutch people find it hard to tell AI images from real photos, according to Norstat research among just over a thousand Dutch people (November 2025). For one AI image, as many as 81% of participants thought it was a real photo. Kjell Massen, commercial director at Norstat, explains why: "Real photos often contain imperfections, such as poor lighting or odd shadows. It is precisely those imperfections that make something real." (translated from Dutch)
That number is relevant, but also misleading. It is about the conscious question "is this real?". The question that drives conversion is a different one: does this image convince me the product does what it promises?
Research that Twinkle covered in February 2026 shows that contextual product images can lift conversion considerably: in one case by 94 percent compared with images without context, and on a specialised retail site under controlled conditions by up to 435 percent. Interactive presentation (zoom, 360 degrees) delivered up to 54 percent more sales, with the strongest effect on more expensive products. The same research saw a different picture on Amazon: there, products that stuck to the platform's white background did slightly better. Context wins, but only within what the platform expects.
A second reader has also arrived in 2026. In May 2026 Amazon replaced its Rufus assistant with Alexa for Shopping, for now for U.S. customers. That assistant compares products, writes AI summaries on product pages and lets customers search with a photo. At the same time Amazon shows images it generates itself based on search queries in the U.S. (TechCrunch, June 2026). How such an assistant reads your listing is covered in optimising your Amazon listing for AI.
The practical split per image position:
| Position | What Amazon asks | Strongest production method | Why |
|---|---|---|---|
| 1, main image | Photograph of the actual product on pure white | Photography, retouched to 255 | The rule says photograph; a deviating render is a risk |
| 2 and 4, views and detail | Accurate image of the product | Photography or CGI | Material and finish must match what the customer receives |
| 3, scale | Accurate image, context allowed | Photography or CGI with correct proportions | Proportion counts, not technique |
| 5 and 8, use and mood | Accurate image, context allowed | Photography where trust carries the purchase, AI background for functional products | Trust weighs more the more personal the product |
| 6 and 7, information and variants | Readable and accurate | CGI or infographic | One 3D model delivers every variant and angle |
| Any image with an AI person | Label in the metadata | Photography with a real model, or AI with a label | Required on Amazon since 22 July 2026 |
For images with context (positions 5 and 8): real photography performs better in categories where trust carries the purchase, such as clothing, jewellery, baby products and personal care. AI backgrounds work better for purely functional products, where a clean image is more believable than a messy real setting.
For information images (positions 6 and 7): the production method hardly matters. CGI or a schematic infographic is the strongest choice here.
Anyone using AI imagery is, by the way, buying a licence rather than a file, with terms on ownership and exclusivity that differ per tool. What those terms say is covered in AI product photos: you are buying a licence, not an image.
What Does Amazon-Compliant Product Photography Cost in 2026?
A standard packshot costs 8 to 15 euros per image at Dutch studios, a studio day a little over a thousand to almost two thousand euros, and an AI variant a fraction of that. The deciding factors are production method, volume and how much retouching it takes to get the main image to pure white.
An earlier version of this article had a price table here that we could not back up, including a fixed amount per AI image and a floor of five euros per packshot that appears in no Dutch price list. We now only use amounts that studios publish themselves.
| Method | Cost | Source |
|---|---|---|
| Packshot on white, per image | 8 to 15 euros, usually with a minimum order | Published price lists, see our cost overview |
| Range of 20 to 30 products | Roughly 600 to 1,500 euros | Derived from the same price tiers |
| Studio day | 1,150 to 1,799 euros | Published day rates, see our cost overview |
| Complete lifestyle project | 1,500 to 3,000 euros | Published rate of a studio, see our cost overview |
| AI background or AI variant | A fraction of a packshot, per attempt | Subscriptions worked out in AI product photos: a licence, not an image |
| CGI and 3D model | By quote, highly dependent on complexity | No public price lists found |
The saving with AI sits in the execution, not the concept. Someone still has to decide which angle, which light and which promise the image makes. Without those rules you get forty images that all look slightly different.
The right choice depends on your category, your volume and the role of trust in the purchase. A generated background is enough for a functional product in a price-driven category. For a jewellery brand where craft and material are the selling point, it does not work.
CGI as a Hybrid Option for Products with Variants
CGI (computer-generated imagery) is strongest on Amazon in the additional image positions, for products with many variants. One 3D model delivers every colour, size and angle, without a new shoot. For the main image a photograph remains the safe choice, because Amazon asks for a photograph of the actual product there.
The maths is simple. Traditional photography for a product in eight colours and three sizes means 24 setups, each with preparation and retouching. With CGI you build one 3D model and then swap colour and size in hours. The model itself is a bigger first investment than a single shoot, but from a handful of variants onwards you earn it back.
For hard-surface products CGI comes very close to photography in our experience: electronics, household appliances, furniture, sporting goods. The material is predictable and light can be simulated consistently. With fabric, skin, food and anything with a soft or organic texture, photography is still more convincing.
The hybrid approach combines the best of both: real photography for the main image and the mood shots (positions 1, 5 and 8), CGI for variants, angles and technical views (positions 2, 3, 4, 6 and 7). Your brand feels real through the photography, while the number of variants stays manageable.
At Evite & Sammit the foundation is real photography: packshots and lifestyle shots with representative models, supplemented with infographics, dimension diagrams and a comparison table within the same product line. That structure keeps the listings consistent across the whole catalogue.
For products still in development, CGI is also the only way to have early imagery: the physical product does not need to exist yet. How CGI compares with AI imagery, and where each belongs, is covered in CGI versus AI product visualisation.
When Should You Invest in Professional Photography?
Professional photography is not a budget decision but a strategic one, and it depends on three factors: how much trust the purchase requires, which price segment you are in, and how many variants you have.
Factor 1: trust as a purchase criterion. In categories where customers doubt authenticity (how does this feel, how does it fall, is this quality?) real photography is stronger. Daylight on fabric, the grain of wood, the sheen of a metal finish: those are signals AI imagery and CGI do not yet reproduce convincingly everywhere.
Factor 2: price segment. Premium products need premium imagery. Customers who pay more expect more proof of quality. A premium product with mediocre photos loses its price justification in the first second on the page.
Factor 3: scale and variation. Few products in few variants: traditional photography is efficient and the quality is high. Many products in many variants: CGI or AI becomes more efficient above a certain volume, with photography for the main image and the images that have to provide proof.
The question is not "real photo or AI". The question is: what convinces my customer, what does Amazon allow in which position, and what does that cost per sale? Answer those three questions for your own category and you reach the right conclusion.
For tailored advice on which approach fits your products and Amazon strategy, see our product photography service.
