Amazon Image Generator: The 8 Rules That Suppress You

The Best Amazon-Specific AI Image Tools, Compared

None of these tools, including StoreClaw, can certify that a specific image will pass Amazon's review — that call belongs to Amazon's own automated and manual systems, not any third-party tool. What they differ on is how close the default output lands to compliant before you check it yourself: a dedicated white-background editor like Photoroom or Pixelcut starts closer than a general-purpose lifestyle generator, and a planning tool like GreenOnion helps you avoid missing an image type across a full set, but every single image still needs a check against the exact rules below before it goes live.

The right pick usually comes down to catalog shape more than raw feature count. A solo seller with a handful of SKUs gets more out of a cheap, fast cutout tool used carefully than a full planning platform; a private-label brand running dozens of SKUs with the same visual system benefits more from something like Nightjar's consistency controls, where the cost of getting one rule wrong multiplies across the whole catalog instead of one listing.

Amazon's Own Technical Specs for Product Images

These are Amazon's own documented numbers, not general web guidance.

Amazon's Seller Central Product Image Guide sets a 500px minimum on the longest side just to upload at all, and a 10,000px maximum. Zoom functionality — the feature that lets a shopper hover and see fine detail — only activates at 1,000px or more on the longest side, and Amazon's own recommendation runs higher, toward 1,600px and beyond. Resolution needs to be at least 72 DPI, file size capped at 10MB, and accepted formats are JPEG (Amazon's stated preference), PNG, TIFF, or non-animated GIF — layered PSD or layered TIFF files aren't accepted; everything needs to be flattened first. Color profile should be sRGB; CMYK is technically accepted for standard listing images but never for A+ Content, which is RGB only.

On image count, Amazon's own guidance recommends at least 6 additional images plus 1 video alongside the main image — commonly summarized as 7 images and a video per listing. If you're including a video, the recommended dimension is at least 1920×1080px at a 16:9 aspect ratio. An AI-generated image can technically be exported at any of these settings, but most generators default to a resolution and format chosen for on-screen preview quality, not for passing Amazon's exact upload gate — worth checking the export settings directly rather than assuming the default is already compliant.

The 8 Reasons Amazon Suppresses a Listing Image

Amazon's own "Image suppression" page documents exactly these 8 reasons.

This is the exact list from Amazon's own Seller Central documentation, not a general summary of "what usually goes wrong":

1. Text, logo, graphics, or a watermark on the main image.
2. A background that isn't pure white — not exact RGB (255, 255, 255). Amazon's automated check can fail an image at RGB (254, 254, 254), a value that looks identical to the eye.
3. The product still in its packaging, or shown with brand or swing tags attached.
4. The product too small in the frame — failing the 85% fill requirement.
5. Prurient image content.
6. A human model in the main image for a non-clothing product — this applies to Softlines categories like Kids, Baby, Shoes, Luggage, and Watch and Jewelry, with Adult Apparel as the exception.
7. A mannequin or hanger in the main image — not permitted for Apparel or Accessories.
8. A model in a non-standing position on any Apparel main image.

Reason 2 is the one that catches the most generated images, since an AI tool's "white" output can land close to white without being exactly RGB (255, 255, 255) — worth checking pixel values directly with a color picker rather than trusting how it looks on a monitor, especially since different screens render the same file with a slightly different color temperature. Reasons 6 through 8 are the ones a generic "good product photo" prompt has no way of knowing about on its own, since they depend entirely on which category the ASIN sits in — the same pose and setup that's required for one category is the exact thing that gets a different category suppressed.

The New Rule for AI-Generated People in Your Images

Mandatory since June 6, 2026, following New York's synthetic-performer disclosure law.

This is specific to AI-generated images and directly relevant if you're using an AI model shot instead of a real one: since June 6, 2026, Amazon requires sellers to tag any photorealistic AI-generated person in listing images or videos, following a New York state disclosure law. The tag itself is technical — an rdf:li value of contains-synthetic-performer inside the image's XMP dc:subject metadata field, added before upload. This is metadata embedded in the file itself, not a caption or a visible watermark, so it doesn't change how the image looks to a shopper.

The rule covers product listing images, A+ Content, Brand Story, Store images, and both A+ and advertising videos. For A+ Content specifically, there are two valid paths: embed the metadata yourself before uploading, or check the "AI-generated people" box in A+ Content Manager's Creative Assets tool and let Amazon apply the tag. For everything outside A+ Content — including your main and secondary listing images — tagging the file before upload is the only option; there's no equivalent checkbox there. An AI-generated image that doesn't contain a photorealistic person at all isn't affected by this rule, so a white-background product shot with no model in it is unaffected regardless of how it was produced.

Apparel: Live Model or Flat, Never Both

Men's and women's clothing: on a live model, standing upright, nothing else.

Men's and women's clothing main images must show the garment on a live human model, standing upright — no sitting, kneeling, leaning, or lying down, legs not spread wide, eyes open, and a neutral expression. Kids' and baby clothing runs the opposite rule: main images must be photographed flat, off-model, with no model at all, for safety and consistency reasons. Multi-pack apparel follows the same flat, off-model rule. Clothing accessories can't show any part of a mannequin in the main image, with stockings and socks as the usual exception.

This split matters most exactly where a seller sells across both segments — a brand selling adult and kids' versions of the same garment style needs two different main-image setups for what is otherwise the same product, and reusing one photography approach across both listings is a documented way to trigger reason 6 or 8 on the suppression list above.

Apparel main image prompt: "Photograph of a live human model standing upright, facing forward, neutral expression, wearing [garment], full body visible, pure white background RGB 255 255 255, studio lighting, no props, e-commerce main image style."

If the model itself is AI-generated rather than a real person, this is exactly the image that needs the synthetic-performer tag above before upload.

Jewelry and Grocery: Two More Category Exceptions

Jewelry: 80% fill, not 85% — and a slight angle beats flat top-down.

Grocery: the nutrition panel belongs in a secondary slot, not the main image.

Jewelry runs its own set of numbers, separate from the general rule: no mannequin or model at all, the image's longest side must fall between 500 and 2,100px rather than the general 500–10,000px range, and the product only needs to occupy at least 80% of the image area — not 85%. Necklaces are the one category allowed to have the product cut off by the frame edge on the main image, letting a seller highlight a pendant and part of the chain without showing the full loop. Rings tend to photograph better at a slight angle than flat top-down, which also helps show depth and setting detail that a flat shot compresses away.

Grocery and consumables carry a smaller but easy-to-miss category quirk: nutritional and ingredient panels belong in secondary image slots, not the main image, which should still just show the product itself against the pure white background like any other category.

Jewelry main image prompt: “Product photography of [jewelry item], no model or mannequin, product filling 80% of the frame, pure white background RGB 255 255 255, slight angle rather than flat top-down, sharp macro detail on metal and stone.”

Catch a Suppression Risk Before You Upload

Checked against Amazon's own 8 reasons: Background color, frame fill, and category-specific rules like model posture or mannequin use.

StoreClaw checks a new product image against Amazon's own documented suppression reasons — background color, frame fill, and category-specific rules like model posture or mannequin use — before it goes live, and flags a mismatch for review. It doesn't generate or edit the image and can't guarantee Amazon's approval; the fix and the final call stay with you, the same way they do for every other tool on this page.

A+ Content Image Specs by Module

Every module has its own exact upload size — none of them match the standard listing-image spec.

A+ Content also runs its own file limits, separate from the standard listing-image rules above: 2MB maximum per image rather than 10MB, RGB only (never CMYK), and JPG, PNG, or BMP as the accepted formats. Video modules within A+ Content are more forgiving on file size, allowing up to 200MB. Always confirm the exact pixel size in Seller Central's own upload window for the specific module you've selected — module sizes can vary slightly by template even within the same broad category, and the upload window's stated requirement overrides any general chart, this one included.

Draft the Listing Content Around a New Image Set

Once a new image set is ready, StoreClaw drafts the surrounding title, description, and alt text across your catalog, so the copy stays aligned with what the images actually show instead of describing an angle or feature the new set doesn't include. This matters more than it sounds across a catalog of category-mixed SKUs — a title referencing a lifestyle scene that got swapped for a compliant white-background shot is a small mismatch that adds up across dozens of listings if nobody's tracking which copy paired with which image version.

Common Mistakes

Trusting a monitor's "white" instead of checking exact RGB values — Amazon's automated check can fail an image at RGB (254, 254, 254), which looks identical to pure white on screen. Applying the 85% fill and pixel-range rules to jewelry — jewelry runs 80% fill and a narrower 500–2,100px range, not the general specs.

• Uploading an AI-generated model image without the synthetic-performer tag — mandatory since June 6, 2026 for any photorealistic AI-generated person in a listing image or video.
• Using a flat off-model shot for men's or women's clothing, or a live-model shot for kids' clothing — the two apparel categories run opposite main-image rules.
• Putting a nutrition or ingredient panel in a grocery listing's main image — that content belongs in a secondary slot; the main image should stay a clean product shot like any other category.
• Reusing the same A+ module image size across a different module template — module dimensions aren't interchangeable, and the exact requirement can vary even within what looks like the same general category of module.