Google Shopping Image Requirements for 2027: Size, Framing and AI Images

Google Shopping Image Requirements in Short

Google Merchant Center will require product images of at least 500 × 500 pixels for all products from 31 January 2027. These are the main rules on Google's image requirements page, checked on 7 November 2026:

The size rule is the one with a deadline, so the practical job between now and January is to find every product image below 500 × 500 and replace it, without breaking the other rules on the way.

About the images in this guide: the cover image is AI-generated.

The 500 × 500 Minimum From 31 January 2027

Google's page announces "new image size requirements of at least 500 x 500 pixels for all products beginning January 31, 2027". It gives no separate minimum for clothing or any other category, so the 500 × 500 floor applies to every product in your feed.

Images below the minimum are the ones to look for first:

• Old catalog photos taken for small web thumbnails years ago.
• Supplier images downloaded at the size the supplier's site showed them.
• Images resized by your shop platform or a plugin, where the URL in your feed points to a small version instead of the original.
• Cropped variant photos, cut from a larger group shot.

Check the image URL your feed actually sends in image_link, not the photo you remember uploading. If your shop serves several sizes of the same image, the feed may point at a small one even when a large original exists.

Recommended Size and Framing

Google recommends images of 1500 × 1500 pixels or above. Two framing rules matter as much as the pixel count:

1. Fill the frame, but not completely. The product should take up no less than 75% and no more than 90% of the image. A small product in the middle of a large empty background meets the pixel rule and fails the framing rule.
2. Show the product clearly and alone. Google asks for "an unobstructed, clean view of the item for sale" and no products that are not sold together with it. Add other angles and close-ups through additional_image_link, and room-setting photos through lifestyle_image_link.

Google also notes that it "may test subtle cropping on your image", so keep important parts of the product away from the edges.

What Gets an Image Disapproved

Google lists these as reasons for disapproval or as things not to do:

• Promotional text and graphics added after the shoot: calls to action, prices, shipping details, promotional adjectives or condition text such as "New".
• Watermarks, copyright symbols and graphic stamps, including stock-photo watermarks.
• Logos or store names added in post-production. A logo that is a physical part of the product is fine.
• Barcodes, QR codes and serial tags laid over the photo.
• Borders around the image.
• Placeholder images, or one general photo reused for distinct products that need their own images.
• A single-colour square instead of a product photo.

Google's page lists exceptions to the generic-image rule for three categories: Hardware, Vehicles & Parts, and Computer Software.

AI-Generated and AI-Edited Product Images

Google's page sets a metadata rule for generative AI: "All images created using generative AI must contain meta data indicating that the image was AI-generated", with the IPTC Digital Source Type tag as the example. It also says not to remove embedded metadata such as the IPTC Digital Source Type property, and it names three codes to keep: TrainedAlgorithmicMedia, CompositeSynthetic and AlgorithmicMedia.

Two practical consequences:

• Don't strip the tag on the way to your feed. Image optimizers, export presets and some shop platforms remove metadata when they compress or resize a file. Check that setting in every step between the AI tool and the URL in your feed, and spot-check a final image with a metadata viewer such as ExifTool.
• Accuracy still comes first. Google requires the photo to reflect the product's exact colour, pattern, finish and material. An AI image that changes any of these misrepresents the product, whatever its metadata says.

What Roomagen writes into its results. Every Roomagen result carries machine-readable AI metadata. Results that Roomagen writes as a new file carry the IPTC Digital Source Type tag, and results saved exactly as generated keep the generator's signed Content Credentials instead. In the IPTC tag, images created from a plan or sketch are marked as created with generative AI (TrainedAlgorithmicMedia). Results of the tools that edit a photo, including upscaling, are marked as edited with generative AI, the IPTC code for a composite with AI elements. Google's page does not mention that "edited" code. If your feed contains AI-edited product photos, check Merchant Center's current guidance on edited images, and keep the tag in place either way. Our guide to AI labels and metadata explains the marks and how to check them.

Fixing Small Product Photos

Work through the options in this order, from most to least faithful:

1. Find the original file. The camera file or the supplier's full-size image is often still somewhere: the photographer's delivery, the supplier's download area, or your shop's original upload before resizing.
2. Reshoot the product. For your best sellers, a new photo at full resolution solves the size, framing and accuracy rules at once.
3. Upscale with AI, then check it. An AI upscaler enlarges a small image and adds detail as it goes. With Roomagen's AI Image Upscaler, a run costs 2 credits and renders at 4K on paid plans, never smaller than the original's long edge. It works best from about 1,000 pixels on the longest side; a very small image gives the AI less to work from. Because the AI generates new detail, look at the result at full size before it goes into your feed: text on labels and packaging, stitching, wood grain, fabric weave and fine patterns are the parts most likely to change. If anything on the product looks different from the real item, don't use that image.
4. Re-frame after enlarging. Once the image is large enough, crop or extend the canvas so the product fills 75–90% of the frame.

Google's own page also points merchants to Product Studio in Merchant Center, which it says can increase resolution, remove backgrounds and generate scenes.

Lifestyle and Room-Setting Images

For furniture, lighting and decor, Google's lifestyle_image_link attribute takes an in-context photo of the product. These images are where AI room settings are most useful: a sofa shown in a furnished living room tells shoppers more about its size and colour than a cut-out on white.

If you place a product into a room photo with an AI tool, such as Roomagen's Add Furniture & Objects, which can take a reference photo of the piece, check the product itself carefully. Its shape, colour, legs, handles and fabric must match the real item. Keep the AI metadata in place, and keep the plain product photo as your main image_link.

Checklist Before 31 January 2027

• List every image_link in your feed with its pixel size, and mark those below 500 × 500.
• For each small image, look for the original file before you upscale or reshoot.
• Check that the product fills 75–90% of every main image.
• Remove promotional text, prices, watermarks, added logos, barcodes and borders.
• Give every variant its own image showing only that variant.
• Compare AI-upscaled or AI-edited images with the real product at full size: colour, pattern, finish, material, any text.
• Make sure no step between your AI tool and your feed strips the IPTC metadata.
• Add alternative angles in additional_image_link and room-setting photos in lifestyle_image_link.
• Re-check Google's page before the deadline: the rules can change, and this guide reflects them as of 7 November 2026.