How should I turn product photos into main and additional images for Google Shopping?

Start with the product, then build the image set

Use Dreamina to prepare product-led compositions while keeping the actual SKU as the source of truth. Start with the strongest real photo rather than asking a generator to invent a similar product from a description.

Google requires a main image through the image_link attribute. You can supply up to ten additional images through additional_image_link. That is a maximum, not a requirement to create ten pictures for every item. Four genuinely different views can be more informative than ten nearly identical renders.

1. Sort your original photos by SKU and variant

Choose clear photographs of the exact product the shopper will receive. Before editing, separate each color, pattern, material and configuration into its own group. A navy shirt and a black shirt need correctly matched product images even if they share the same layout.

For each group, identify a complete front or three-quarter view, genuine photographs of the other sides, useful detail shots, and a view of everything included in the purchase. Keep the unedited originals so you can compare labels, proportions and colors after processing.

Do not generate a missing back view if you do not know what the back looks like. AI can create convincing seams, connectors and controls that the item does not have. Photograph that angle instead. The same applies to accessories, packaging contents and a product’s customized appearance.

For packaged goods, show the actual package rather than using a flat label as a substitute for the product. Make sure the quantity shown matches the offer: one bottle, a pair or the correct multipack.

2. Make the main image clear and complete

Show the whole correct product with a simple composition and enough detail to inspect it. A white or light neutral background is a useful default for many products because it makes the shape easy to read. Keep the product’s proportions intact and leave enough space around its edges.

Google recommends framing the product so it occupies roughly 75–90% of the image. Use that as a composition guide, rather than stretching a tall or unusually shaped item to fill a square. The product should remain complete and recognizable at a small display size.

Keep promotional elements out of the image: price badges, shipping claims, calls to action, warranty messages, added watermarks, borders and overlaid logos. A real brand name printed on the product or its packaging is different from a logo pasted onto the photograph; preserve the physical branding that identifies the item.

Avoid props that could be mistaken for included products. A charger shown beside a device, for example, can suggest that it comes with the purchase. If it does not, choose a clearer composition.

There are category-specific exceptions to Google’s general preference for an actual product image. Certain hardware, vehicle parts and software can use appropriate generic graphics, while software logos and some paint color swatches have specific allowances. Use the rule for the item’s category rather than extending an exception across your whole catalog.

Choose an export size that will last

Use a high-resolution source and distinguish Google’s upcoming minimum from a useful production target. Google’s image guidance announces a 500 × 500-pixel minimum for all products from January 31, 2027; its format table already lists 500 × 500. For work prepared in November 2026, keep that transition date explicit instead of assuming the notice describes an already completed rollout.

A practical working target is 1,500 × 1,500 pixels or larger when your source supports it. That is a production choice, not a universal Google minimum. A rectangular product can use a suitable rectangular composition; do not distort it to meet a square template.

Google supports JPEG, WebP, PNG, non-animated GIF, BMP and TIFF images, with limits of 64 megapixels and 16 MB. The file extension should match the actual format. Enlarging a blurry source can increase its pixel count without restoring label detail, so return to a better photograph when the product itself is unclear.

3. Prepare the product image in Dreamina

Start with the real product photo, then edit the setting or composition with tightly specified instructions. Dreamina is useful for exploring a cleaner presentation or a contextual background while using the product image as a reference.

1

1. Upload the clearest photograph of the correct variant as your reference.

2

2. Describe the intended background, framing and lighting. Name the product details that must remain recognizable.

3

3. Generate or refine the composition using the image-editing tools available in your selected workflow.

4

4. Compare the result with the original before choosing it for the listing.

For a photographed coffee pouch, an instruction could be:

“Use the uploaded coffee pouch as the product reference. Place one upright pouch against a clean, light neutral background, with soft even light and a subtle contact shadow. Show the complete package. Preserve its shape, label layout and visible brand text. Add no other products, promotional text, badges or decorative graphics.”

This is a starting prompt, not a guarantee that every detail will be preserved. Inspect the zipper, bag shape, label lettering, weight statement and color. If the result changes the product, refine the edit or use the protected-layer approach below.

Once the main view is settled, use a separate prompt for a contextual image. Specify a plausible surface and scene, and keep other objects from looking like part of the offer. For a larger campaign, the same source-first approach also helps you keep one product consistent across AI-generated ad images.

Keep a protected product layer when details are critical

Reuse the original product pixels over a separate background when a generated redraw would risk changing the item. This is especially useful for small packaging text, intricate jewelry, precise stitching and recognizable hardware.

Make a clean cutout from the original photograph, refine its edges, and place it above a suitable background in a layer-based editor. Add a restrained contact shadow and adjust the scene around the product. Avoid reshaping the product to make it fit an implausible perspective.

This gives you more control over which parts change, although the final export still needs inspection for cropping, color and edge artifacts. If choosing the editing route is the main difficulty, compare the approaches in this guide to changing a product background while preserving product detail.

4. Give each additional image a different job

Use additional images to answer questions the main image cannot answer alone. Google allows more flexibility here, including staged or in-use images, close-ups, and images showing part of a bundle. Additional images still inherit the main-image requirements except where Google explicitly provides an exception.

These allowances do not make the gallery an unrestricted advertising space. Do not assume that discount graphics, delivery promises or a text-heavy promotional infographic become suitable simply because they are an additional image.

Google also provides a lifestyle_image_link attribute for suitable lifestyle imagery. Use the relevant field in your existing product-data setup, rather than treating every styled scene as the main image. Do not rely on a fixed gallery order to explain essential information: the selected main image should stand on its own.

An example image set for a ceramic mug

A useful set shows the actual mug from complementary views before adding a styled scene. For a hypothetical single-mug listing, you could prepare:

1

1. Main view: the complete mug against a light background, showing its true glaze and handle.

2

2. Opposite side: a real photograph that reveals any different pattern or finish.

3

3. Detail: a close-up of the actual rim, glaze or maker’s mark.

4

4. In-use view: a genuine photograph showing the mug held naturally, if scale matters to the purchase.

5

5. Optional styled scene: a carefully checked Dreamina background composition that keeps the mug accurate and does not imply that nearby items are included.

This is an example production plan, not a finished or approved image set. Skip any view that adds no new information. If you sell a pair, photograph the pair accurately rather than reusing the single-mug main image unchanged.

5. Choose supporting tools for the part of the job you need

Keep Dreamina as the creative starting point and use a specialist editor or Merchant Center tool where it saves a concrete step. The most useful choice depends on whether you need a new setting, manual edge control, a repeatable canvas or editing alongside your product data.

Google Product Studio: edit images inside Merchant Center

Product Studio is useful when you want image editing close to the product data you already manage. Its image tools can remove backgrounds, generate scenes and increase resolution. In Merchant Center, look under Creative content, then Product Studio.

The current guide lists Product Studio as available to Merchant Center users outside France and Russia, with eligibility and product limitations still applying. The tools are free to use where available. Merchants using Shopify can also encounter Product Studio through the Google & YouTube app.

After creating a suitable image, you can download it or add it to Merchant Center. The workflow distinguishes choosing a single image as the main image from adding multiple images as additional images. Check that selection deliberately; generating a scene does not establish that it is the best main view or that Google will approve it.

Adobe Photoshop: protect detail with layers and masks

Photoshop is a useful finishing route when the product edge, label or transparent material needs manual control. Keep the original photo on its own layer, use background removal or a selection to create a mask, and refine the boundary rather than repeatedly regenerating the entire product.

Place a solid background or separate scene below the product layer. Check small openings, translucent edges and reflections at full size. Save a layered working file so that a later background change does not require another product cutout.

This route is particularly useful when exact package text matters more than creating a completely new composition. Photoshop is a paid editing option; you do not need generative editing for every cleanup or background change.

Canva: standardize spacing and simple layouts

Canva can help you place a clean product cutout on a consistent canvas and prepare a repeatable layout. Its current background-remover page provides a free entry point and options to refine the cutout or continue editing.

Keep the product’s aspect ratio, use consistent margins across the catalog, and save the design as a reusable working project. A template is a starting layout: remove decorative badges, added logos and promotional text before preparing Shopping images.

Do not assume that every resizing, brand-management or bulk feature has the same free availability as the advertised background-removal entry. If you only need a reusable cutout, this comparison of free AI background removers can help you choose a suitable export route.

Photoroom: prepare product backgrounds efficiently

Photoroom is suited to product cutouts and background variations that you can compare against the source photo. Use a plain color for a simple presentation, a suitable uploaded background for controlled placement, or an AI background for a scene that needs more creative direction.

An AI reference image can guide the background’s look without being inserted literally. Generated scenes can also adapt lighting, so compare the product itself after each edit rather than assuming the original appearance is untouched.

Photoroom’s connected Shopify Product Catalog guide describes a store-specific workflow; its screens are not the same as a standalone upload for every user. Choose the route that matches your setup and inspect the exported file before adding it to your product data.

6. Preserve AI-origin information in the final files

Keep the required IPTC information when an image was created using generative AI. Google’s image guidance requires the appropriate DigitalSourceType metadata and says to preserve it in AI-generated images.

Relevant values include trainedAlgorithmicMedia for model-generated media, compositeSynthetic for a composite containing synthetic elements, and algorithmicMedia for media created by an algorithm that is not based on a trained model, as applicable to the file’s origin. A real product placed into an AI-generated setting still needs the appropriate treatment for the resulting composite.

Check the final export and the copy served by your image host. An editor’s project file, an exported image and a CDN-delivered version can carry different metadata. Do not assume your image tool or store automatically retains the required information, and avoid blanket “remove all metadata” export settings for this workflow.

Correct origin information does not make an inaccurate product depiction acceptable. Product accuracy and AI labeling are separate checks, and both matter.

7. Host the images and map them to the right fields

Use one main-image URL and put the other image URLs in the additional-image field. Upload the final files to a stable image host or your store’s image system, then use the direct image URLs in the corresponding product item.

A useful URL opens the image itself, not an editor preview, login page or product page. It should remain reachable without an expiring session. Make sure Googlebot and Googlebot-image can fetch it, and map every URL to the correct SKU and variant.

Keep filenames and URL handling straightforward. Use a supported HTTP or HTTPS URL and encode characters where necessary. If a URL itself contains a literal comma, encode that comma so it is not confused with the separator between multiple additional-image URLs.

A simple feed mapping example

Keep the main image separate from the list of additional images. For the hypothetical mug, the mapping could look like this; the example addresses below are placeholders, not uploaded files.

In a CSV or tabular data source, the additional URLs belong in the same additional-image field, separated by commas. For a CSV, quote that whole field so its commas do not create extra columns. In XML, repeat the additional-image element for each image instead of putting every URL into one element.

If your store or existing Merchant API integration supplies product data, use its supported image fields rather than starting a separate integration just to submit this set. Image creation and product-data delivery are connected steps, but generating an image does not automatically upload or map it.

8. Check the hosted files before expanding the batch

Inspect the exact images Google will fetch, then review the item’s image issues in Merchant Center. Begin with a small, representative group of products so a repeated crop or variant error does not spread across the catalog.

Open the final hosted images and compare each one with the original product. Check the complete outline, label text, color, included quantity and any accessories. Then check actual dimensions, format, file size, accessibility and the AI-origin metadata where needed.

In Merchant Center, look for issues associated with the affected item and image field. Processing an image is separate from creating the file, and a successful download is not the same as product approval. An additional image that fails its minimum requirements may not appear; do not assume every additional-image issue means the whole item has been disapproved.

Fix the cause of an image problem

Correct the file, product mapping or access setting responsible for the issue. Changing a filename alone will not fix misleading content or an inaccessible host.

Frequently asked questions

Do Google Shopping images have to be square and white?

Square crops and light backgrounds are useful defaults, but they are not a universal rule for every image. The product must be clear and accurately represented. Additional images can use permitted staged or in-use views, and category-specific rules can differ.

Can I use AI-generated lifestyle images?

Use a generated scene only when it depicts the actual item accurately and carries the required AI-origin information. Do not invent product features or create uncertainty about which objects are included. Use real photographs for details and angles you cannot verify from the source.

Do I need all ten additional images?

Use the number of images needed to answer useful product questions. Ten is the limit, not a quality target. Prefer distinct, truthful views and choose the main image deliberately rather than depending on a particular additional-image display order.

Can I use the same image set for every color?

Reuse the production layout, but show the correct color and variant in each item’s images. A real photo of the relevant variant is a stronger source than a generated recolor when finish, pattern or material also changes.

Build a clear listing from a truthful product image

Start in Dreamina with the actual product photo, then assemble a main image and a focused set of supporting views. Keep the main image easy to understand, use additional images to answer real purchase questions, and preserve the product details that distinguish the SKU.

Finish by checking the exported files, AI-origin information and product-data mapping. That gives you a repeatable process you can apply to the next item without treating an attractive generated scene as a substitute for an accurate product photograph.