Updated August 2026. This guide walks you through Google Merchant Center image requirements, the IPTC metadata rules for AI-generated content, and a step-by-step workflow for producing compliant product images at scale using AI tools.
Jump to:
Google Shopping Image Specs | AI Image Policy | Step-by-Step Workflow | Avocado AI for Google Shopping | FAQ
TL;DR
Google Shopping accepts AI-generated product images, but they must meet strict size, composition, and metadata rules. The minimum size is 500x500 px (rising from 100x100 for non-apparel and 250x250 for apparel), Google recommends 1500x1500 px or larger, and every AI-generated image needs IPTC DigitalSourceType metadata. Tools like Avocado AI let you generate compliant product images from a single photo at 1 to 2 credits per image, with no per-image licensing fees.
Google Shopping Image Specs
Google Merchant Center enforces specific image requirements for every product listing. Get any of these wrong and your product gets disapproved.
Primary Image (image_link)
| Specification | Requirement |
|---|---|
| Minimum size | 500 x 500 pixels (enforced from Jan 31, 2027; warnings active since Apr 14, 2026) |
| Recommended size | 1500 x 1500 pixels or larger |
| Maximum resolution | 64 megapixels |
| Maximum file size | 16 MB |
| Supported formats | JPEG, PNG, WebP, non-animated GIF, BMP, TIFF |
| Product fill | 75-90% of the image area |
| Background | Solid white or transparent recommended |
| Aspect ratio | Square (1:1) recommended |
Source: Google Merchant Center Help - Image link
Lifestyle Image (lifestyle_image_link)
| Specification | Requirement |
|---|---|
| Minimum resolution | 600 x 600 pixels |
| Aspect ratio | Between 2:0 and 2:3 |
| Promotional elements | Not allowed |
| Required? | No, but products with lifestyle images see higher click-through rates |
What Google Prohibits
These violations trigger immediate disapproval:
- Promotional text or overlays (sale badges, price information, calls to action)
- Watermarks, brand logos, or borders on the primary image
- Placeholder or generic images that do not show the actual product
- Products that fill less than 75% of the image frame
- Thumbnails or artificially upscaled images
- Multiple products in the primary image (unless sold as a bundle)
Upcoming Change: Universal 500x500 Minimum
Starting January 31, 2027, all product images must be at least 500 x 500 pixels regardless of category. Google has been issuing warnings since April 2026. If your current images are below this threshold, update them now to avoid disapprovals when enforcement begins.
Source: Google Merchant Center - Image requirements update
Google's AI Image Policy
Google explicitly allows AI-generated product images in Merchant Center. The company even built its own tool, Product Studio, directly into Merchant Center Next for background swapping, resolution upscaling, and image generation.
However, all AI-generated images must include IPTC metadata identifying them as AI-created. This applies to primary images, additional images, and lifestyle images.
Required IPTC DigitalSourceType Tags
| IPTC Code | When to Use |
|---|---|
| TrainedAlgorithmicMedia | Images created using AI models trained on sampled data (most AI generators) |
| CompositeSynthetic | Real product photo combined with AI-generated elements (e.g., AI background replacement) |
| AlgorithmicMedia | Images created purely by algorithm, not based on training data |
Source: Google Merchant Center - AI-generated content policy
Practical Implications
Most e-commerce use cases fall under two categories:
- Full AI generation (white-background product shots from text prompts or reference images): Use
TrainedAlgorithmicMedia. - AI editing (swapping backgrounds, removing objects, enhancing a real product photo): Use
CompositeSynthetic.
If you strip the IPTC metadata from an AI-generated image before uploading it to Merchant Center, you are violating Google's policy. Some AI tools embed this metadata automatically; others do not. Always check before uploading.
Step-by-Step Workflow
Here is a practical workflow for creating Google Shopping-compliant product images with AI.
Step 1: Start with a Clean Product Photo
The best AI-generated Shopping images start from a real product photo. Even a smartphone snapshot works if the product is well-lit and clearly visible. If you only have a text description, AI tools can generate images from prompts, but a reference photo produces more accurate results.
What to capture:
- The product on a neutral or white surface
- Even lighting without harsh shadows
- The product filling most of the frame
- No promotional text or overlays in the source image
Step 2: Choose an AI Image Model
Different AI models excel at different product types. For Google Shopping images, you need models that produce clean, photorealistic output at 1500x1500 px or larger.
| Model | Best For | Cost (Avocado AI) |
|---|---|---|
| GPT-Image 2 | Text rendering, photorealistic products, complex compositions | 2 credits per image |
| Nano Banana 2 | Fast, clean product shots with sharp detail | 1 credit per image |
| Recraft v4 | Design-focused renders, vector-ready product shots | 1 credit per image |
| Seedream V5 Pro | Dense layouts, multi-language text on images | 2 credits per image |
| SeedDream 5.0 Lite | Lifestyle and campaign product shots | 1 credit per image |
| Flux 2 Flex | Stylized social and product renders | 1 credit per image |
All of these models are available on every Avocado AI plan, from Intro (EUR 19.99/mo) to Pro (EUR 249/mo). See the full model list at Avocado AI pricing.
Step 3: Generate the Image
Use your AI tool to create a product image that meets Google's specifications. Key prompt guidelines:
- Background: Specify "solid white background" or "clean neutral background" in the prompt
- Product fill: Ask for the product to "fill 75-90% of the frame"
- Resolution: Request "1500x1500 pixels" or "high-resolution square product photo"
- No text or logos: Explicitly exclude promotional elements: "no text, no watermarks, no logos, no overlays"
Example prompt for a white-background product shot:
"Professional product photography of [product] on a solid white background. The product fills 75-90% of the frame. Clean, even studio lighting with soft shadows. Sharp detail, no text, no watermarks, no overlays. Square format, 1500x1500 pixels."
Step 4: Verify IPTC Metadata
Before uploading to Merchant Center, confirm the AI-generated image has the correct IPTC DigitalSourceType tag embedded. You can check this with:
- ExifTool (command line):
exiftool -IPTC:DigitalSourceType image.jpg - Adobe Bridge or Lightroom: Check IPTC metadata panel
- Python script: Use the
Pilloworexifreadlibrary to read IPTC fields
If the metadata is missing, add it before uploading. Tools like ExifTool can write IPTC tags: exiftool -IPTC:DigitalSourceType="http://cv.iptc.org/newscodes/digitalsourcetype/TrainedAlgorithmicMedia" image.jpg
Step 5: Resize and Format for Merchant Center
Ensure the final image meets all technical requirements:
- Dimensions: At least 1500x1500 px (square recommended for primary images)
- Format: JPEG or PNG (most reliable across Merchant Center)
- File size: Under 16 MB
- Product fill: 75-90% of the frame area
If your AI output is not the right dimensions, use a resize tool. Most AI platforms return images at 1024x1024 or similar sizes. Upscaling to 1500x1500 is acceptable as long as the image remains sharp.
Step 6: Upload to Merchant Center
Upload the image as the image_link attribute in your product data feed. You can also submit up to 10 additional images via additional_image_link and 1 lifestyle image via lifestyle_image_link.
Products with multiple images (3 or more) see significantly higher engagement than single-image listings.
Using Avocado AI for Google Shopping Images
Avocado AI is a workspace for generating product images, videos, music, and more. Here is how it fits the Google Shopping image workflow.
What Avocado Offers
- Multiple image models in one workspace: GPT-Image 2, Nano Banana 2, Recraft v4, Seedream V5 Pro, and more. Pick the model that fits your product type.
- Prompt-to-edit canvas: Generate a base image, then refine it with draw-to-edit tools without switching platforms.
- Batch generation: Generate images for multiple products in a single session using the Studio or Flows.
- No watermarks on any plan: Every generated image is clean and ready for Merchant Center upload.
- Commercial usage rights: All plans include commercial licenses for generated content.
Cost per Image
At 1 credit per image for Nano Banana 2 or Recraft v4, the cost works out to approximately EUR 0.10 to 0.13 per image on the Starter plan (300 credits for EUR 39/mo). For a catalog of 100 products needing primary and additional images, that is roughly EUR 10 to EUR 13 total for all product images.
What Avocado Does Not Do
Avocado AI is not a dedicated Google Shopping feed manager. It does not:
- Automatically upload images to Merchant Center
- Manage product data feeds
- Handle IPTC metadata embedding (you need to add this post-generation)
- Replace a feed management tool like DataFeedWatch or Channable
The value is in the image generation itself: fast, high-quality product images at scale, from a single workspace that also handles your video ads, social content, and brand assets.
Workflow with Avocado
- Upload a product photo to Avocado Studio
- Use the prompt-to-edit canvas to remove the background and place the product on white
- Generate additional lifestyle images using text prompts
- Download all images, verify IPTC metadata, and add it if missing
- Upload to Merchant Center via your feed management tool
Common Mistakes That Get Products Disapproved
Based on Google Merchant Center data, image issues account for roughly 65% of all product disapprovals. Here are the most common mistakes:
- Promotional overlays: Adding "SALE", "20% OFF", or price badges to the primary image. Keep promotional text in your ad copy, not on the product image.
- Watermarks: Even subtle brand watermarks trigger disapproval. Remove all watermarks before uploading.
- Low resolution: Images below 500x500 px will be disapproved starting January 2027. Use 1500x1500 or larger now.
- Wrong variant images: Submitting a blue product image for a red variant. Each color, size, and pattern variant needs its own image.
- Missing IPTC metadata on AI images: Stripping metadata from AI-generated images violates Google's policy. Always verify before uploading.
- Product does not fill the frame: If the product takes up less than 75% of the image area, crop tighter or regenerate.
FAQ
What is the minimum image size for Google Shopping in 2026?
The current enforced minimum is 100x100 pixels for non-apparel and 250x250 pixels for apparel. Starting January 31, 2027, all products must meet a universal 500x500 pixel minimum. Google recommends 1500x1500 pixels or larger for best results.
Does Google Shopping allow AI-generated product images?
Yes. Google explicitly allows AI-generated images and even built Product Studio into Merchant Center for AI image editing. The requirement is that all AI-generated images must include IPTC DigitalSourceType metadata indicating they were created with AI.
What IPTC metadata do I need for AI product images?
Use TrainedAlgorithmicMedia for images generated entirely by AI, and CompositeSynthetic for real product photos edited with AI (e.g., background replacement). This metadata must be embedded in the image file before uploading to Merchant Center.
Can I use AI to remove the background from my product photos for Google Shopping?
Yes. Google recommends white or transparent backgrounds for primary product images. AI background removal is a common and accepted practice. Use the CompositeSynthetic IPTC tag if you are editing a real photo with AI.
How many images can I submit per product in Google Shopping?
Up to 12: 1 primary image (image_link), up to 10 additional images (additional_image_link), and 1 lifestyle image (lifestyle_image_link). Products with 3 or more images typically see higher engagement.
What image formats does Google Merchant Center accept?
JPEG, PNG, WebP, non-animated GIF, BMP, and TIFF. JPEG and PNG are the most widely used and reliable formats.
How much does it cost to generate product images with AI?
On Avocado AI, 1-credit models like Nano Banana 2 and Recraft v4 cost approximately EUR 0.10 to 0.13 per image on the Starter plan. A 100-product catalog with multiple images per product would cost roughly EUR 10 to EUR 30 depending on the models used.
Do I need to be on a specific Avocado AI plan to generate product images?
All Avocado AI plans, from Intro (EUR 19.99/mo) to Pro (EUR 249/mo), include access to all image models. The difference is the monthly credit allocation: 100 credits on Intro, 300 on Starter, 800 on Growth, and 2,000 on Pro.
How to Pick the Right AI Image Approach in Under 30 Seconds
- Need white-background packshots fast? Use Nano Banana 2 (1 credit, clean detail, fastest turnaround).
- Need lifestyle or campaign images? Use SeedDream 5.0 Lite (1 credit) or GPT-Image 2 (2 credits) for richer scenes.
- Have a real product photo that needs editing? Use Avocado's prompt-to-edit canvas to swap backgrounds, remove objects, or enhance lighting.
- Generating images for 100+ SKUs? Use Avocado Flows to batch-generate from a template prompt.
- Selling apparel? Generate on-model images with lifestyle context. Google strongly recommends showing clothing on a person rather than flat-lay.
- Selling on Amazon too? Amazon's requirements overlap with Google Shopping but are not identical. Verify both sets of specs before using the same images.
If you want one workspace for product images, video ads, and brand content, start with Avocado AI. Plans start at EUR 19.99/mo with access to all image models and commercial usage rights.
Written by Wanderson Jackson, founder of Avocado AI. Wanderson has spent the last two years building AI-powered creative tools for e-commerce brands and marketing teams.