Google Gemini AI is the multimodal model that has changed how UK e-commerce teams edit product photography in 2026. Instead of learning Photoshop, you describe the change in plain English and the AI handles the rest. This guide shows how to use Gemini AI photo editor for professional product retouching, what it costs in pounds, and where its limits sit for British online sellers.
What is Gemini AI and why it changes the game for photo editing
Google Gemini is not just another filter slapped on top of an image. It is a natively multimodal AI model, trained from the ground up on text and images together. In practice, this means it understands both your text instructions and the actual content of the photo - context, lighting, composition and even the surface materials of the product itself.
Traditional editors require mastering dozens of tools - layers, masks, tonal curves. The free Gemini AI photo editor lets you describe an edit in a single sentence: "swap the background for clean white studio" or "add warm lighting from the left". The model interprets the instruction and performs the edit, preserving the original detail of the input image.
In the Gemini app, images are created and edited by Nano Banana 2 (Gemini 3.1 Flash Image), or by the faster Nano Banana 2 Lite when the model is set to Flash-Lite. Nano Banana Pro (Gemini 3 Pro Image) is available on paid plans as a way to redo an image with more detail. In the Gemini API, Google recommends Nano Banana 2.1 for new projects, and the Imagen models have been shut down there (as of 22 October 2026).
What Gemini can do - a tour of the photo editing features
Background swap and removal
The most common use case in e-commerce. You upload the product shot and write something like "place the product on a marble countertop in warm side lighting". Gemini processes it in seconds, preserving the original shadows and reflections. This is ideal when you need background variants without booking another studio session - particularly useful for Amazon UK, eBay UK and Etsy listings where multiple lifestyle variants drive conversion.
Adding and removing objects
Need to remove an unwanted cable from the frame? Or add seasonal props - pinecones and evergreen branches for a Christmas campaign? Gemini AI rebuilds the background naturally, with no obvious artefacts. Important rule: one edit per prompt. Stacking multiple instructions in a single prompt usually means only one of them is executed cleanly.
Style and aesthetic transformation
Take a single base shot and turn it into a series of completely different aesthetics - editorial Kinfolk style, clean Scandinavian minimalism, vivid maximalist Vinted-friendly compositions. The same product, ten different mood boards, in minutes. Gemini handles colour grading, prop swaps and style transfer with consistency that beats older diffusion-only models.
Text and labels in images
One of the underrated strengths of Nano Banana 2 is reliable text rendering. You can ask the model to add a price tag, a "Sale" badge, a label on a packaging mockup - and the typography stays sharp and legible. This solves the classic problem with older AI generators where text usually came out as scrambled gibberish. For UK sellers preparing campaign assets in pounds (£) or with British English copy, the difference is significant.
Multi-image fusion
You can upload up to 14 reference images at once. Gemini combines elements from multiple sources - take the product from one shot, the lighting from another, the background from a third. This is closer to how creative directors actually think than traditional single-prompt generation.
Step-by-step: how to edit product photos in Gemini
The whole process boils down to four steps. No installation, no plug-ins, no learning curve - just a browser and a Google account.
- Sign in to Gemini. Go to gemini.google.com and log in with your Google account. The interface is identical for free and paid users.
- Open Create Image. From the Tools menu, pick "Create image". This is the dedicated image-editing surface.
- Upload your photo. Drag and drop a product shot - up to 10 files, each up to 100 MB. JPEG, PNG and WebP all work.
- Describe the edit. Write your instruction in plain English. Iterate by sending follow-up messages to refine the result. Download the final version.
The 4-element method - the key to a consistent series
To keep an entire product series visually consistent, repeat the same four parameters in every prompt: light direction (e.g. "soft light from left"), colour temperature (e.g. "5500K neutral daylight"), camera parameters (e.g. "85mm f/2.8, shallow DoF") and background (e.g. "pure white RGB 255,255,255"). UK sellers running cross-platform listings on Amazon UK and eBay UK find this approach saves hours when scaling catalogue uniformity.
5 copy-and-paste prompts
- Hero shot: "Transform this product image into a professional e-commerce hero shot. Place on pure white background (RGB 255,255,255). Three-point softbox lighting. 45-degree elevated camera angle. Ultra-sharp focus."
- Office lifestyle: "Place this product on a modern minimalist home office desk with a MacBook and coffee cup. Warm natural window light from the left. Shallow depth of field. Editorial photography style."
- Winter campaign: "Create a winter holiday scene with this product. Add pinecones, evergreen branches, warm bokeh lights. Cosy warm lighting 3200K. Premium gift-giving context."
- Floating product: "Product appearing to float in mid-air with subtle grounding shadow below. Clean gradient background white-to-light-grey. Dramatic rim lighting. Premium luxury feel."
- Flat lay set: "Arrange all product components in a top-down flat lay composition on a clean white surface. Even soft lighting from above. Each item clearly visible and labelled. Minimalist editorial style."
How much does Gemini photo editing cost? Free vs paid plans
One of the biggest selling points of the free Gemini AI photo editor is that basic editing costs nothing. In the UK, Google offers a free tier and three Google AI plans: AI Plus at £4.49, AI Pro at £18.99 and AI Ultra from £79.99 per month (as of 22 October 2026). Even the free tier can generate and edit images.
| Plan | Price (UK, per month) | Gemini app usage limit | Nano Banana 2 | Nano Banana Pro | Image download |
|---|---|---|---|---|---|
| Free | £0 | Standard | Yes | No | 1K |
| AI Plus | £4.49 | 2x standard | Yes | Yes (redo) | 2K |
| AI Pro | £18.99 | 4x standard | Yes | Yes (redo) | 2K |
| AI Ultra | from £79.99 | 5x or 20x AI Pro | Yes | Yes (redo) | 2K (4K only via the Gemini API) |
Free plan - what you actually get
Without a Google AI plan you get image generation and editing with Nano Banana 2 at standard limits. Google does not publish a fixed number of images: Gemini app limits are compute-based, depend on prompt complexity and chat length, and refresh every five hours until you reach a weekly limit. That is enough for testing and small projects. Downloads are 1K, fine for social media but too low for serious online shop photography. "Redo with Pro" (Nano Banana Pro) is only available on paid plans (as of 22 October 2026).
Google AI Studio and the Gemini API - what is free and what is paid
Google AI Studio is Google's developer tool for testing models and working with the Gemini API. Google says AI Studio usage stays free of charge unless you link a paid API key for paid features. In the Gemini API pricing, the image models have no free tier. You pay per image in US dollars: Nano Banana 2 (gemini-3.1-flash-image) costs about $0.067 per 1K image and $0.101 per 2K image, and Nano Banana 2.1 (gemini-nano-banana-2.1), recommended for new projects, about $0.0336 per 1K image (as of 22 October 2026). The API makes sense for large batches and automation, but it requires basic programming knowledge.
Monthly cost comparison
| Tool | Cost/mo | Images/mo | Cost per image |
|---|---|---|---|
| Gemini (free) | £0 | no fixed number (standard limit) | £0 |
| Gemini AI Plus | £4.49 | no fixed number (2x limit) | - |
| Gemini AI Pro | £18.99 | no fixed number (4x limit) | - |
| Adobe Photoshop | $22.99 | unlimited | - |
| ChatGPT Plus | $20 | higher than Free, no published number | - |
| Canva Pro | $12.99 | 500 AI credits | ~$0.026 |
Gemini vs ChatGPT vs Photoshop - AI editor comparison
Tests by techradar.com show the picture clearly: Gemini preserves the original photo better than ChatGPT. When you ask for a background change, ChatGPT often quietly alters a person's hairstyle, the colour of clothing, or product details. Gemini minimises unintended changes - critical when you need to publish on Amazon UK or John Lewis where image accuracy matters.
| Aspect | Gemini AI | ChatGPT | Photoshop |
|---|---|---|---|
| Original preservation | Best - minimal drift | Often alters details | Full control |
| Generation quality | Realistic, consistent | Artistic, stylised | Perfect (manual) |
| Edit speed | Seconds | 10-30 seconds | Minutes to hours |
| Learning curve | Zero | Zero | Steep (hundreds of tools) |
| Price (entry) | £0 | £0 (Free plan, image limit) | $22.99/mo |
| Batch processing | Yes (API) | Limited | Yes (Actions) |
| IP indemnity | None in the app and Gemini API (offered in Google Cloud) | None | Firefly only, on qualifying plans |
| Best for | Precise photo edits | Creative generation from scratch | Advanced retouching |
The takeaway from the tests? "ChatGPT is an idea machine, Gemini is a photo editor, Photoshop is a surgeon's scalpel." Each tool has its place. For most UK e-commerce sellers, Gemini offers the best balance of price, speed and quality.
Curious how it compares to the broader landscape? See our ranking of the 8 best AI image generators for a complete breakdown of where Gemini sits among Midjourney, Firefly and DALL-E.
Google Product Studio - free editing inside Merchant Center
If you sell through Google Shopping, you already have access to Product Studio - a free AI tool built into Google Merchant Center Next. You don't even need to open Gemini directly - you edit product photos straight inside the merchant dashboard. For UK retailers running Google Shopping campaigns, this is the lowest-friction entry point.
Product Studio offers one-click background change, quality upscaling for low-resolution photos, seasonal scene generation, and - new in 2026 - short video creation from single product photos. All of it free, with no per-image cap. Read our full Google Product Studio guide for hands-on examples.
Google's own data is clear: products with more than one image get +76% impressions and +32% clicks. Product Studio lets you generate several variants from one master shot in minutes - so there's really no excuse not to do it.
Gemini's limitations - what you need to know before you start
Gemini AI is not perfect, and you should know its weak spots before going all-in. The safety filter blocks editing of recognisable people - celebrities and politicians. If your photo includes a famous face, Gemini will refuse the request.
Detail hallucinations are a frequent issue. The model can add small elements (an extra button, a slightly altered material tone) even when the prompt says "preserve the original". Always compare the output to the input pixel-by-pixel before publishing on a marketplace - especially under Amazon UK A+ Content rules where misleading imagery can trigger listing suspension. When the shape or colour of the product drifts, correct it with manual photo retouching instead of more prompting.
A critical legal point: Google offers no IP indemnity for the Gemini app or the Gemini API. Google lists such commitments for selected Google Cloud and Workspace business services, including Gemini models on the Gemini Enterprise Agent Platform (formerly Vertex AI); the Gemini app and the Gemini API are not on that list. Adobe covers Firefly output only where the customer's agreement with Adobe provides for it, only for eligible Firefly features, and not for features powered by non-Adobe models or beta features. For larger UK brands, this can be a problem. Privacy is another factor - all photos are uploaded to Google servers for processing, which has data-protection implications under the UK GDPR. The UK's pro-innovation AI regulation framework is lighter-touch than the EU AI Act, but the underlying data law still applies.
- Download quality - the preview often looks better than the downloaded file (varies by subscription tier)
- Quality instability - users report quality drops after model updates
- Regional limits - Personal Intelligence (Google Photos integration) is not yet available in the UK or EU
- No offline mode - every edit needs an internet connection
- Large file compression - Google automatically compresses images above a certain size
For the broader regulatory picture across the UK and EU, see our explainer on the EU AI Act and product photography - rules that affect British sellers exporting to the EU.
Why AI cannot replace a strong base photo
Gemini AI is a powerful post-processing tool. But there is one rule no AI model can break: garbage in, garbage out. Even the best prompt cannot fix blurry focus, bad lighting or weak composition in your source photo.
AI cannot recover information your photo does not contain. At best it can draw in missing detail, and that detail does not have to match the real product. If you upload a dark, grainy phone snap, Gemini can change the background and add effects, but the quality of textures, details and product colours will stay low. Google's own documentation says it plainly: "start with moderate-sized, high-quality images".
The best ROI in e-commerce comes from combining professional base photography with AI editing. One well-lit studio session is an investment from which Gemini can spin out dozens of variants - different backgrounds, lifestyle contexts, seasonal campaigns. Without that foundation, AI can produce a hundred variants and none of them will look properly professional.
Need base photography that lets you squeeze the most out of Gemini AI? Get in touch with marszalstudio - we deliver the source material that AI can transform into dozens of variants for your e-commerce needs. You can also browse our overview of the top 5 AI photo editors if you are still narrowing down your stack.
2026 updates - Gemini in Photoshop and Google Photos integration
Adobe Photoshop 2026 integrates Gemini 3 (Nano Banana Pro) as one of the available AI models alongside Adobe Firefly and FLUX. This confirms Gemini's position as a reference model for image editing - even Adobe acknowledges it as a valuable addition to its own ecosystem.
The second big update is Personal Intelligence - Gemini reaches into your Google Photos library and recognises people and pets from labels. Instead of describing "a woman with brown hair" you write "my partner". For now this feature is US-only for AI Plus/Pro/Ultra subscribers, but UK and European rollout is a matter of months.
Worth tracking too: the new markup tools - lets you draw on the image to indicate the area to edit, replacing positional descriptions in words. You highlight a region and write the instruction - edit precision goes up dramatically. This feature is currently in beta. For deeper context on Nano Banana itself, see our dedicated piece on Nano Banana in product photography.
If you want to compare Gemini to a tool built specifically for product photography, read our review of the Photoroom AI Photo Editor.
Frequently asked questions
Yes - basic photo editing in the Gemini app is free. Without a Google AI plan you get Nano Banana 2 at standard limits (Google does not publish a fixed number of images) and 1K downloads. Paid Google AI plans (in the UK from £4.49 per month, as of 22 October 2026) raise the limits, allow 2K downloads and let you redo images with Nano Banana Pro. Image models in the Gemini API are paid.
Go to gemini.google.com, sign in with your Google account, and pick "Create image" from the Tools menu. Upload your product photo (up to 10 files, 100 MB each). Describe the change in plain English - for example, "swap the background for clean white studio, keep the original shadows". Iterate by sending follow-up messages to refine the result, then download the final version using the download button.
For precise editing of existing photos, Gemini is the better choice - it preserves more original detail and rarely introduces unintended changes. ChatGPT performs better in creative image generation from scratch. Both have a free plan that includes image generation with a limit: neither Google nor OpenAI publishes a fixed number of images there, and paid plans raise the limits (as of 22 October 2026).
For 80% of typical e-commerce edits - yes. Gemini handles background swaps, object removal, style changes and variant generation. Photoshop remains essential for advanced layered retouching, RAW workflows and precise colour correction - work that requires a specialist. Notably, Photoshop 2026 itself integrates Gemini as one of its AI models.
Google does not publish a fixed number. Gemini app limits are compute-based, depend on prompt complexity and chat length, refresh every five hours until you reach a weekly limit, and the app tells you when you are close. In the UK, AI Plus (£4.49 per month) gives 2x higher limits than the free tier and AI Pro (£18.99) 4x higher (as of 22 October 2026). Image models in the Gemini API have no free tier.
