How to Edit Your Photos with Gemini AI

Before you start: account and age rules

Google's help page for generating and editing images sets clear age limits for personal accounts. You must be 18 or over to generate and edit images. People aged 13 and over (or the applicable age in their country) can generate new images, but the page says editing images is not available to users under 18. School accounts may have different restrictions.

You also need to be signed in, and the feature depends on the languages and countries each Gemini app supports. If you are a student or a younger family member wants to try a festival edit, the safer route is for an adult to do the edit on their own account with their own photo.

• Account: a personal Google Account works. Work and school accounts are controlled by the organisation.
• Where: the Gemini web app at gemini.google.com and the Gemini mobile app both support images.
• Models: Google's help page lists Nano Banana 2 for Flash and Pro, a lighter Nano Banana 2 Lite for the Flash-Lite setting, and Nano Banana Pro for people on a Google AI plan.
• Downloads: the same page says downloads come at 1K resolution without an AI plan and 2K with one.

How to edit a photo in Gemini, step by step

The exact buttons move between updates, so treat these steps as a general route rather than a tap-by-tap map. This is how it works as of October 2026.

1. Open Gemini in your browser or the app and start a new chat. A fresh chat keeps old instructions from leaking into this edit.
2. Use the upload or add button next to the text box and pick your photo from your phone or computer. Choose the clearest, best-lit version you have. Our guide on the best photo to upload for AI editing explains what to look for.
3. Type or paste your edit request. Say what to change and what to keep. For example: "Change my background to a Diwali balcony with diyas and soft evening light. Keep my face, hairstyle, skin tone and clothes exactly as they are."
4. Send the message and wait for the image. If Gemini replies with text only, ask directly: "Please create the edited image."
5. Check the face, hands and edges closely at full size before you share anything.
6. Ask for one small fix at a time in the same chat, for example "make the lighting warmer" or "remove the extra lamp on the left".
7. Download the version you like. Google's help page says that on a computer you hover over the image and choose the download option for full size.

You can also edit an image Gemini made earlier, or upload more than one photo and ask Gemini to build a new image from them. Google's help page says Nano Banana 2 can accept multiple reference images, while the Lite version is not designed for several references or many edits in a row.

Writing an edit request Gemini understands

Google's own Nano Banana tips say you can give simple, conversational instructions, and its Nano Banana Pro tips add that edits to an existing image should be direct and specific. In practice that means short sentences that name one change, then a clear list of things that must stay the same.

A useful pattern for portraits has four parts:

• The change: "Put me in a maroon silk saree with a gold border."
• The setting and light: "Stand me in a palace courtyard in warm golden-hour light."
• What stays fixed: "Keep my face, expression, skin tone, hairline and body shape unchanged."
• The format: "Make it a vertical 4:5 image for an Instagram post."

Google's Pro tips also suggest thinking like a photographer: name the framing (close-up, half body, wide), the lighting (golden hour, soft window light) and the style (photorealistic, watercolour). If you want words inside the picture, write the exact text and how it should look. When you upload two photos, say what each one is for, such as "use photo 1 for my face and photo 2 for the background". Our guide on how to write AI photo prompts breaks each part down further, and festival pages like Diwali-ready traditional wear prompts or Christmas prompts give you full examples to copy.

Fixing face drift and other mistakes in follow-ups

Google says Nano Banana was built to keep the likeness of people and pets consistent across edits, and that it can change specific parts of an image while keeping the rest. Results still vary from one attempt to the next, so expect to nudge it. The trick is to correct in small steps rather than paste a whole new prompt.

If three or four follow-ups have not fixed the face, start a new chat and upload the original photo again. Each edit builds on the last output, so small changes can pile up. Our guides on keeping your face the same and fixing common AI photo problems go deeper.

Watermarks and labels on Gemini images

Google adds identification to images made or edited in Gemini, and it is worth knowing what is there before you post. Google's watermark help page says all AI-generated media created or edited with Gemini Apps includes an invisible SynthID watermark, and Gemini Apps also include Content Credentials (C2PA) metadata about the content's origin.

There can also be a visible watermark. The same help page says that if you are in India, South Korea or Vietnam, you will only see the setting to turn visible watermarks off if you have an AI Ultra subscription; otherwise visible watermarks are applied automatically. That setting controls only the visible mark and does not affect SynthID or Content Credentials.

These marks help people tell that an image was made or changed with AI. That matters for honesty on social media, so we do not share ways to strip them. If you want to check an image, Google's help page says you can upload it to Gemini and ask whether it was made with Google AI, with the limit that Gemini can only recognise content from Google's own tools.