Gemini AI photo prompts: How to edit and generate images, with examples to copy

Plans, prices and features checked on the official pages on October 20, 2026.

A good Gemini photo prompt names the subject, the change you want and what must stay the same, for example: “Change the sofa to dark green. Keep everything else, including the faces, exactly the same.” Gemini can both generate new images and edit photos you upload, and it is free to use with a Google account, with higher limits on paid plans.

This guide covers how photo editing in the Gemini app works, what it can do, prompts based on Google’s own examples, and what each plan includes.

How Gemini edits and generates images, and who can use it

Gemini’s image tools run on Google’s image models, which Google calls Nano Banana. According to Google’s Gemini pages, Nano Banana 2 is free for image generation and editing, while a newer model, Nano Banana 2.1, is described as Google’s most advanced and is aimed at users with a Google AI plan. Nano Banana Pro, which also needs a Google AI plan, can be used to redo an image when more detail is needed.

Google’s help center also lists a lighter Nano Banana 2 Lite, used with the Flash-Lite model. It is optimized for speed and, according to Google, is not designed for multiple image references or multiple edits.

You work in plain language, in a chat. You can ask for a new picture, upload a photo and describe a change, or upload several photos and ask Gemini to combine them. Google says image generation is available in all languages and countries where the Gemini app is available, and that you must be signed in. Age rules apply: the overview page says 18+, and school accounts can have different restrictions.

What Gemini can do with a photo you upload

Google’s overview of image generation and editing lists these capabilities:

• Targeted edits: change a background, replace an object or add an element while keeping the rest of the image intact.
• Lighting, camera angle, focus and aspect ratio: adjust how the shot looks without starting over.
• Style and texture transfer: apply the color, texture or style of one photo to another.
• Combining images: merge several pictures into one. Google’s blog says you can combine up to three images in the original Nano Banana editing flow.
• Character consistency: keep a person’s or character’s appearance across generated images and edits.
• Text in images: clearer, more accurate text for logos, invitations, posters and comics, in many languages.
• Infographics and diagrams, 2K output and resizing a design to different formats.

Google’s tips post adds that you can turn a sketch into a realistic scene, annotate objects, and restore and colorize old photos. Its example of a pixel-level edit is the simple prompt “the dog’s mouth is closed.”

Prompts that get results: The structure Google recommends

Google’s help center says to start with a verb such as draw, generate or create, name a style, and describe the subject, action and setting in detail. Its overview page suggests the formula subject, action, scene, then adding detail. Specific beats vague: “a young woman in a red dress running through a park” works better than “a woman in a red dress.”

For Nano Banana Pro, Google’s prompt guide lists six elements: subject, composition, action, location, style and, for edits, direct editing instructions. It also suggests controls such as aspect ratio (a 9:16 poster or a 21:9 wide shot), camera and lighting wording (“shallow depth of field (f/1.8)”, “golden hour backlighting”), and giving each uploaded image a role such as pose, art style or background.

Editing prompts to copy and adapt

The first three follow wording used in Google’s own examples. The rest apply the same structure to common jobs.

• Change the man's tie to green. (Google’s direct-edit example.)
• Turn this scene into nighttime. (From Google’s Nano Banana Pro guide.)
• Focus on the flowers. (Changes depth of field, per the same guide.)
• Restore this photo. Make it less blurry, remove the white streaks and add bright, vivid color. Keep his face the same. (Based on Google’s restoration trend examples.)
• Replace the background with a quiet beach at golden hour. Keep the person, their clothes and the lighting on their face unchanged.
• Use image A for the pose, image B for the art style and image C for the background. Make it a 16:9 cinematic image.
• Translate the text on the packaging into Korean. Keep everything else the same.

Generation prompts to copy and adapt

• Create a photorealistic close-up of a stoic robot barista with glowing blue optics, low angle, shallow depth of field, warm cafe lighting, 4:5 aspect ratio.
• Create an infographic that shows how to make elaichi chai. (A Google example. Google advises verifying facts in data-driven visuals.)
• Create a storyboard for this scene. (With a reference image attached.)

How to start editing a photo in the Gemini app

1. Sign in to Gemini with your Google account on the web or in the mobile app.
2. Open the sidebar and choose Images, or use “Create images” in the tools menu, then pick a template or type a prompt.
3. To edit, upload your photo, or open the Library, select an image and choose Chat under it.
4. Write one clear instruction: what to change, and what to keep.
5. Refine in follow-up messages. Google says that if a result is not right you can simply ask Gemini to modify it.
6. Download the result. Hover over the image and choose Download full size. Google says the resolution is 2K with a Google AI plan and 1K without one.

Paid subscribers can also choose More, then “Redo with Pro” to regenerate an image with Nano Banana Pro. Google notes this is not available once you have reached your daily Nano Banana 2 quota.

Plans and pricing: Free tier and what the paid plans change

Prices checked on October 20, 2026.

Google does not publish a number of images per plan. Its plans page says limits are based on compute, depend on prompt complexity and the features used, reset every 5 hours until a weekly limit is reached, and can be extended by buying AI credits. Plans also include monthly credits for Flow, Google’s creative studio for film-style scenes, which is separate from the Gemini app.

Limits and things to know before you rely on it

Google is open about the weak spots. Its Nano Banana Pro guide says small text, fine details and accurate spelling may not render perfectly, that multilingual text can contain grammar errors, and that complex edits and image blending can produce unnatural artifacts, especially when lighting changes. Character consistency across edits may also vary.

Every image comes marked. Google says Gemini adds an invisible SynthID watermark and a visible watermark to show content is AI-generated. You can upload an image to the Gemini app and ask whether it was made with Google AI; Google says this works for images, audio and video. That is the official route for people searching for a Gemini image detector, and it covers Google’s own watermark only.

Google’s help center also says images may be removed if the prompt appears to violate its Terms of Service or Prohibited Use Policy. For people in photos, ask for changes to a face only when you have the right to use that picture.

Gemini compared with other free image tools

Gemini is one of several free options. Our comparison of free AI image generators sets Gemini beside Bing, Firefly and FLUX, and our Midjourney guide covers a paid alternative. If you want motion rather than stills, see free AI video generators.

What sets Gemini apart, according to Google, is the chat-based editing loop: one conversation where you generate, edit, combine and refine, plus tight links to Canvas and Google AI Studio, where Google says Nano Banana is built in for developers and app builders. Google’s pricing page also lists Gemini 4 Argon access on paid plans, a model covered in this week’s new models roundup.

Common questions