How to Create and Edit Images with Gemini (Nano Banana)

The short version

Describe the picture the way you’d brief an illustrator: what’s in it, where, and in what style. Gemini’s image model, Nano Banana 2, is good at text in images, keeping a character consistent across pictures and making small, local edits to a photo.

Step by step

Create an image

1. Go to gemini.google.com and sign in.
2. Open the sidebar and click Images.
3. Pick a template or type your prompt. Google suggests starting with create, draw or generate, naming a style (photorealistic, watercolor, cartoon) and describing the subject, what it’s doing and the background.
4. Ask for changes in plain words: “Make the sky darker”, “Move the logo to the bottom right”.

Edit a photo

1. Upload a photo with Add files, or open the sidebar, click Library and pick an image you made.
2. Under the image, click Chat.
3. Say exactly what to change and what to keep.

You can also upload several photos and ask Gemini to combine them into one scene.

Save it

Hover over the image and click Download full size, or click More > Export to Docs. Google says downloads are 2K with a Google AI plan and 1K without.

Prompts to copy

A product photo:

Create a photorealistic image of a ceramic coffee mug on a wooden cafe counter,
morning light from the left, shallow depth of field, plain background, no text.

A sign with text:

Create a simple poster that says "Fresh Bread Daily 7am" in large bold letters,
warm brown and cream colors, flat illustration of a loaf. Spell the text exactly as written.

A careful edit:

Replace the cloudy sky in this photo with a clear blue sky.
Keep the building, people and lighting on the street exactly the same.

A consistent character:

Using the cartoon fox from the last image, show it reading a map at a train station.
Same colors, same scarf, same drawing style.

Example

You give: a phone photo of your shop window and the prompt “Remove the reflection of the parked car and brighten the display, keep everything else.”

A good result: the same window with the reflection gone. If the edit spills over (the sign changes too), reply “Only change the reflection; restore the sign exactly as in the original.”

Mistakes to avoid

• Vague prompts. “A nice picture of a dog” gets a generic dog. Breed, setting and style get your dog.
• Not saying what to keep. In edits, list what must stay the same.
• Trusting text and details. Check spelling, hands, labels and logos before you print anything.
• Using real people without permission. Google asks you to respect others’ privacy; get consent before editing someone’s photo.

Good to know

Google’s Help Center says you must be 13 or over (or the applicable age in your country) to generate images and 18 or over to edit them. Paid subscribers can click More > Redo with Pro to remake an image with Nano Banana Pro, which Google says adds detail, especially for text and infographics. Gemini may remove an image if it detects a policy violation.

Related: make slides with Gemini and write better Gemini prompts.