To create an AI image in Gemini, go to gemini.google.com, sign in with your Google account, click Images in the sidebar and describe what you want, starting with a verb like “create,” “generate” or “draw.” Gemini returns the image in seconds, and you can download it or refine it with follow-up requests. The steps below also cover editing your own photo, account limits and how to spot images made by Google’s AI.
What do you need to get started?
You need a Google account and a browser. Google’s help page says you must be signed in to Gemini Apps, and that generating images on a personal account requires being at least 13 (or the applicable age in your country). Editing images requires being 18 or older. School accounts may have different restrictions, and availability depends on language and country.
One recent change matters here: since October 9, people without a paid plan only get the Flash-Lite model in Gemini, as we explained in our story on the free tier changes. It matters because each text model has an associated image model, as you’ll see below.
How do you create an image from scratch?
- Go to gemini.google.com on your computer and sign in with your Google account.
- Click Open sidebar, then Images.
- Pick a model if the selector shows up on your account, or just type your request in the text box.
- Start the request with a verb like “draw,” “generate” or “create,” as Google recommends, and include the style and visual details you want.
- Send it and wait for the result. If it’s not what you pictured, ask for an adjustment in the same chat, such as “change the background to a late-afternoon sky.”
- When you like it, download the image. Download resolution is 1K without a plan and 2K with a Google AI plan, according to the help page.
These steps are for desktop. If you’d rather use the phone app, see our guide on how to use Gemini on your phone.
Which image models are available?
Google’s help page lists three options:
- Nano Banana 2: the default for requests made with the Flash or Pro models.
- Nano Banana 2 Lite: optimized for speed and paired with the Flash-Lite model. It isn’t built for multiple reference images or several edits in a row.
- Nano Banana Pro: available only with a paid Google AI plan, used to redo images with more detail.
There’s a daily quota of Nano Banana 2 images. Once you hit it, you can’t redo images with Nano Banana Pro. The quota can differ from account to account, so if the work matters, test before you depend on it.
How do you write a good prompt?
Vague requests produce generic images. To get better results, describe four things in a sentence or two:
- The subject: who or what appears (“an orange cat sleeping on a velvet armchair”).
- The style: realistic photo, illustration, watercolor, cartoon, 3D.
- Light and color: “soft late-afternoon light,” “blue and gold tones.”
- Framing: close-up, wide shot, overhead view.
A full example: “Create a realistic photo of an orange cat sleeping on a green velvet armchair, soft late-afternoon light, close-up.” If the first result gets one detail wrong, don’t rewrite everything. Ask for just the change, since Gemini continues from the previous image.
Avoid asking for text inside the image, such as posters and logos, because image tools still get letters wrong often. If you need text, add it afterward in an editor.
How do you edit your own photo with Gemini?
The help page says you can edit an image you made in Gemini or upload one from your device. Here’s the path:
- Open Library in Gemini.
- Click the image you want to change.
- Select Chat, below the image.
- Type what you want changed, for example “remove the people in the background” or “make the photo black and white.”
- Click Send.
You can also upload several images at once and ask for a new image based on them. Nano Banana 2 is described as a good fit for personalized images, such as trying a different haircut on a photo of yourself. Respect other people’s privacy and copyrights: the page itself asks you not to violate them, and Gemini may remove images when it detects a possible violation of the terms of service or the prohibited use policy.
How can you tell if an image was made by Google’s AI?
Gemini also works as a checker. According to Google’s help page, you sign in to Gemini, upload the file and ask something like “Was this image created or edited by Google AI?” You can also type @synthid to start the check. Gemini looks for an invisible digital watermark called SynthID, which Google embeds in images made with its tools.
A few rules for the check:
- Upload one file at a time, up to 100 MB.
- If it’s a screenshot, crop it tightly and don’t upload collages of several images.
- A positive result means all or part of the image was created or edited by Google’s AI.
- A negative result doesn’t prove the image is real: it may have been made by another AI tool, because Gemini only recognizes content generated by Google’s AI.
Frequently asked questions
Can Gemini create images for free?
Yes, you can generate images on an account without a paid plan, but with limits. The daily quota and download resolution (1K without a plan, 2K with a Google AI plan) depend on the account.
Can I use Gemini images for work?
Google asks that you not violate anyone’s copyright or privacy. For commercial use, read the current terms of service before publishing, because the rules can change.
Why did Gemini refuse or remove my image?
The help page says Gemini may remove images when it detects a possible violation of the terms of service or the prohibited use policy. Rewrite the request without the part that may have triggered the block.
What’s the difference between Nano Banana 2 and Nano Banana Pro?
Nano Banana 2 is the default model. Nano Banana Pro is only available on paid Google AI plans and is used to redo images with more detail.
