If you’ve scrolled through social media in the last year and seen someone’s selfie turned into a tiny collectible figurine, or a bedroom instantly repainted a different color without touching anything else in the photo, there’s a good chance Nano Banana was behind it. Google’s viral image model, built into the Gemini app, has become one of the most talked-about AI editing tools around and unlike a lot of AI hype, this one is genuinely useful for everyday photo editing, not just novelty effects.
This step-by-step guide walks through exactly how to use Nano Banana for image editing, from your first upload to advanced multi-photo blending, plus the prompting habits that make the biggest difference in your results.
What Is Nano Banana, Exactly?
Nano Banana is the nickname that stuck for Google DeepMind’s Gemini Image model technically Gemini 2.5 Flash Image when it first went viral, now succeeded by more advanced versions. Since its debut, Google has released two major upgrades: Nano Banana Pro (Gemini 3 Pro Image), built for complex, studio-quality compositions with accurate text rendering, and Nano Banana 2 (Gemini 3.1 Flash Image), which combines Nano Banana Pro’s advanced reasoning and world knowledge with the speed of Gemini’s Flash model. As of 2026, Nano Banana 2 has become the new default across most Google products, with Nano Banana Pro still available for specialized, high-fidelity tasks.
What made the original model go viral, and what still defines the whole Nano Banana family, is that it was built from the ground up to process text and images together. Instead of only generating a picture from a written prompt, it can look at an image you upload, understand what’s actually in it, and make targeted, contextual changes while keeping everything else in the scene untouched. Google’s product team refers to this as “pixel-perfect editing”: the ability to change one specific detail, like the color of a sofa, without disturbing anything else in the shot.
That distinction matters. A lot of AI image tools treat every prompt as a blank slate, regenerating the whole picture and often changing details you wanted to keep. Nano Banana is designed to hold onto context across an editing conversation, so you can make a series of small, precise adjustments instead of starting over each time.
Where to Find Nano Banana
Nano Banana is built into a growing list of Google products, and it helps to know which one fits your task before you start.
- Gemini app (web and mobile) the main, free-to-try home for Nano Banana. Best for casual editing, photo transformations, and quick creative experiments.
- Google AI Studio the developer-facing option, useful if you want to build a custom app or automate image editing through code.
- AI Mode in Search accessible from the Google app or browser, useful for quickly visualizing an idea without leaving Search.
- Google Workspace (Slides and Vids) lets you generate and edit visuals directly inside a presentation or video project, including a “Beautify this slide” feature that redesigns a slide using its existing content as context.
- Google Ads advertisers can use Nano Banana to generate and edit campaign creative directly inside the Ads platform.
- Flow and Mixboard Google Labs tools for filmmaking and visual brainstorming that include Nano Banana image generation.
- Developer platforms Vertex AI, the Gemini API, Firebase AI Logic, and Antigravity, for teams building Nano Banana into their own products.
For most people who just want to edit a photo, the Gemini app is the simplest starting point, so that’s what the rest of this guide focuses on.
Step-by-Step: How to Use Nano Banana in the Gemini App
Step 1: Open Gemini and Select the Image Tool
Open the Gemini app or go to gemini.google.com, then select the image tool from the tools menu usually labeled with a banana icon and “Create images.” From the model menu, you can choose between “Fast,” “Thinking,” or “Pro,” depending on how much speed versus precision your task needs. For quick, casual edits, Fast is usually enough; for anything with text, complex lighting, or a detailed composition, Thinking or Pro will generally produce a cleaner result.
Step 2: Upload the Photo You Want to Edit
Tap the upload or attachment icon and select the image you want to work with. You can upload a personal photo, a product shot, a screenshot, or even a rough sketch. Nano Banana can also accept two or more images at once if you want to blend or combine them, which is covered in more detail below.
Step 3: Write a Clear, Specific Prompt
This is the step that determines whether your result looks intentional or generic. Describe exactly what you want changed, using natural, conversational language rather than technical jargon you’re talking to the model like you would a person, not typing search-engine keywords.
A useful formula for both generation and editing prompts is to describe the subject, the action or change, and the scene: instead of “change the background,” try “replace the background with a sunlit beach at golden hour, keep the subject and lighting on their face exactly the same.” The more specific detail you provide, the more accurately Gemini follows your intent, and the less likely it is to make unwanted changes elsewhere in the image.
Step 4: Review the Result and Refine Conversationally
Once Nano Banana returns an edited image, you don’t need to start over if it’s not quite right. Because the model holds context across the conversation, you can simply reply with a follow-up instruction “make the lighting warmer,” “move the object slightly to the left,” “make the jacket navy instead of black” and it will apply just that change while preserving the rest of the edit.
This conversational refinement is really the core of what makes Nano Banana different from a typical single-shot image generator, so it’s worth leaning into rather than rewriting your prompt from scratch every time.
Step 5: Try Redoing With a Higher-Quality Model (Optional)
If you’re a Google AI Plus, Pro, or Ultra subscriber and want a more polished version of a result, open the three-dot menu on the image and select “Redo with Pro” to regenerate it using Nano Banana Pro’s higher-fidelity model useful for final versions of anything with text, fine detail, or a complex composition.
Step 6: Download or Continue Editing
Once you’re happy with the result, download the image directly from the chat. If you want to keep building on it turning a still image into a short video, for example you can upload the edited image back into Gemini and continue from there.
Common Nano Banana Editing Techniques
Changing the Background
Upload a photo and prompt something like: “Change the background to a modern minimalist office, keep the subject and their pose exactly the same.” This works well for previewing home decor changes, professional headshot backgrounds, or product photography without a physical reshoot.
Removing or Adding an Object
Nano Banana’s pixel-perfect editing is well suited to targeted removals and additions describing exactly what should disappear or appear, and specifying that everything else should stay the same. For example: “Remove the trash can in the background, leave everything else in the photo unchanged.”
Swapping an Outfit or Style
You can prompt a costume, outfit, or era change while keeping the person’s likeness consistent something the model was specifically tuned to handle well. Try: “Change the outfit to a formal navy blazer, keep the person’s face and pose exactly the same.” This also works for decade-style makeovers, like 90s grunge or 80s preppy looks, using the same likeness-preserving approach.
Restoring or Colorizing Old Photos
Upload a scanned or damaged photo and ask Nano Banana to repair scratches, sharpen detail, or add natural color to a black-and-white image. Being specific about what’s damaged “repair the crease across the middle and restore natural skin tones” produces better results than a vague “fix this old photo.”
Blending Multiple Photos Into One Scene
Nano Banana can accept more than one image and combine them into a single, coherent picture for example, placing yourself in a photo alongside a pet from a separate picture, or merging two product shots into one composite. Upload both images and describe how they should be combined: “Put the person from the first photo into the living room scene from the second photo, standing near the window.”
Turning a Photo Into a Figurine or Stylized Character
One of the most popular Nano Banana prompts takes a regular photo and turns it into a miniature collectible figurine, complete with a display base and packaging-style presentation. A simple prompt like “turn this photo into a custom miniature figurine on a display base” is usually enough to get a strong first result, which you can then refine conversationally.
Generating Text-Heavy Designs (Posters, Mockups, Infographics)
This is where Nano Banana Pro specifically shines, since accurate, legible in-image text was one of its core upgrades. You can request posters, product mockups, or infographics with specific text, including multiple languages, and even ask the model to translate or localize text already present in an uploaded image.
Turning Notes or Data Into Diagrams
Upload a photo of handwritten notes or a rough data table, and ask Nano Banana Pro to turn it into a clean diagram, flowchart, or infographic. The model’s real-world reasoning helps it organize the information logically rather than just prettifying the raw text.
Prompting Tips for Better Nano Banana Results
- Be specific, not vague. “Change the background” invites a random reinterpretation of the whole photo. “Replace the background with a foggy forest at dawn, keep the subject’s lighting and pose unchanged” gives the model clear boundaries.
- State what should stay the same, not just what should change. Explicitly saying “keep everything else the same” reduces the chance of unwanted side effects elsewhere in the image.
- Use natural language. You don’t need special syntax or keyword stacking describe the result the way you’d explain it to a person.
- Iterate instead of restarting. Small follow-up instructions inside the same conversation tend to produce more consistent results than repeatedly rewriting the entire prompt.
- Match the model to the task. Use Fast for quick, casual edits; switch to Thinking or Pro (or “Redo with Pro”) for anything involving legible text, complex lighting, or a composition you plan to actually use professionally.
- Double-check text and small details. Even the newest models can misspell words, distort small faces, or misrepresent fine details, especially in complex infographics always review before you publish or print anything.
Nano Banana vs. Nano Banana Pro vs. Nano Banana 2: Which Should You Use?
- Nano Banana (original) the fast, free-tier default for casual editing, quick creative experiments, and everyday photo tweaks.
- Nano Banana Pro (Gemini 3 Pro Image) built for complex, studio-quality compositions: accurate multilingual text, detailed infographics, and high-fidelity branding work, generally with higher usage limits reserved for paid subscribers.
- Nano Banana 2 (Gemini 3.1 Flash Image) the newer default across most Google products as of 2026, combining Nano Banana Pro’s advanced reasoning and world knowledge with much faster generation speed, effectively narrowing the gap between the fast and the high-quality options.
In practice, most casual users won’t need to think much about which version they’re on Gemini automatically selects an appropriate model based on your plan and the tool you’re using but if you’re on a paid tier and want the sharpest possible result for a specific image, the “Redo with Pro” option is worth using deliberately.
Limitations to Keep in Mind
Nano Banana is genuinely capable, but it isn’t flawless, and Google is upfront about a few recurring weak points:
- Small faces and fine detail in busy or distant compositions can come out distorted or lower quality.
- Spelling and grammar in generated text can still be inconsistent, particularly for longer passages or non-English languages.
- Data-driven outputs, like infographics or charts based on real information, should always be fact-checked rather than trusted outright, since the model can misinterpret or misrepresent data.
- Advanced edits heavy lighting changes like day-to-night conversions, masked editing, or blending several images together can occasionally produce unnatural artifacts that need a manual touch-up.
- Every image is watermarked. All Nano Banana outputs include a visible watermark plus an invisible SynthID digital watermark, so viewers and platforms can identify the image as AI-generated worth knowing if you’re using outputs commercially.
Troubleshooting Common Nano Banana Problems
The edit changed more than I asked for. This is the most common frustration, and it usually comes down to prompt scope. If you only say “change the shirt color,” the model has to guess how much else counts as fair game. Add an explicit boundary “change only the shirt color to red, keep the face, background, and lighting exactly the same” and you’ll get far more predictable results.
Text in my image is misspelled or garbled. Switch to the Thinking or Pro model before generating, since text accuracy was a specific focus of those upgrades over the original Fast model. If you’re editing an existing image with a “Redo with Pro” option available, use it for anything with a headline, label, or on-image copy you plan to actually publish.
A person’s face looks slightly “off” after an edit. This tends to happen more with small or distant faces in a busy composition, or after several rounds of iterative editing. Try re-uploading the original source photo rather than continuing to edit an already-edited version, since repeated edits can compound small inconsistencies.
The model won’t make the change I’m asking for. Nano Banana includes safety filters that can decline certain requests, particularly around realistic depictions of real people in misleading contexts. If an edit is being blocked, try rephrasing with more neutral, descriptive language, or reconsider whether the request falls into a sensitive category the model is designed to avoid.
My blended photo looks unnatural. Multi-image blending is one of the more advanced capabilities and can occasionally produce mismatched lighting or perspective between the source images. Choose source photos with reasonably similar lighting conditions and camera angles for the cleanest blend, and expect to do a round or two of conversational refinement to smooth out inconsistencies.
Frequently Asked Questions
Is Nano Banana free to use?
Yes, Nano Banana is available to try for free in the Gemini app, though free-tier users have limited quotas before falling back to the standard model, while Google AI Plus, Pro, and Ultra subscribers get higher usage limits and access to Nano Banana Pro.
Do I need design skills to use Nano Banana?
No. The entire workflow is built around plain-language prompts and conversational follow-ups no editing software knowledge required.
Can Nano Banana edit photos of real people accurately?
Yes, and maintaining a consistent likeness across edits was a specific focus of the model’s development, so photos of people (and pets) tend to keep looking like themselves through outfit changes, style transformations, and scene changes.
What’s the difference between Nano Banana and Nano Banana Pro?
Nano Banana is optimized for fast, everyday edits, while Nano Banana Pro is built for more complex, high-fidelity compositions with better text rendering useful for posters, infographics, and branded content.
Can I combine two photos into one with Nano Banana?
Yes. You can upload multiple images and prompt Gemini to blend them into a single coherent scene, such as placing two people from separate photos into the same picture.
Are Nano Banana images watermarked?
Yes, every image created or edited includes both a visible watermark and an invisible SynthID watermark to identify it as AI-generated.
Where can I use Nano Banana besides the Gemini app?
It’s also available in AI Mode in Search, Google Workspace (Slides and Vids), Google Ads, Flow, Mixboard, and through developer platforms like Google AI Studio and Vertex AI.
