Creative editing has taken a huge leap forward thanks to the arrival of Nano Banana integrated into Gemini, Google's system that lets you manipulate images with natural language instructions. You no longer need to open a complex editor or download apps: upload a photo, type what you want, and get the result in seconds.
The best is that only what you ask for is modifiedThe rest of the photo remains unchanged, respecting framing, lighting, and style. Plus, the model understands the instructions much better than before, allowing you to combine changes (color, background, clothing, text, etc.) without breaking the coherence of the image. All of this is available for free in the Gemini experience.
What is Nano Banana and how does it fit into Gemini?
During its internal testing, Google used the codename “Nano Banana” for your Gemini 2.5 Flash Image model, specializing in image editing and generation. This technology lives within Gemini and works on both the web and mobile, without any additional installations or complicated configurations.
The user experience is direct: You can generate images from scratch or upload a photo and edit it. with natural phrases. Ask "make this photo black and white," "remove the person from the background," or "add a clear sky," and you'll instantly see accurate versions of what you requested.
At the access level, there is free use with reasonable daily limitations and, for those who want to integrate it via API into products or business flows, the reference cost is $0,029 per processed imageThis way, creators and organizations can scale without any surprises on the bill.
Essential tricks to improve photos without external apps
Edit and enhance your images
For basic adjustments you no longer need Photoshop: ask Gemini to Increase contrast, boost colors, or improve clarityYou can even make technical requests like sharpening, adjusting exposure, or balancing whites.
- Prompt example: “The image looks dull; raises the contrast by a point, enhances the colors and slightly improves the sharpness.”
Apply creative styles and artistic filters
The traditional filters are just a click away, but now with extra power: you can order “charcoal”, “mosaic with tiles”, “anime drawing” or almost any style, and the system will adapt the photo respecting its key features.
- Prompt examples: “I want the photo look like a charcoal portrait”; “Turn the image into a tile mosaic”; “Turn this scene into an anime drawing.”
Change colors of specific objects
Do you want to test how an object would look in a different shade or visualize an aesthetic change? Indicate which element and what color, and the system recolors the area while maintaining texture and light of the material to make it look believable.
- Prompt example: “In the photo, puts the vinyl in dark green without altering the shadows of the plate.”
Eliminate objects and also people
The element erase function is surprisingly robust: it detects the area, removes it and reconstructs the background in a coherent manner, even taking care of shadows or occlusions so that the retouching is not noticeable.
- Prompt example"Remove the cars that appear in the background of this image.”
- Prompt example: “Eliminate the person on the right and rebuilds the environment without leaving a trace.”
Replace one element with another
In addition to deleting, you can replace an object with a different one while maintaining geometry, lighting, and perspective. It's useful for both aesthetic touch-ups and polishing sensitive or advertising content.
- Prompt example: “In this photo, exchange the small bottle of beer for a bottle of water preserving the frame.”
Restore old photographs
AI helps recover damaged photos: it removes cracks, removes stains, reinterprets lost areas and, if requested, apply natural-looking colorThe result depends greatly on the original quality, but the improvements are noticeable.
- Prompt example: “Restore this old photo: erases cracks and imperfections, color it with realistic tones and improve clarity and definition.”
Advanced changes to control the scene
Zoom out (outpainting)
If your photo is very closed (only face or half body), you can ask the system to “zoom in” to show more of the sceneThe more you have to invent, the lower the fidelity, so describe in detail what you want to see.
- Prompt example"Enlarge the frame up to full body: I'm sitting on the ground, with a festival fence behind me and an empty stage in the background.”
Modify facial expression
Without touching the rest of the photo, you can ask Changes in gesture: slight smile, surprise, excitement…The engine respects the structure of the face so that it doesn't look artificial.
- Prompt example"Make him look moved and sad“without modifying the hairstyle or lighting.”
Costumes, hairstyles and disguises
Try different looks without going through your closet: you can change clothes, try out new hairstyles or dress someone in outfits you upload as a reference. It helps a lot if the clothing and the person have a similar angle.
- Prompt example"Dress him with the tracksuit which appears in the second photo, maintaining pose and light.”
- Prompt example: “Change your hairstyle for shoulder-length black curls in the foreground.”
- Practical tip: better half-body or full-body shots and, if you share reference clothing, try to make it bottomless and the point of view matches.
Change the background or move characters to other locations
Cut and paste has become trivial: describe the environment and The system remakes the background without touching the subjectIdeal for varying settings, creating fictional scenes, or adjusting the tone of a session.
- Prompt example"Move this woman and her dog to a meadow next to a dairy farm in the evening light.”
- Prompt example: “Change the background of the first photo to the place that appears in the second".
Merge and combine images
Nano Banana allows you to integrate elements from several photos into a new scene: you can seat a character in front of objects from another image or overlay artwork on frames and screens with visual coherence.
- Prompt example"Combine these images so that the woman is having that toast with that coffee for breakfast.”
Add elements based on others (chained prompts)
A useful technique is to work in steps: first you generate the base and then you add logos, labels or brands as a second instruction, preserving the previous composition.
- First prompt: “Portray a person with great CEO look, dark background and confident gaze; it retains its features from the original photo.”
- Second prompt: “Keep the previous image and adds relief and shadow the logo attached at the bottom without covering the face.”
Text editing in images and editorial compositions
Another very practical function is change numbers or words on signs, posters or t-shirts, maintaining the original typography and style so that the retouching is not noticeable.
- Prompt example"Replace the number 2 with a 3 with the same source and texture.”
If you want to go further, you can create your version of a magazine cover replacing the main photo and some lines of text. It's more complex and there may be minor flaws, but the results are amazing.
- Prompt example: “The first image is a magazine: replace the cover photo for the second photo I attached. Change 'Living on the Moon' to 'Satanic Cats' and the subtitle to 'They're Watching You from the Darkness.'"
Recommended workflows in Gemini
Start by uploading your image and write a clear instruction. Sometimes it's enough to simply "make it black and white"; other times, it's better to start with “In the original photo…” to make it clear that you want to keep the content and only change one detail.
Work in iterations: make an edit, observe the result and add a second prompt with the next change, without reloading the photo from scratch. This step-by-step approach gives you fine-grained control and avoids surprises.
The more specific you are, the better: provide the exact color, precise location, lighting style, texture, relative size… Detailed prompts improve quality and homogenize shadows and perspective.
3D modeling and product mockups
In addition to edit photos, there are product-oriented flows that can speed up your work. You can request that an image be generate a 3D model with a scene desktop and the result appears on the monitor showing the modeling interface.
- Prompt example: “Turn this photo into a 3D model and create a background with a 3D modeler's desktop; place the figure on the table in front of the monitor with the CAD interface visible.”
- Council: if you are not looking for a complex look, use simple backgrounds for a cleaner result.
For packaging, upload a white mockup and the design pattern, and ask for it to be apply the chart to the template with a minimalist aesthetic, for example on a white table.
- Prompt example: “Apply the pattern in Image 1 to the packaging mockup in Image 2 and place the pack on a white table with a simple and fresh style.”
Practical tips for realistic results
For believable changes, ensure consistency of light, shadows, and scale. If you move a person from one location to another, Describe the time of day and the type of lightingIf you change clothes, try to have the reference share the same angle and focal length.
- Use clear and concrete phrases; avoid ambiguities in the “make it pretty” style.
- When adding objects, indicate relative size (“the size of an A4 book”) and position (“on the left, on the ledge”).
- If the system fails on a detail (e.g. complex hands or fonts), asks for a new variation or clarify the prompt more precisely.
For creators, brands and content professionals, this versatility allows test versions in minutes without the costs and time of a 100% manual flow, opening the door to more iteration and better visual decisions.
FAQs
It's free?
Editing with Nano Banana within Gemini can be used for free with daily limitsIf you need to integrate the technology into a product or backend, the price reference via API is $0,029 per image.
Where and how is it used?
It is integrated into the Gemini app and web version; you don't have to install anything additionalYou can also access the Nano Banana project site, log in, and start creating without any hassle.
The typical flow is: you upload your photo, you give an instruction in natural language, you review the result, chain another prompt if you want another change and download the final image in high resolution.
Are there ways to save money if I take the career leap?
If you are going to consume API, consider annual plans (they are usually cheaper than monthly ones), be aware of promotions and discounts, and check if your Google One subscription gives you extra credits or benefits.
What about the videos?
In the Gemini ecosystem you can summarize YouTube videos or ask questions about its content on Android. The “Nano Banana” layer we're dealing with is focused on images, but it fits into a set of multimodal capabilities also geared toward video.
Sample prompts ready to copy
- Basic improvement: “Increases contrast and color richness maintaining natural lighting.”
- Artistic style: “Turn the photo into a charcoal portrait with paper texture.”
- Recolored: “paint the furniture of navy blue respecting veins and reflections.”
- Deleting objects: “Remove the trash cans in the background and rebuild the wall.”
- Deleting people: “Delete the person which is on the right and fills the background naturally.”
- Substitution: “Change the car for a van white with the same perspective.”
- Old retouch: “Restore and color “This old photo improves sharpness and removes scratches.”
- Zoom out: “Zooms the frame up to show full body“I’m standing next to a festival fence with an empty stage behind me.”
- Expressions: “Softens the smile and adds a gesture of slight surprise.”
- Clothes: “Put the suit from the second image and maintain the original pose.”
- Hairstyles: “Generates four different hairstyles in a 2×2 grid in the foreground.”
- Backgrounds: “Change the background to a meadow next to a farm with golden light.”
- Fusion: “Combine these two photos so that she can have the other one's toast and coffee for breakfast."
- Logo in two steps: “Portray it as a CEO with a dark background and a confident look." Then: "Keep the image and add This embossed logo “at the bottom without covering the face.”
- Text change: “Replace the number 2 with 3 with the same typography and thickness.”
- Cover: “In the magazine, replaces the cover for my second photo and change the headlines to the ones I indicate.”
- Packaging mockup: “Apply the pattern from Image 1 to the white box in Image 2 on a white table, clean style.”
- 3D: “Turn this photo into 3D model and place it on a desktop with the CAD interface on screen.”
If you ever thought it was science fiction to erase people from a photo or change the background completely, now you can do it with a phrase and in seconds; with Nano Banana inside Gemini You can enhance, stylize, or reconstruct images, combine scenes, and play with costumes and text, all without breaking what already works in your photo. The key is to clearly describe what you want, iterate with chained prompts, and take advantage of the fact that AI increasingly understands and respects the visual context. Share this tutorial and more users will learn about the potential of Google's Nano Banana..