Nano Banana Prompts
Nano Banana is the name of Gemini's image model, and its strength is editing: it keeps a face across edits and can merge several uploaded pictures. These prompts lean on that, with the upload named at the top of each block.
Action Figure Toy Box
Use the attached photo. Keep my face, hairstyle, skin tone and outfit recognisable, and turn me into an articulated action figure sealed in retail packaging: a…
Colourise a Vintage Photo
Use the attached black and white photograph. Keep every face, expression, pose, item of clothing and background detail exactly as they are, and add only…
Consistent Character Sheet
Use the attached photo as the reference for one character. Keep the face, skin tone, hair and outfit identical in every panel. Produce a four panel character…
Dramatic Sky Replacement
Use the attached photo. Keep the landscape, buildings, people and foreground exactly as they are, and replace only the sky. Put in a [time of day] sky with…
Extend The Photo Frame
Use the attached photo. Keep every part of the original untouched in the middle of the new frame: the same people, faces, poses, colours and grain. Extend the…
Isometric Building Miniature
Use the attached photo of a building. Keep its real shape, number of storeys, window pattern, roof line, colour and any distinctive detail. Rebuild it as a…
Logo On A Mug Mockup
Use the two attached images: the first is my logo, the second is a plain mug. Keep the logo's exact shapes, colours and proportions. Place it centred on the…
Merge Two Photos Into One
Use the two attached photos. Keep each person exactly as they appear in their own photo: the same face, skin tone, hair and clothing. Combine them into one…
Outfit From Second Photo
Use the two attached photos. The first shows me, the second shows a garment. Keep my face, skin tone, hair, body shape and pose exactly as in the first photo,…
Product In Lifestyle Scene
Use the two attached photos: the first is my product, the second is the room I want it in. Keep the product's shape, colour, label and proportions exactly as…
Remove Unwanted Objects
Use the attached photo. Keep the people, their faces, poses and clothing exactly as they are, along with the framing and the light. Remove only the clutter I…
Room Furniture Makeover
Use the attached photo of my room. Keep the walls, windows, doors, floor, ceiling and the camera angle exactly as they are. Replace only the furniture: a…
Sketch To Product Render
Use the attached sketch. Keep the design exactly as drawn: the same silhouette, proportions, part lines and details, with nothing added and nothing simplified…
Style Transfer From Reference
Use the two attached images: the first is my photograph, the second is a style reference. Keep the subject, composition, faces and layout of the first image…
Nano Banana Prompts: questions people ask
What is Nano Banana and where do I paste these prompts?
It is the nickname that stuck to Gemini's image model, so you work inside Gemini rather than in a separate tool. Attach whatever pictures the block asks for, paste the text, and send both in one message so the instruction arrives with the files. If the reply comes back as words instead of a picture, say you want an edited image returned and it will usually comply.
How does it know which uploaded photo is which?
By the order you attach them and by what the text says about each one. Outfit From Second Photo opens by stating that the first picture is you and the second is a garment, and Style Transfer From Reference does the same with a photograph and a style. Attach them in that order. If you need more than two, check what the interface currently accepts.
Why does it change parts of the photo I wanted left alone?
Because anything you do not protect is treated as open. Write the hold list out, as Remove Unwanted Objects and Room Furniture Makeover do, naming the background, the pose or the face as untouched. Then make one change per message and keep going in the same conversation, so each result builds on the last rather than starting again from your original upload.
Do I get my original resolution back, and is the output marked?
Not usually. It generates a fresh image rather than patching your file, so the result comes back smaller than a modern phone photo and a touch softer. Plan for that before editing something you mean to print. Pictures from this family of models also carry an invisible marker identifying them as generated, which survives ordinary editing. Check the current documentation for size limits.
More prompt collections
AI Photo Prompts AI Image Prompts Sora Prompts AI Video Prompts Midjourney Prompts
New prompts in your inbox
One email a week: the best new prompts from the library and one short guide on prompting. No spam, unsubscribe any time.