239 prompts by 57 creators, free to copy, each credited to the person who wrote it. 150 of them have editable fields you can fill in before you copy.
Google's image model, and the largest collection here. The prompts lean heavily on identity-preserving edits and on structured, JSON-shaped instructions — a sign the model responds well to being told exactly what goes where.
Many of these carry editable placeholders, so you can swap the subject without rewriting the prompt.
This is the largest collection in the library, and it has a clear character. A big share of the prompts edit a photo you supply rather than inventing a scene from nothing: "keep the face from the reference image", "use the person in image 1", "turn this selfie into…". Many are written as structured instructions, sometimes literally as JSON, with separate fields for subject, outfit, background, lighting and camera.
That structure is worth copying even if you use another model. Splitting a prompt into labelled parts makes it easy to change one thing, such as the outfit, without disturbing the rest.
The prompt: "Using the person in image 1: keep their face, hairstyle and glasses exactly. Subject: the same person as a 1990s bookshop owner. Outfit: a moss-green cardigan over a white shirt. Setting: a cramped bookshop at closing time, warm lamps. Camera: 50 mm, eye level, shallow depth of field."
Points at the photo, the habit this collection relies on.
Subject, outfit, setting and camera are separate, so you can change one at a time.
The identity lock comes before any change.
Press Generate this on any prompt to open Create Image with the prompt filled in. Nano Banana Pro is one of the models there, and prompts that need your photo ask for it before they run. Generating uses credits; copying the prompt to use elsewhere is free. Prefer one tap and no prompt at all? The AI photo apps carry tuned versions of the most popular looks.
Google's image model, and the largest collection here. The prompts lean heavily on identity-preserving edits and on structured, JSON-shaped instructions — a sign the model responds well to being told exactly what goes where. This page collects 239 prompts written for it by 57 different people.
From YouMind OpenLab's open collection of Nano Banana Pro prompts on GitHub, shared under CC BY 4.0. Each was first posted by an individual creator, who is named and linked on every card and prompt page here, as the licence requires.
Open the prompt, fill in any fields, and press Generate this. Create Image opens with the prompt loaded and asks for your photo when the prompt needs one. You can also copy the text into any other app where you use Nano Banana Pro.
No. Copying needs no account and is always free; only generating an image on PixelShine uses credits.
150 of the Nano Banana Pro prompts here have fill-in fields, such as the subject or the text on a poster. They appear as input boxes on the prompt's page; change one and Copy gives you your version.
Prompt data adapted from YouMind OpenLab — YouMind OpenLab — licensed CC BY 4.0. Each prompt credits its original author and links to their post.