Awesome-Nano-Banana-images

What it is: A curated gallery of example images and the exact text prompts used to create them, all generated by Google's Nano-banana and Nano-banana Pro image models. Think of it as a showroom: each entry displays an impressive output image alongside the prompt that produced it, so you can see what the model can do and learn how to replicate similar results.

Who it's for: Anyone curious about what these image models can accomplish — designers, hobbyists, researchers, or people just exploring AI image generation. The examples come from social media posts (Twitter/X, Xiaohongshu, etc.) and are organized by use case.

What's inside:

• 52 "Pro" examples — advanced use cases like generating a card in the style of a Japanese woodblock print, creating a flowchart from a research paper, turning an article into a slide deck, or adding a giant creature to a city photo.
• 110 standard examples — broader creative tasks like turning an illustration into a figurine, recoloring old photos, designing a chess set, or creating a custom phone wallpaper.
• Each example shows the output image and the full prompt used. Some also list required inputs (e.g., "upload a reference photo of a person").
• Prompts are often in Chinese (the README is primarily Chinese), with some containing placeholder fields like {角色名字} that you fill in.

Bonus dataset: The project also introduces Nano-consistent-150k — a dataset of over 150,000 images built on Nano-Banana, designed to keep a person's identity consistent across many different editing tasks. For the same portrait, it provides 35+ different edited versions (different poses, styles, contexts), which enables building "interleaved" data that links multiple editing tasks around one consistent character. It's available on Hugging Face.

Also included:

• A news log tracking updates from August 2025 through September 2026.
• A link to a companion Editable-Design project — an interactive design tool combining a coding agent with image generation.
• Translations of the README into English, Japanese, Korean, Spanish, and Turkish.
• Licensed under CC BY 4.0 (free to share and adapt with attribution).

In short: It's a practical, community-sourced prompt library and showcase for Google's Nano-banana image models — plus a large research dataset for identity-consistent image editing.