I built Banana Studio, an image editor that lets you modify specific regions inside an image using simple text instructions. It runs fully client-side in the browser and uses Google’s Gemini Nano Banana for fast, clean edits.

Demo video: https://youtu.be/-DbDDsyT2MM Link: https://banana-studio-nano.vercel.app/

Why I built this Text-only prompts make it challenging to instruct the model precisely where to apply changes, so I developed a simple bounding-box interface that eliminates all the guesswork. The idea was to keep everything lightweight, browser-based, and still get clean, high-quality edits.

How it works You upload an image, draw a box on the canvas, and type what you want changed. The app maps your box to the 1000x1000 coordinate grid Gemini expects. It then builds a structured prompt telling the model exactly which region to modify, sends it to the API, and gets an image back.

Features - Edit specific regions by drawing one or more boxes - Different prompts for different boxes - Global enhancements when no region is selected - Works entirely client-side - Bring your own Gemini API key (stored locally in your browser)

If you want to try it, I would love feedback from the HN crowd.