Exploring the mysteries of Nanobanna (aka Gemini 2.5)

Google's new image model was codenamed nanobanana. Why? We don't know, but we agree it's bananas. Like OpenAI’s GPT4o, this is a model that can take instructions as well as descriptions. It's a big step forward, especially if you are working with existing images.

This model can understand your image, edit specific parts while preserving details, and reimagine it. Gemini 2.5 hints at the future of image making. PictureStudio gives you an easy, consistent way to work with this model, and in the near future, we'll give you specialized tools to tap the power of this model.

We're excited because for the first time, you can include an image of a person or a product in a prompt and consistently get reimagined while retaining likeness. This is a big deal.

Editing is easy with this model once you crack the instructive prompting. You can add an image of a person and a prompt for a different style of clothing, hair, or setting. Change the seasons or the weather. It's easy.

We particularly like how good this model is at transforming drawings and paintings. Sketch a character and ask for different poses, or request another scene drawn in the same style. It's all dead easy, fast, and cheap. In the progression above, we started with a painterly style and then decided to illustrate a couple in that style and then redressed and reposed them.

What will you do with this tool?!