Nano Banana I2I

Nano Banana I2I (image-to-image mode) allows users to input an existing image plus a text prompt to transform, edit, or restyle parts of the image. The model interprets your instructions to apply changes—such as altering backgrounds, swapping objects, adjusting lighting or colors, or repositioning subjects—while preserving the integrity of elements not involved in the edit. It supports identity consistency (e.g. keeping a person’s face recognizable across edits) and style coherence, so edits feel natural and harmonious. The system handles fused image operations too, such as merging elements from multiple reference inputs. Nano Banana I2I is designed to understand spatial relationships and context, making localized edits (e.g. “change the car color to red”) without distorting unrelated areas.

August 26, 2025

Google

Image to Image (I2I)