Nano Banana: Google DeepMind upgrades Gemini app with advanced AI image editing model

Google DeepMind has rolled out an upgraded image editing model in the Gemini app, offering global users new tools such as photo blending, multi-turn editing, and design mixing. The update ensures likeness accuracy and includes visible and invisible AI watermarks.

Hyderabad: Google DeepMind has unveiled a major upgrade to image editing in the Gemini app, integrating its latest model that early users have rated as the world’s top-performing tool in this category. The feature is now available to both free and paid users globally.

The update focuses on maintaining likeness in photos, ensuring that subtle details of people, pets, and objects remain consistent across edits. For example, users can now experiment with hairstyles, outfits, or pet accessories while retaining a realistic resemblance.

---

Among the new features is photo blending, which allows users to merge multiple pictures into a single scene. For instance, a user can combine a personal photo with one of their pet to create a shared portrait.

The upgrade also supports multi-turn editing, enabling step-by-step modifications. A user can start with an empty room, change wall colors, and then add furniture without disrupting the overall image.

Another addition is design mixing, which lets users transfer visual elements such as colors or textures from one image to another. This could allow someone to create rainboots inspired by flower petals or design a dress using butterfly wing patterns.

Google confirmed that every image created or altered in Gemini will carry both a visible watermark and an invisible SynthID digital watermark to indicate that it is AI-generated.

With the new model, users can combine photos, preview home décor changes, or even place themselves in different locations and eras, while retaining a realistic likeness.