Google Gemini now produces images by looking at your photos

Google is adding a new feature to its Gemini artificial intelligence platform that makes visual production more personalized. The company now carries its “Personal Intelligence” approach, which previously customized responses with Gmail, Search and YouTube data, to the visual creation side. In this context, the model called Nano Banana 2 can produce more contextual and personalized images by using data connected to the user’s Google account.

With this update, the need for users to provide detailed explanations to Gemini is reduced. The system directly contributes to the visual production process by analyzing YouTube history, e-mails and especially the content in the Google Photos archive. For example, when a command such as “draw my desert island items” is given, the model can create an image shaped according to the user’s interests. This approach allows text-based commands to be kept shorter and more natural.

Google Photos integration stands out as one of the most striking components of this system. Person or pet tags in users’ photos can be used as references in the visual production process. In this way, when given a command such as “create a hand-drawn portrait of my mother”, Gemini can find the right person and prepare a suitable visual for him. Despite this, the fully automatic operation of the system can sometimes produce unexpected results, and user intervention still plays an important role.

More contextual images with Google Photos integration

When the created images do not meet expectations, users can reshape the result with additional commands. In addition, it is also possible to direct the production by selecting a different image from Google Photos. The “Sources” button on the interface provides transparency in terms of showing which images the artificial intelligence uses as a reference. If they wish, users can also directly request the system to explain the resources used.

In addition, the use of personal data on this scale also brings about privacy debates. Google states that users have control over this feature and can manage what data is used. Nevertheless, the inclusion of personal photo archives in artificial intelligence processes is considered an area that needs to be carefully monitored, especially in terms of sensitive data security.

The new feature is currently available to AI Pro and AI Ultra subscribers in the Gemini application. However, Google states that the integration will soon be opened to the Gemini experience accessed via Chrome and to a wider user base. It will become clearer over time how this expansion process will proceed and what the performance results will be in different usage scenarios.

Although the system in its current form provides obvious ease of customization, the balance of accuracy and control still seems to depend on user interaction. In particular, situations such as incorrect matches or context errors may require manual corrections. Despite this, Google’s extensive data ecosystem provides an important basis for such integrations to become more advanced.