Gemini can now browse a vast library's worth of images about you and produce its own AI creations that are based on them. The AI chatbot's Personal Intelligence features, designed to help Gemini adapt to your preferences and personality, are now, with your permission, able to browse your Google Photos library.
The Nano Banana 2 image-making model can then use that information to shape its visuals without lots of explanation on your part.
Instead of carefully writing prompts that explain your appearance, your environment, or your general vibe, you can simply say “me” and let the system handle the rest. Once you give Gemini permission to access your photos, you type a prompt as usual. The difference is that Nano Banana 2 now fills in the blanks using what it already knows. The output is not just based on your words, but on a version of you assembled from your own images.
Cinematic life
To see how well Gemini now knew me, I gave it a fairly open-ended prompt. “Create a cinematic scene from my life as if it were a movie.”
The result leaned heavily into an atmosphere of anticipation. A very good AI likeness of me stares pensively out at the rain in some sort of old hotel. I am apparently planning a big adventure, judging from the notebook, compass, and small plane outside, but I look as if I haven't made a decision.
It's essentially a more dramatic version of many photos of me on trips or in hotel lounges with guidebooks. A cinematic, more narratively interesting version of many of my photos.
Fantastic me
Next, I wanted to see if Gemini would still be able to recognize me if it put me in a less realistic setting. So I asked it to "turn me into a character in a fantasy adventure.”
This time, realism disappeared in favor of scale and spectacle. I was placed on a winding stone path that stretched toward a towering castle carved into a mountainous landscape. My clothes had transformed into layered armor and leather, complete with a satchel marked by glowing symbols. A staff replaced the compass, emitting a faint light, while strange creatures hovered nearby as if they were part of the same ecosystem.
Despite the setting, the continuity was impressive. The face was clearly mine, not a generic placeholder but a direct translation from my photos. The expression, slightly serious, slightly focused, carried over without exaggeration. Even the posture felt familiar.
That consistency made the image more effective than it might have been otherwise. It did not feel like a fantasy character loosely inspired by me. It felt like me inserted into a fantasy world, which is a very different experience. The system was not inventing a new identity. It was adapting an existing one.
Hobby extreme
As Gemini had proven it knew what I looked like and what a movie version of me might be like, I asked for a more holistic, if over-the-top interpretation. I asked Gemini to “create an image of me doing my favorite hobbies in an extreme or exaggerated way.”
The result was far less restrained. I appeared in a vividly stylized scene that blended multiple interests into a single, energetic composition. Books floated midair, pages turning as if caught in a controlled storm. A small plane cut across the background, while futuristic cityscapes and distant planets filled the horizon. I was mid-motion, holding objects that hinted at different pursuits, reading, exploring, creating, all dialed up beyond anything that would happen in real life.
The tone was exaggerated but not random. Each element seemed to connect back to something the system had inferred from my photos and context. Except for the wings. I don't know what they were about, and Gemini seemed at a loss to explain the reason behind it. The illustrative style was bold, but even so, the underlying likeness remained consistent.
The feature sets up a much more powerful way to teach Gemini about yourself. Traditional AI image generation depends heavily on how well you describe what you want. Here, description becomes secondary. The system is already working from a base layer of information.
Prompts become shorter and easier to write, and results feel more specific without requiring extra effort. How comfortable you are in sharing that information with Google is another question, of course. The company is good at encouraging people to upload their life story, but in this case, if you already use Google Photos, you're not adding to the database Google already hosts for you. And the frictionless way you get from a brief idea to a personalized artwork is undeniably appealing. Even if it makes you part bird for no particular reason.
