Google’s new AI image creator took my shirt off

I tried Google’s new Nano Banana Pro and it instantly took off my clothes. I didn’t ask for it, but the AI ​​model apparently decided that my greeting card would look better with more skin.

As the name suggests, Nano Banana Pro is for professionals. Powered by Gemini 3, it’s effectively an upgrade to the company’s popular image generation and editing tool that became a viral social media trend that turned selfies into surreal 3D sculptures. Google says it lets you create high-quality images that you can print, render legible text on images, and blend multiple images together into a single composition. “It’s also for people who want to feel like professionals,” said Naina Raisinghani, product manager at Google DeepMind. The VergeIt feels good, because I’m not a professional by any means, To me, the results were dazzling, but silly, It looked good, but felt amateurish,

Using the Nano Banana Pro is very simple: you go into the Gemini app, select “Draw a picture” and toggle on ‘Thinking’ mode. Just plug in your prompt (and image, if you’re using one) and go. It is also free, although it has some limitations, the quota is expanded for Google AI Plus, Pro and Ultra customers.

Google makes some bold claims, promising “Studio-quality design,” “Flawless text rendering,” and a number of great and creative edits. To test these, I uploaded a simple photo of my The Verge’s Office in New York with the Brooklyn Bridge in the background. I asked Gemini to change the lighting from day to night and it worked great. The results seem reliable. It also handled details that often bogged down the image generator, such as cars going in the right direction. Adjusting the camera angle was equally easy. I asked Gemini to recreate the shot as if it was taken from a higher angle on the right and it did.

Google also says the Nano Banana Pro can create infographics and diagrams to help visualize real-time information like weather or sports. Being British, I asked about the weather for the next four days in Washington, DC and New York City, where I am currently. Visually, the infographic would have been at home on a basic forecasting site. The text and numbers appeared normal – a far cry from the garbled nonsense you often see in AI-generated images – and Gemini gave me a list of quotes at the end that helped me confirm it was accurate.

The model faltered a bit in more complex tasks. I asked him to summarize his recent the verge The story is about how Europe is rolling back its AI and privacy laws, in comic book-style format. The illustrations and text were indeed flawlessly rendered in a cartoonish font, but the comic did not summarize the story, instead giving a vague overview of Block’s AI act. The problem probably occurred because I gave Gemini a link to the story instead of pasting the text.

When I did this it gave me a summary of a popular comedy genre. It gave the gist of the actual story, although I don’t think I would have understood as easily if I hadn’t written the source material. It also contained phrases that did not appear anywhere in my article.

To really feel like a professional designer, I tried my hand at creating greeting cards. Christmas is finally coming. Considering I only uploaded three selfies, Gemini clearly did an amazing job creating three full-body versions of me, each in a different outfit and with a different facial expression. It also created a realistic, snowy setting with Christmas trees, as I asked it to do, and “Merry Christmas!” On top like I asked her.

Gemini took liberties when I asked her to transform the card’s snowy background into a summer beach for an Australian-style vacation. Those liberties were my deepfake clothes: two of my clones were topless. It was strange. There were also some prominent AI-generated legs and a smiley Sandman to replace the snowman from the winter scene (being created by my topless lookalike). However, there were a few issues – unlike other presented objects in the picture, the Sandman had no shadow, and the Christmas lights in the palm trees were magically glowing in the bright sun. I tested its precise editing skills by simply asking it to add some muscles to a clone, which it did in seconds (if only it were that easy in the real world). Overall, the quality was fantastic, and the image would have been somewhat believable (except for the abs) if you didn’t know I was missing a large tattoo on my chest.

However, it was not all great. The model failed to preserve the exact text on my card that I asked it to. “Merry Christmas!” Instead of “Australian Summer Christmas!” Opted for. It also seems to conflict with the animals: My sister’s cat is sitting in exactly the same crouched pose as the reference image in every version of the card I provided (though, she was given a whimsical Santa hat).

Overall, I was impressed. The Nano Banana Pro is a clear upgrade over the original model. I was able to ask for more precise edits and it actually produced sensible text, removing a major barrier preventing the use of generative AI tools like this in the real world. But alas, these qualities were not enough to make me a good designer.