Google has released Nano Banana 2, an upgraded AI image generator that is faster and more powerful than its predecessors. It is now the default model in the Gemini chatbot and integrates real-time web search for tasks like creating infographics, though early testing shows it can sometimes produce inaccurate information.
The tool excels at detailed photo editing and generating photorealistic images from user prompts and uploaded photos, convincingly altering scenes and replicating fine details. However, its outputs can be inconsistent, with occasional unconvincing or flawed generations.
The article serves as a reminder of the advanced, yet imperfect, capabilities of AI image tools and the need to critically assess such content. It also provides instructions on how to access Nano Banana 2 through various Google services.
Main Topics: Google Nano Banana 2 AI image generator release, its features and capabilities, hands-on testing results and examples, access methods, and commentary on AI realism and reliability.
Google just debuted Nano Banana 2, an updated version of its AI image generator. It combines the abilities of Google’s previous release, Nano Banana Pro—like text rendering and web searching—with speedier image generation. This tool will be the new default in Google’s Gemini chatbot.
The first image model from Google under the Nano Banana moniker dropped last August, and the Pro version arrived three months later. The AI tool was widely adopted online to alter photos of real people, from generating custom action figures to nostalgic images of people hugging younger versions of themselves.
Nano Banana 2 is not only faster at crafting images, it’s also a more powerful photo editor. Despite some rough edges and unconvincing generations in my initial hands-on experience through Gemini, Google’s latest release marks the continued improvement of photorealistic AI tools that can manipulate existing images and serves as a stark reminder to always scrutinize unverified images you see online.
Getting Started
If you want to try the new image model, the easiest way to access Nano Banana 2 is through the Gemini app or website. You can either click the banana emoji to generate images or just put the request in your prompts to the chatbot. This new image model is also available through Google’s Search tools, AI Studio, Cloud, and other services.
Google says the Nano Banana 2 image generator pulls real-time information from the web, which can be useful for generating infographics. To test this, I asked Gemini to generate a custom weather report for my upcoming weekend getaway. Here’s my prompt:
I'm going skiing in Dodge Ridge this weekend with some friends. Could you create an infographic that covers the weather conditions?
At first glance, the result looks decent. No wobbly text or disfigured skiers in the background. The forecast for each day includes expected temperatures as well as wind and snow conditions. A small disclaimer at the bottom of the infographic reads, “Weather and conditions subject to change. Check official sources.”
I’m glad I did! When I looked up the forecast for this weekend from a different source, I realized that Gemini had messed up the dates and pulled the Google Weather context from last week. When I pointed out this mistake to the bot, it used Nano Banana 2 to replace the text from its first attempt with the correct weather data.
Tub Time
If you want more details about my getaway, I’m headed to a cozy ski lodge with friends who are skiers. I’m a novice and still deciding whether to actually hit the slopes or just turn into a wrinkly prune sitting in the hot tub all day long. Maybe Nano Banana 2 could make a dumb meme to send to the group chat? I uploaded a photo of myself to Gemini with this prompt:
Take this image and put me in a cozy outdoor jacuzzi surrounded by snow. Make my skin comically wrinkly from sitting in there for hours.
This result is upsetting to look at. The pruny wrinkles make me look like I’m 80 years old, more than just over-hot-tubbed. Also, the bot kept my shirt on for some reason. Still, the small details Nano Banana 2 mimicked from the original photo, a random bar bathroom selfie, are impressive.
The shirt’s oddball design was intact and faithfully re-created for the sections not visible in the original snap. I even had the same chain jewelry on my underwater hand in the AI-generated image. Overall, the outdoor jacuzzi scene looked photorealistic and convincing, with a snowcapped cabin and evergreen trees in the background.
Photo Real
How good is Nano Banana 2 at generating some delusional, fantasy images for social media? If I decide not to ski this weekend, I still need a couple of images of me looking hot to post on my Instagram stories. It’s not relaxation if you don’t document it. Or, at least, faux-document it. I uploaded another photo of myself with an unhinged ask:
Create a photorealistic image of me hitting the slopes. I’m ripped and shirtless, powder-blasting everyone with my intense skiing speed.
I laughed out loud at this one. If you cover up my face, it’s a cool action shot of snow flying everywhere. Each hand has the normal number of fingers, a stumbling block for AI of yesteryear. The ski goggles even appear to have an Oakley brand insignia on the side. That face though! It looks like the bot cut a paper printout of my head with scissors and decoupaged it on top of a fitness model. Complete cringe.
While this attempt was a flop, Google is getting even better overall at manipulating images with Nano Banana 2. And even though the company watermarks its outputs to identify these images as AI-generated, it's easy to overlook these signs while scrolling through social media, especially as the outputs improve in quality.
Nano Banana 2 is free to use and widely accessible in the Gemini app. With this release, Google is making it even easier to alter images of people into situations that never existed, and anyone can pump out highly specific, yet fictitious, images faster than ever.
Meridiano
