Google Improves Gemini AI Image Editing With 'Nano Banana' Model 23
Google DeepMind's new "nano banana" model (officially named Gemini 2.5 Flash Image) has taken the top spot on AI image-editing leaderboards by delivering far more consistent edits than before. It's being rolled out to the Gemini app today. Ars Technica has the details: AI image editing allows you to modify images with a prompt rather than mucking around in Photoshop. Google first provided editing capabilities in Gemini earlier this year, and the model was more than competent out of the gate. But like all generative systems, the non-deterministic nature meant that elements of the image would often change in unpredictable ways. Google says nano banana (technically Gemini 2.5 Flash Image) has unrivaled consistency across edits -- it can actually remember the details instead of rolling the dice every time you make a change.
This unlocks several interesting uses for AI image editing. Google suggests uploading a photo of a person and changing their style or attire. For example, you can reimagine someone as a matador or a '90s sitcom character. Because the nano banana model can maintain consistency through edits, the results should still look like the person in the original source image. This is also the case when you make multiple edits in a row. Google says that even down the line, the results should look like the original source material.
Gemini's enhanced image editing can also merge multiple images, allowing you to use them as the fodder for a new image of your choosing. Google's example below takes separate images of a woman and a dog and uses them to generate a new snapshot of the dog getting cuddles -- possibly the best use of generative AI yet. Gemini image editing can also merge things in more abstract ways and will follow your prompts to create just about anything that doesn't run afoul of the model's guard rails.
This unlocks several interesting uses for AI image editing. Google suggests uploading a photo of a person and changing their style or attire. For example, you can reimagine someone as a matador or a '90s sitcom character. Because the nano banana model can maintain consistency through edits, the results should still look like the person in the original source image. This is also the case when you make multiple edits in a row. Google says that even down the line, the results should look like the original source material.
Gemini's enhanced image editing can also merge multiple images, allowing you to use them as the fodder for a new image of your choosing. Google's example below takes separate images of a woman and a dog and uses them to generate a new snapshot of the dog getting cuddles -- possibly the best use of generative AI yet. Gemini image editing can also merge things in more abstract ways and will follow your prompts to create just about anything that doesn't run afoul of the model's guard rails.
Re: Nano banana model? (Score:2)
It is actually designed for the specific use case of generating images of lifted pickups.
Nano banana? (Score:2)
Is that a banana in your . . . never mind.
I am a banana! (Score:2)
https://www.youtube.com/watch?... [youtube.com]
Kids, this is what art looked like before CrapGPT and other AI tools.
The full version in 4K
https://www.youtube.com/watch?... [youtube.com]
Re: (Score:2)
Yup watch this classic and laugh then just click right there to watch World of Tomorrow and cry your face off. (It is very good though)
Bad choice of name? (Score:2)
Re: (Score:2)
A Nano-Banana, 10^-6 Banana would thus convert to 0.1*10^-6*10^-9, 0.1*10^-15 sieverts, 0.1 femtosieverts well played
Welp, there went Adobe (Score:3)
Sounds like this could be a powerful alternative to Photoshop... at this rate, I wonder how much longer Adobe will be around. I was never a fan of their subscription model anyway...
Re: (Score:2)
At least their Firefly model needs a level-up or they can remove it altogether.
Crap (Score:1)
Re:Crap (Score:4, Insightful)
It's going to be crap anyhow. I have yet to see a decent AI picture which does not look fake. People might wax lyrical about AI images and videos, but so far, it can only fool people with reduced mental capability, which is what AI is turning humans into, I guess. luckily, the bubble seems to be bursting now.
The actual evidence is that most humans are pretty bad at telling apart AI and not AI generated images. See https://www.astralcodexten.com/p/how-did-you-do-on-the-ai-art-turing [astralcodexten.com]. That link also includes the test set of images used so you can run through and see how you do on that set.
Limited dominance (Score:2)
Re: (Score:2)
Google improves... (Score:1)
It doesn't work. (Score:2)
Re: (Score:2)
Did you try the other versions?
vim blueberry
emacs peachpit
Re: (Score:2)
For the vim example, don't forget to install the Gemini plugin...
-- lazy.nvim
{
'kiddos/gemini.nvim',
opts = {}
}
Re: (Score:2)
It sucks (Score:2)
It creates a photo with a SIMILAR person, hairstyle etc but NOT the same person, after the 5th try it looks like it could be a cousin or something, useless.
'possibly the best use of generative AI yet' (Score:2)