I read the hype the past week and gave it a try - Mind blown!
Google has created, IMHO, a revolution in image generation. This is a real inflection point in showing the power, and potential danger, of AI based image manipulation.
It is so powerful, Adobe has added it as a model choice in Firefly, which until now, they’ve been developing on their own.
I can’t post a quick photo that will do this tool justice. I suggest you Google for YouTube videos or try it yourself.
I need to add - I’ve been having a ton of luck with modifying, not generating photos.
Upload a photo of your own and use text prompts to modify it.
I can do a lot of Photoshop-ish things much faster and easier, and other things I still have yet to master in Photoshop.
Key thing to try: Upload multiple photos and use text prompts to tell nano banana to make composite image taking things from one photo to put in the other.
(Technical: Harmonizing light, color, and focus is the challenge in compositing. With AI tools, no need to mess with color balance, color grading, depth maps, and other traditional tools to make it look like it wasn’t cut and pasted like a ransome note.)
I can agree it’s very good indeed, to the point that they’ve created their own micro product around it: https://nanobanana.ai independent of AI Studio.
I spent an ungodly amount of time yesterday trying to remove a pair of too-reflective sunglasses off of the top of my subject’s head by throwing the best that Lightroom and Photoshop have to offer at the task. The results were OK for, say, an Instagram post on a phone, but certainly not for anything else.
I threw the image into the Nano Banana prompt box and had a print-worthy result in under a minute. A pixel-peeper who was actively looking for the edit might be able to see what’s been done, but I’m betting that if I made a 8x10 print and put it in a frame on a mantlepiece, no one looking at it would be the wiser.
I want this in Photoshop yesterday.
Note: I should add that the sunglasses perched on her head was part of the subject’s overall look, and I would have been happy to leave them there, but they were simply too reflective to harmonize with the rest of the image. I wouldn’t want to use AI to radically change how someone or something looks, but it’s a gift for this kind of distraction removal.
Ugh. An update: at the moment, the largest image you can extract from Nano Banana is 1024x1024. Gemini recommends using AI to upscale the image if you need something larger.
I hope accessing the model via Photoshop and exporting a high-resolution tiff file solves the problem.
Adobe has already incorporated Nano Banana models into both Photoshop and FireFly. Beta versions, but in the pipeline for release in next version updates.
All AI-based generative image stuff, from all vendors, seems to be capped at 1K resolution.
Not sure if this is a cost thing (cloud computing and GPU’s have to be paid by somebody, sooner or later), or if the diffusion algorithms. can’t yet be scaled up to higher resolution.
The Photoshop gurus have done multiple tutorials on workarounds to slice-up the work into into smaller pieces. Currently, each piece gets 1K resolution, so with the right slicing, you can get higher resolution results.
I think someone has even created a Photoshop plug-in that can do this slicing/generative AI in an automated workflow?
Right now, one has to add a second step of using AI-based up-scaling tools (every company has different marketing names for them).
Photoshop, “of course”, has multiple ways to do that, or you can use dedicated 3rd party upscalers.
I upscaled the 747 × 1024 Nano Banana png into a 1459 × 2000 jpg using Preserve Details 2.0 in Photoshop. The results are OK, but obviously not at good as the result I’d get exporting a high quality jpg from the original raw file in either Lightroom or Photoshop.
I did try using the generative upscale functionality in Photoshop beta and the result was overly smoothed for my taste—it looked like an AI generated image.
Not yet in Photoshop, at least not directly from Adobe. There are third-party plug-ins you can install to use Nano Banana within Photoshop, but it my copy of Photoshop beta for desktop, at least, it’s not available as a model option. I tried the online version too - I still couldn’t find a way to access a model other than Firefly. If you’ve found a way to invoke it as a partner model in Photoshop beta desktop, please share how you do that!
I don’t believe nanobanana.ai is a Google property or product. This header appears at the top of the nanobanana.ai homepage:
Nanobanana.ai is an independent product and is not affiliate with Google or any of its brands
For safety reasons, I’d access the model colloquially known as nano banana directly via Gemini or Google AI Studio. I certainly wouldn’t give nanobanana.ai my credit card.
The online scuttlebutt is that it’s a scam (or at least scam-adjacent) site that wants to sell you credits. And yes, probably vacuum up whatever images it can.
OK! Adobe added Gemini Flash 2.5 (AKA Nano Banana) to v 27 of Photoshop Beta this morning as a partner model along with Flux Kontext. I expect it will roll down to Photoshop itself at some point this year.
