If you've been following the wild ride that is AI image generation over the last year or so, you'll already know the name Nano Banana. Yes, it really is called that. And yes, it really is as impressive as it sounds ridiculous. Google's Gemini-powered image model made quite the splash when it first landed in Adobe Firefly, and now the next generation — Nano Banana 2, running on Gemini 3.1 — is here and available right inside Firefly today.
Let's take a look at what's new, what it can actually do, and why I think this matters for anyone working creatively with Adobe tools.
Wait — What Even Is Nano Banana?
Good question. Nano Banana is the slightly bonkers internal codename for Google's Gemini image generation model. When Adobe integrated partner AI models into Firefly back in 2025 bringing in the likes of OpenAI, FLUX, Runway and others alongside its own Firefly models Google's offering got labelled Nano Banana in the Firefly interface, and the name stuck. The first version, based on Gemini 2.5 Flash, became hugely popular with creators almost immediately.
Adobe reported that users created 5 billion images with the original Nano Banana in just two months. That's not a typo. 5 billion. So it's fair to say there was quite a bit of anticipation for what version 2 would bring to the table.
So What's New in Nano Banana 2?
Built on Gemini 3.1, Nano Banana 2 is a step up in pretty much every area. Here's what stands out:
Crystal clear text in images. If you've ever tried to get an AI to generate a poster and ended up with complete gibberish where the words should be, you'll understand why this is a breakthrough. Nano Banana 2 can now produce sharp, well-integrated text directly inside generated images — useful for mockups, posters, product shots, you name it.
Targeted edits with simple prompts. You can now use conversational text prompts to fix specific parts of an image without touching the rest of it. Got a great shot but the lighting on one side is off? Just describe the change and let Nano Banana 2 sort it. This alone changes how you'd use the tool day to day.
Better reflections, textures and complex lighting. Glass, water, depth of field — these are areas where the first Nano Banana struggled a bit. Version 2 handles them noticeably better.
Factually grounded image generation. By tapping into Google Search's knowledge base, Nano Banana 2 can generate accurate maps, diagrams, and real-world visuals — not just artistic interpretations.
Text localisation. It can now translate text within images — genuinely useful if you're producing assets for international audiences and need a French or German version of an advert without recreating the whole thing from scratch.
Where Can You Use It in Firefly?
Nano Banana 2 is available across several areas of the Firefly platform. In the main Firefly app, you'll find it in the Text to Image module — just select Gemini 3.1 (with Nano Banana 2) from the model picker and you're away. It's also available in Firefly Boards, Adobe's AI-powered collaborative moodboarding space, where creative teams can brainstorm and develop campaign concepts together.
In Photoshop, Nano Banana 2 powers Generative Fill, giving you a fast, prompt-based way to make targeted edits directly in your existing workflow. The key advantage here is that you get Google's model capabilities without ever leaving the Adobe environment — no juggling between tabs, no separate subscriptions.
You can also upload up to six reference images in Firefly and prompt Nano Banana 2 to merge and refine elements from them — handy for building consistent visual identities or campaign assets.
Nano Banana 2 vs Adobe's Own Firefly Models — Which Do You Use?
This is worth addressing, because it comes up a lot. Adobe's own Firefly Image Model 5 is commercially safe — trained on licensed content, meaning you can use the output in client work without worrying about IP issues. That's a big deal in a professional context.
Nano Banana 2, on the other hand, excels at text in images, advanced editing, character consistency across multiple images, and factual accuracy. It's not a replacement for Firefly's own models — it's a complement to them. Adobe's whole strategy here is about giving you the right tool for the right task, all in one place. And that philosophy is starting to really make sense in practice.
Think of it this way: use Firefly Image Model 5 for commercially safe production assets, and reach for Nano Banana 2 when you need stronger results on portraits, complex edits, or text-heavy visuals. Test them both on the same project and see which fits.
How to Get Started
Getting going with Nano Banana 2 in Firefly is straightforward:
Head to firefly.adobe.com and open the Generate Image module.
Click the Model dropdown and select Gemini 3.1 (with Nano Banana 2).
Upload an existing image or write a prompt to generate a starting point.
Use the prompt bar to describe your edits conversationally and refine from there.
For Photoshop users, Generative Fill is the place to find it — make a selection, open the Firefly model picker, and choose Nano Banana 2 from the partner models list.
Final Thoughts
The arrival of Nano Banana 2 in Adobe Firefly is another strong step in Adobe's strategy of bringing the best available AI models into a single creative platform. Rather than forcing creators to choose between the Adobe ecosystem and the latest generation of AI models, you now get both — and the integration is genuinely seamless.
For anyone already working in Firefly or Photoshop, this is a meaningful upgrade. The improvements to text in images and targeted edits alone make it worth exploring, and the fact it sits alongside Adobe's own commercially safe models gives you real creative flexibility without the workflow headaches.
Head into Firefly today and give it a go. And yes — you are allowed to chuckle at the name.
Let me know your thoughts in the comments below!
No comments:
Post a Comment