NANO BANANA (AKA GEMINI 2.5 FLASH IMAGE) JUST KILLED PHOTOSHOP
Why graphic designers should care about Google’s new image editor, and how I made a client-ready wine mockup in five minutes.
TL;DR (for the skimmers)
Nano Banana = Gemini 2.5 Flash Image, a new Google model that does photoreal image editing and generation, now inside the Gemini app and via API.
It supports targeted, prompt-based local edits, multi-image fusion, multi-turn editing, and design mixing (apply the look of one image to objects in another).
Availability: Gemini app (free + paid) for end users; developer access via Google AI Studio & Vertex AI (public preview for the image model).
Cost (API): ~$0.039 per generated/edited image in preview.
Watermarks: Visible label in the app + invisible SynthID watermark on all AI images/edits.
Hot take: For speed-run product mockups, Nano Banana beats a manual Photoshop workflow. For heavy, layered control, Photoshop still rules.
Wait—what exactly is
“Nano Banana”?
“Nano Banana” started as the playful codename for Gemini 2.5 Flash Image, Google’s newest image generation and editing model. It’s now integrated into the Gemini app (web + mobile), so regular users can try it, while developers get access through Google AI Studio and Vertex AI (image model in preview).
Under the hood, this update focuses on precise, local edits with natural language (think “make the table blue, keep the wood grain”), multi-turn edit chains (edit → re-edit → re-re-edit without nuking prior work), and multi-image fusion (blend your product into new scenes). It also emphasizes character/identity consistency across edits—handy for keeping people looking like themselves.
Bonus chatter: Google’s own post points to Nano Banana topping the LMArena leaderboard for image editing; early reviews celebrate the consistency wins, while also flagging some missing “basic” controls in the app UI (like precise cropping).
Case Study: 5-minute wine bottle mockup (real demo)
The brief: Re-tint the set for brand alignment and apply a supplied label—fast.
The moves: three plain-English commands.
The result: client-ready mockup in ~5 minutes.
Edit 1 — Table → Blue
Prompt: “Change table surface color to blue; keep wood grain and natural shadows.”
Why it matters: You get a recolor that respects texture and lighting—no tedious mask-chasing.
Edit 2 — Chair → Match
Prompt: “Change the chair to the same blue as the table; preserve fabric texture and edge lighting.”
Why it matters: Brand-true color cohesion across materials, in one pass.
Edit 3 — Label → Wrap
Prompt: “Apply wine_label.png to the bottle; wrap to glass curvature; keep specular gloss; add subtle label edge shadow.”
Why it matters: You get curvature + gloss interplay that sells “real photograph,” not “flat sticker.”
Time vs. Photoshop (typical)
Nano Banana: 3 prompts → done.
Photoshop (manual): contour mask, displacement map, perspective/warp, targeted color replace, AO/shadow pass, cleanup.
If you’re billing for speed or pitching comps, that delta is the difference between “I’ll send options today” and “Give me an hour.” (Nano Banana is built for fast local edits and multi-turn changes; PS still wins for layered, pixel-level control.)
Where Nano Banana shines for designers
Product mockups in context: Drop packaging into sets, recolor props, and keep textures believable—fast. (Prompt-based local edits + multi-turn.)
Catalog & brand consistency: Style-lock a look and propagate variants without drifting identity or materials. (Character/object consistency.)
Composite sanity: Blend multiple shots—product + environment—without starting a layer circus. (Multi-image fusion.)
Style transfer for materials: Use “design mixing” to map a palette/texture to a different object (e.g., bottle cap finish to box panel).
WHERE PHOTOSHOP STILL WINS
Layered, non-destructive pipelines with smart objects, adjustment layers, blend-modes wizardry.
Production-grade prepress (CMYK, spot colors, true layer exports).
Edge-case controls & exactness: Some reviewers note the Gemini app UI still lacks basics like precise cropping/aspect. (You can always round-trip to PS.)
Access, pricing, and watermarks (what to expect)
Gemini app (consumer): Nano Banana’s editing is live for both free and paid users; edits include visible and invisible (SynthID) watermarks.
Developers & teams: Use Google AI Studio or Vertex AI. The image model is in public preview with image outputs and costs about $0.039 per image right now.
Why watermarks matter: They help audiences and clients identify AI involvement—ethical + pragmatic. (SynthID is applied automatically, and the app adds a visible label.)
Swipe-worthy prompt recipes (copy/paste)
Re-tint with texture intact:
“Change [object] to [brand color]; keep original texture and shadows; match white balance to the scene.”
Material-aware label wrap:
“Apply [label.png] to bottle; wrap to glass curvature; preserve specular highlights; add 5–10% edge shadow; micro-noise 2% to blend.”
Scene match after composite:
“Match exposure and color temp to plate; add contact shadow (soft); AO around base; keep reflections under 30%.
(These leverage the model’s targeted editing + multi-turn chaining; add your brand palette values for consistency.)
Hot takes, responsibly served
Yes, we’re cheeky when we say “killed Photoshop.” The nuance: Nano Banana kills the time it takes to get a convincing mockup, especially when you’re iterating live with clients. For deep, layered post and print workflows, Photoshop remains the endgame. Balance the two and you’ll shave hours off comping without sacrificing craft.
FAQ (designer edition)
Is Nano Banana free?
In the Gemini app, the updated editor is available to both free and paid users globally. Dev/enterprise access runs through the Gemini API and Vertex AI (image model in preview), with usage-based pricing.
Can I export layered PSDs?
Nano Banana outputs final images (raster). If you need layered deliverables, finish in Photoshop after generating/editing the base. (The image model’s outputs are images; no layered PSD.)
Does it do local, multi-step edits without wrecking the image?
Yes—prompt-based local edits and multi-turn editing are core to this release.
Any caveats?
Reviewers note the app UI currently lacks some basic controls (e.g., precise cropping/ratio). Expect this to evolve; you can always round-trip to PS.
Is it safe to use for client work?
The app adds a visible AI label; all outputs include SynthID watermarks. Keep consent + disclosure in your workflow, especially with people images.
Ready to build brand-safe, client-ready comps at banana speed?
👉 Enroll in the AI Branding Masterclass. We’ll cover Nano Banana workflows, prompt recipes, and PS round-tripping—end-to-end.