open instagram or tiktok right now and count how many tiny versions of real people you see standing on desks. little 3D figurines, like collectible toys — same face, same outfit, standing on a little clear base next to a toy box with their name on it. politicians posted them. your cousin posted one. there's a good chance your gym trainer posted one.
it's called the nano banana 3D figurine trend, and it's one of those rare AI trends that actually looks good in the feed. the whole thing takes one photo and one prompt, it's free, and it takes about two minutes. here's exactly how to make yours — plus the one step everyone skips that separates the crisp ones from the cursed ones.
what even is "nano banana"?
quick explainer so you don't feel lost. Nano Banana is the internet's nickname for Google's Gemini 2.5 Flash Image — the image model inside the Gemini app and Google AI Studio. "Banana" was apparently an internal codename that stuck, and honestly it's a way better name than the official one.
the trend works because the model is scary good at keeping your face recognizable while rebuilding everything around it as a little collectible figurine: realistic 1/7-scale toy, glossy plastic finish, round transparent acrylic base, sitting on a wooden desk, usually with a toy-box next to it showing the "packaging". one photo in, mini-you out. no design skills, no payments, no waiting in a queue.
what you need (all free)
- the Gemini app on your phone (Android/iOS), or a browser open to Google AI Studio
- one clear photo of yourself — front-facing, decent light, face visible. a normal selfie works fine
- the prompt below (copy-paste, zero writing required)
- a photo editor on your phone for the cleanup step — PicFix Pro handles it free, more on that in a minute
how to make your nano banana figurine (the actual steps)
step 1: pick a good photo. this matters more than the prompt. bright, front-facing, no sunglasses, no motion blur. the AI builds the figurine's face from this — garbage photo in, melted face out.
step 2: open gemini and attach the photo. in the app, tap the image button and pick your photo. in AI Studio, do the same in the image chat.
step 3: paste the prompt. use the classic one below (or any of the variations). hit send. wait 20-30 seconds.
step 4: regenerate if it's cursed. first tries come out weird maybe 40% of the time — hands merged with the desk, face slightly off, toy box text that says gibberish. just run it again with the same prompt. most people land a good one by try 2 or 3.
step 5: clean it up on your phone. this is the step that makes yours look better than everyone else's. zoom in and check for AI glitches — warped fingers, a smudged ear, weird background junk, text on the toy box that almost spells your name. open PicFix Pro and erase them: the object eraser wipes stray artifacts in a couple of taps, enhance sharpens a soft-looking face, and a quick brightness/contrast pass fixes the flat lighting AI renders usually have. two minutes, huge difference.
step 6: post it. 4:5 portrait for the instagram feed, 9:16 if it's going to stories or reels. hashtags that are actually getting used: #NanoBanana #3DFigurineTrend #AIFigurine #NanoBananaTrend.
4 copy-paste nano banana prompts (steal these)
i wrote these to cover the main flavors people are posting. swap in your own details where it says so.
1. the classic desk figurine
Create a realistic 1/7 scale figurine of the person in this photo, in a realistic style, in a real environment. The figurine stands on a round transparent acrylic base on a wooden computer desk. Behind it, a computer monitor shows the 3D sculpting process of this figurine. Next to the monitor is a toy packaging box in BANDAI style with vibrant 2D illustrations matching the figurine. Natural window light, soft shadows, photorealistic product photography.
2. the toy-box action figure
Turn the person in this photo into a collectible action figure inside a toy box with a clear plastic window. Bold retro graphics on the packaging, the person's first name printed on the box, blister-card style, hanging on a store peg hook. Photorealistic, studio lighting, premium collectible look.
3. chibi mini-me
Turn the person in this photo into an adorable chibi-style 3D figurine with a slightly oversized head, big glossy eyes, soft vinyl toy finish. Standing on a pastel round base, tiny sparkles around it, cute collectible aesthetic, clean studio background, photorealistic render.
4. your pet as a figurine
Turn the animal in this photo into a realistic collectible 3D figurine with a glossy finish, standing on a clear acrylic base on a wooden desk. Next to it, a toy packaging box with cute 2D illustrations of the animal. Soft natural light, photorealistic product photography.
why half the figurines on your feed look weird (and how yours won't)
scroll through the hashtag and you'll notice a pattern. the viral ones look clean. the flops all have the same problems: hands that melt into the desk, a face that's almost you but not quite, toy-box text that's pure gibberish, or a faint smudge where the AI didn't know what to draw.
none of that needs a reshoot. it's all fixable on your phone after the fact:
- warped hands or merged objects — brush over them with an object eraser and they're gone
- soft or blurry face — run an enhance pass to sharpen the details back up
- flat, lifeless lighting — nudge brightness and contrast until the figurine pops off the desk
- gibberish text on the toy box — erase it, or crop the frame tighter so the box is a prop, not the focus
PicFix Pro does all four of these free on Android — no subscription, no watermark on your exports. it's the same cleanup workflow people have been using for the ghostface trend and the younger-self trend: generate with AI, fix the glitches on your phone, then post.
5 mistakes that ruin nano banana posts
after watching a few hundred of these go by, the fails are almost always the same five things. avoid them and you're already in the top 10% of the hashtag.
1. uploading a group photo and hoping the AI picks you. it won't pick right. it blends faces, or builds a two-headed figurine. crop to one person first.
2. sunglasses or heavy filters on the input. the figurine inherits whatever's on your face. take the glasses off, use the unfiltered original.
3. posting the first generation. the first try is a draft. generate 3, pick the best face, that's your post.
4. ignoring the background. AI loves to leave a weird smudge on the desk or a half-rendered monitor behind the figurine. one pass with an object eraser and it's gone.
5. tiny text on the packaging. the model cannot spell your name reliably at small sizes. either accept the gibberish (crop it out), or erase the text area and leave the box clean. PicFix Pro's eraser is honestly perfect for this — one brush stroke over the fake text.
why this trend blew up again this week
the figurine thing has been around for a bit, but it caught a second wave this week — mainstream press picked it up, politicians and celebrity fan pages started posting their own, and suddenly everyone's feed is figurines again. that's the pattern with AI photo trends: they don't die, they hibernate and come back bigger. the 80s glow-up did the same thing in september, and pumpkin head comes back every single october.
the practical takeaway: trends like this are a cheat code for reach. the format is proven, the prompt is free, and the only thing standing between you and a post that performs is that two-minute cleanup pass. most people skip it. don't be most people.
quick posting tips
- crop 4:5 for instagram feed — the figurine reads bigger and the desk scene stays intact
- post in the evening in your timezone; AI trend content tends to peak 7–10pm when people are scrolling
- if you're doing a carousel, put the original selfie first and the figurine second — the before/after swipe gets way more engagement
- don't over-filter it. the charm is that it looks like a real product photo. one cleanup pass, done
- The Gemini Ghostface AI Photo Trend Is All Over TikTok — Here's How to Do It Without the Creepy AI Glitches
- The Pumpkin Head Halloween Photo Trend Is Back: How to Nail the Shot With a Free Photo Editor
- "Me and My Younger Self" AI Photo Trend: The Prompt, the Steps, and How to Fix It on Your Phone
- Google's SynthID Detector: How to Check If a Photo Is AI-Generated (and What It Can't Catch)
- The Viral 80s AI Photo Trend: How to Get the Retro Look Without Risking Your Photos
that's the whole trend. one selfie, one prompt, two minutes of cleanup — and you've got a post that looks like a $200 collectible photoshoot. if you want the cleanup step handled properly, grab PicFix Pro free on Google Play and your figurine will be feed-ready in about two minutes.
related reading
Frequently Asked Questions
What is the Nano Banana 3D figurine trend?
It's a viral AI photo trend where people turn a selfie into a realistic 3D collectible figurine using Google's Gemini 2.5 Flash Image model (nicknamed 'Nano Banana'). The AI renders you as a 1/7-scale toy on an acrylic base, usually on a desk next to a toy-box. One photo and one prompt, free.
Is the Nano Banana figurine trend free to make?
Yes. The Gemini app and Google AI Studio let you generate the figurine image for free with no subscription. Fixing AI glitches afterward is also free — apps like PicFix Pro clean up warped hands, smudged faces, and gibberish packaging text on your phone at no cost.
Why is it called Nano Banana?
'Banana' was an internal codename for Google's Gemini image model that stuck as a nickname online. 'Nano' comes from Google's lightweight Gemini Nano model family. The full official name is Gemini 2.5 Flash Image.
What is the best prompt for the Nano Banana figurine trend?
The classic prompt: 'Create a realistic 1/7 scale figurine of the person in this photo, in a realistic style, in a real environment. The figurine stands on a round transparent acrylic base on a wooden computer desk.' Add a toy-box line for the packaging look. If the first result looks off, regenerate — most people get a good one by try 2 or 3.
Why does my AI figurine look weird or melted?
Common causes: a blurry or dark input photo, hands in complex poses, and the model struggling with tiny text on the toy box. Use a bright front-facing selfie, regenerate 2-3 times, then fix leftover glitches (warped fingers, smudged ears, gibberish text) with a free photo editor like PicFix Pro's object eraser and enhance tools.
Is there a free photo editor to fix AI figurine glitches?
Yes — PicFix Pro is free on Google Play and covers the exact cleanup this trend needs: object eraser for warped bits and background junk, enhance for soft faces, and brightness/contrast sliders for flat AI lighting. No watermark on exports.