blackdark
AI ImageNano Bananaai 3d figuresGeminipromptscollectible figurineimage editing

Nano Banana Prompts for 3D Figures: How to Create Your AI Collectible Figurine

A gallery of Nano Banana prompts to turn a photo into a 3D collectible figure with AI: blister box, base, anime or realistic style, consistency tricks, and where to do it free.

By BlackdarkUpdated on 8 min read

You've seen the trend: someone uploads their photo and, two minutes later, there they are, turned into a perfect collectible figurine. Plastic render, on its base, inside a blister box with the name printed on it, like something you'd buy in a collectibles shop. It's not Photoshop or an expensive 3D scan. It's Nano Banana, and the trick is in the prompt.

This guide gets straight to it: what Nano Banana is, why it stands out, and a gallery of copy-paste prompts so you nail your 3D collectible figure on the first try, with variations for the box, the base, anime or realistic styles, and the touch-ups almost nobody explains.

Note

"Nano Banana" isn't an app you download: it's the nickname for Gemini's image model (Google). You use it inside the Gemini app or in Google AI Studio. When you see "Nano Banana," think "Gemini's image generator."

What Nano Banana Is

Nano Banana is the pet name that stuck to Gemini's native image generation, Google's model. It started as an internal nickname and got so popular that even Google now uses it in its messaging. Technically it's the image capability inside Gemini: in its base version it launched as Gemini 2.5 Flash Image, and since then it's had more powerful successors (the Nano Banana 2 and Nano Banana Pro lines, built on Gemini 3).

What matters isn't the version number, but what makes it different from other generators. Two things:

  • Character consistency. It can keep the same face, the same character, across several images. Where other models give you a different person on every generation, Nano Banana holds the likeness. This is exactly what makes the figure trend work: your face stays your face.
  • Conversational editing. It doesn't just generate from scratch: it edits. You upload an image and talk to it: "change the background," "add glasses," "make the box say a different name." You refine with text, like directing a designer, instead of praying the prompt comes out perfect on the first shot.

The Pro versions add two advantages that matter for figures: they render legible text inside the image (key for the name on the box) and bump up to 2K and 4K resolution, enough to print or post large.

The 3D Figure Trend, Explained

The viral format is always the same, and that's why it works so well: you take a photo of a person (you, a friend, a pet, a character) and ask the model to turn it into a photorealistic collectible figure. The result doesn't look like a drawing: it looks like the photo of a real PVC figure placed on a desk, with its base and, often, inside a blister-style box with commercial-looking packaging.

It works because it combines three things people love: personalization (it's you), collector nostalgia (the boxes, the bases, the "limited edition" vibe) and realism (it looks like a product photo, not a cheap render). And because Nano Banana keeps the facial likeness, the "that's me in miniature" effect is instant.

From here, it's all about the prompt. On to the gallery.

The workflow is always the same: upload the photo to the Gemini app or Google AI Studio, paste the prompt, then iterate with touch-ups. Copy, paste and adjust the details in brackets.

1. The 3D Collectible Figure (the base)

This is the foundational prompt. It does 80% of the work.

3D collectible figure — base
Turn the person in the attached photo into a photorealistic 1/7 scale collectible figure, PVC render style. Full body, standing on a circular black base. Matte plastic finish with hand-painted details. Place it on a wooden desk with soft natural light, like a real product photo. VERY IMPORTANT: keep the exact facial likeness from the original photo, do not stylize the face.

Two details make the difference: the scale ("1/7" tells the model it's a small figure, not a statue) and the final instruction about the face. Without that last sentence, the model tends to generate a generic face.

2. Inside the Blister Box (the packaging)

What turns a good figure into a viral figure is the box. This is where the Pro versions' text-rendering strength comes in.

Figure in a blister box with packaging
Same collectible figure of the person in the photo, but now shown inside its collector's blister box. Cardboard box with a clear plastic window at the front. At the top, the printed text large and legible: "BLACKDARK EDITION". Limited-edition style packaging design, dark colors with gold accents. The box sits on a desk, product photo. Keep the exact facial likeness.

Swap "BLACKDARK EDITION" for whatever name you want. If the text comes out blurry, that's a sign you're on the base model; explicitly ask for "sharp, legible text" or use a Pro version.

3. Base and Pose Variation

The base completely changes the figure's character. Try different materials and shapes.

Figure with premium base and dynamic pose
1/6 scale collectible figure of the person in the photo, in a dynamic action pose, cape or jacket caught in the wind. Black marble base with a gold metal plaque engraved with the text "BLACKDARK". Cinematic studio lighting, blurred dark background. Premium PVC finish with reflections. Keep the exact facial likeness from the photo.

4. Anime / Garage Kit Style

If you want to step away from realism and into Japanese anime figure territory, say so clearly.

Anime / garage kit style figure
Turn the person in the photo into a Japanese garage-kit anime figure, 1/8 scale. Slightly stylized proportions but keeping the recognizable facial features from the photo. Vivid colors, glossy finish, detailed eyes. Clear acrylic base with a glass plate. Product photo on a light neutral background. Keep the likeness to the original person.

5. Realistic / Luxury Collectible Style

And the opposite extreme: a hyperrealistic figure, almost a sculpture.

Realistic / luxury sculpture figure
Hyperrealistic collectible figure of the person in the photo, 1/4 scale, premium resin sculpture finish with realistic skin and fabric texture. Standing on a marble base with museum lighting. Sober, elegant aesthetic, high-end product photography, neutral gray background. Precisely preserve the facial likeness from the original photo.

6. Touch-ups: Iterating Is Where You Win

This is Nano Banana's real edge. You don't toss the image and start over: you edit on top of the result by talking. These are the touch-ups you'll use most.

Touch-ups on the already-generated figure
On the previous image, make these changes: 1) the face isn't quite right, adjust it to match the original photo better; 2) change the background to a collector's shelf with other blurred boxes; 3) make the box text sharper and centered; 4) add a small price tag in the corner that says "LIMITED EDITION". Keep everything else the same.

Tip

Ask for changes one at a time or in small batches. If you drop ten corrections at once, the model prioritizes some and forgets others. And always close with "keep everything else the same" so it doesn't redo the whole figure.

Consistency Tricks (so it lands on the first try)

The difference between a figure that looks like you and one that looks like a distant cousin is in these details:

  1. A front-facing, well-lit starting photo. No backlit shots, extreme profiles or pixelated images. The model can only keep what it sees clearly.
  2. Repeat the face instruction. "Keep the exact facial likeness, do not stylize the face" isn't optional: it's the single highest-impact sentence in the whole prompt.
  3. Be specific with scale and material. "1/7 scale, matte PVC render" gives the model a clear mental frame. "Make me a figure" alone leaves it too much freedom.
  4. One person per figure. If you upload a photo with several people, the model gets confused. Crop to whoever you want to convert.
  5. Iterate, don't restart. If something almost works, fix it with an edit instruction instead of changing the whole prompt and rolling the dice again. That's Nano Banana's superpower.
  6. For box text, go Pro. The base model sometimes spits out misspelled or blurry text. If the packaging matters, the Pro versions render legible text far better.

Limits (what the viral tutorial won't tell you)

It's not all perfect, and it's worth knowing before you get frustrated:

  • Text fails on the base model. Made-up letters, half-finished words. It's the most typical limitation. Fix: iterate, ask for "legible text," or go Pro.
  • Hands and small accessories. As with almost all image AI, fingers and tiny objects can come out weird. Always check and fix by iterating.
  • Likeness isn't cloning. It keeps features, but it's not a scan. For very distinctive faces it may land at 90%. Accept it or iterate.
  • Free quotas. Free access exists, but it has a daily limit, especially on the Pro versions. If you're doing a long session, ration your Pro attempts and prototype on the base model.
  • Watermark / provenance. Generated images usually carry provenance marks (visible or invisible like SynthID). Don't present them as real photos of a product you're selling.

Pros

  • Real character consistency: your face stays the same across images.
  • Conversational editing: you refine by talking, no starting over each time.
  • Free access in the Gemini app and in Google AI Studio.
  • The Pro versions render legible text and bump up to 2K/4K.
  • Zero learning curve: upload a photo, paste a prompt, done.

Cons

  • The base model fails with box text (blurry or misspelled).
  • Free access has a daily quota, tighter on the Pro versions.
  • Hands and tiny objects can come out deformed and need fixing.
  • The likeness is very good, but it's not an exact 3D scan.
  • Images carry provenance marks: don't pass them off as real photos.

Who Is This For?

You'll be interested if: you want to jump on the trend and get your figurine for social with no payment; you're a creator who needs quick merch mockups or packaging; you run a brand or comic and want a consistent character across many images; or you simply want an original personalized gift without touching Blender or a 3D printer.

You won't be interested if: you need an actual 3D model to print or animate —this generates 2D images that look like figures, not 3D files—; or if you're after a forensic, millimeter-perfect facial match, because the AI keeps features but doesn't clone.

The honest takeaway: Nano Banana isn't magic, it's structure. The figure trend looks complicated and is one of the easiest to clone once you get the formula —photo + scale + material + base + box + facial likeness— and accept that the first shot doesn't always nail it and that iterating is part of the game, not a failure. Copy the prompts above, tweak the brackets, and your collectible figure is out in five minutes.

FAQ

Nano Banana is the nickname for Gemini's native image generation, Google's model. It's not a separate app: it's the ability to create and edit images inside Gemini. It stands out for two things: keeping the same character across images (consistency) and editing by talking, meaning you upload a photo and request changes in plain language instead of starting from scratch.

You upload a photo of yourself (or anyone) to the Gemini app or Google AI Studio and give it a prompt describing a collectible figure: PVC render style, a scale (1/7), a base, a blister box with its packaging and, crucially, the instruction to keep the facial likeness. Then you iterate: ask to change the background, the pose or the text on the box until it lands.

Yes, there's free access. In the Gemini app you can generate images without paying, and in Google AI Studio the base model has a generous daily quota for creating several figures a day. The higher versions (Nano Banana Pro, with better text rendering and 2K/4K resolution) give fewer free passes and, once you exhaust them, you usually drop back to the base model; Google AI subscriptions raise those quotas.

It's almost always the prompt. If you don't explicitly tell it to keep the features, the model 'stylizes' the face toward a generic figure. Add a clear instruction to keep the facial likeness, upload a front-facing, well-lit photo, and if it still drifts, fix it by iterating: 'the face doesn't match, adjust it to match the original photo.'

Beyond a figurine of your face, the same workflow works for product mockups, merch packaging, consistent characters for a brand or comic, personalized gifts and social content. The powerful thing about Nano Banana isn't the figure itself, but being able to keep the same character across many images by editing with text.

Keep digging into the same topic.

Share
Newsletter

Get the next guides in your inbox

AI and marketing ideas and resources, no filler. What works and how to apply it.

Ideas and resources, no spam. Unsubscribe anytime.

Finding this guide useful?

Subscribe