How I turn a selfie into a viral 3D AI figurine.
TL;DR: I used Google’s Gemini 2.5 Flash Image (nicknamed Nano Banana) to turn a photo into a hyper-real 3D AI figurines, tuned the prompt for toy/packaging details, checked the SynthID watermark rules, and exported print-ready references for 3D printing or mockups. The whole viral trend — people sharing “3D AI figurines” made from their photos — exploded across social platforms this week.
Table of Contents
My step-by-step guide — how I made a 3D AI figurine photo (exact steps I follow)
Prep (what I gather before I start)
Image Credits: Saurabh Chauhan/YT
- Pick a clean, well-lit photo of the subject — front three-quarter works best for figurines. (High res, no severe motion blur.)
- Decide the style: vinyl toy, wooden carving, chibi, action figure, or realistic resin.
- Gather reference images (toy base, packaging, lighting) if I want consistent results.
Step 1 — Open Gemini 2.5 Flash Image (or the Gemini app)
Image Credits: Saurabh Chauhan/YT
I upload my photo into Gemini’s image edit tool (the image-to-image mode). Gemini 2.5 Flash Image supports fusion of reference images if I want to combine multiple references.
Step 2 — Use a focused prompt formula
I start with a simple formula then iterate:
Image Credits: Saurabh Chauhan/YT
Prompt formula: “Create a photorealistic collectible figurine of the subject from the uploaded photo. Style: [vinyl / resin / wooden toy]. Pose: [standing/arms down/hero pose]. Base: clear acrylic round base with small label. Lighting: soft studio key + rim light. Background: white product studio with subtle shadow. Include packaging box mockup in the second image.”
Examples I used:
- “Photorealistic vinyl toy figurine of subject, 1:12 scale, detailed face, soft plastic texture, studio product shot on white, acrylic base, 45° three-quarter view, high detail, shallow depth of field.”
- “Action-figure style, removable accessories, articulated joints, packaging with logo, brand photography, 4k, ultra-realistic.”
Tips: start generic, then add or remove texture words (glossy, matte, carved grain). Iteration is key.
Step 3 — Use reference images to lock likeness
If the app accepts multiple inputs, I drop a toy texture reference (e.g., a vinyl figure shot) and a packaging mockup. Gemini 2.5 Flash Image can fuse multiple inputs to keep likeness and style consistent.
Step 4 — Tweak lighting, scale, and props
I prompt for camera type, lens (e.g., “50mm, f/2.8”), and lighting specifics. That gives me consistent depth of field and studio look. I adjust prompts until the face reads like the original person but on a toy scale.
Step 5 — Generate packaging and hero shots
Image Credits: Saurabh Chauhan/YT
I ask Gemini to output both (A) close-up figurine on base and (B) packaged box mockup for social posts. Two images make a better social carousel.
Step 6 — Check synth labeling & export
Before sharing, I remember Gemini images include an invisible SynthID watermark per Google’s policy — don’t try to remove or spoof it. Export the highest available resolution for social or for a 3D-printer reference sheet.
Also Read: Studio Ghibli Movies: Trending Merch, Art & Gifts
Step 7 — Share with the right caption & hashtag
I post the carousel with a short process caption, mention “created with Gemini / Nano Banana,” and include trend hashtags. That’s how posts ride the existing virality.
The internet’s favourite Coke paglus.#NanoBanana #WithGoogleGemini pic.twitter.com/77Ng2azogv
— Google India (@GoogleIndia) September 11, 2025
Prompts from the video (Credits: @AsapGuide/YT)
ORIGINAL:
Create a 1/7 scale commercialized figurine of the characters in the picture, in a realistic style, in a real environment. The figurine is placed on a computer desk. The figurine has a round transparent acrylic base, with no text on the base. The content on the computer screen is a 3D modeling process of this figurine. Next to the computer screen is a toy packaging box, designed in a style reminiscent of high-quality collectible figures, printed with original artwork. The packaging features two-dimensional flat illustrations.
ARTISAN:
Create a 1/7 scale commercialized figurine of the characters in the picture, in a realistic style, in a cozy artisan workshop environment. The figurine is placed on a wooden craft desk with warm ambient lighting. The figurine has a round transparent acrylic base, with no text on the base. The content on the computer screen is a detailed sculpting process with material nodes visible. Next to the computer screen is a handcrafted toy packaging box, designed in an artisanal boutique style, printed with hand-drawn artwork. The packaging features vintage two-dimensional flat illustrations.
RETRO:
Create a 1/7 scale commercialized figurine of the characters in the picture, in a realistic style, in a vintage collector’s den environment. The figurine is placed on a retro wooden desk with warm incandescent lighting. The figurine has a round transparent acrylic base, with no text on the base. The content on the computer screen is a clay sculpting process reference sheet. Next to the computer screen is a nostalgic toy packaging box, designed in a classic 80s collectible style, printed with retro-themed artwork. The packaging features vintage two-dimensional flat illustrations.
MODULAR:
Create a [SCALE] scale [STYLE] figurine of the character, displayed in a [ENVIRONMENT]. The figurine has a [BASE_TYPE] base [BASE_DETAILS]. The [SCREEN_DEVICE] shows [SCREEN_CONTENT]. Nearby is [PACKAGING_STYLE] featuring [PACKAGE_DETAILS]. [LIGHTING_MOOD].
Prompt cheatsheet (copy-paste and tweak)
- Base prompt (vinyl toy):
Create a photorealistic vinyl collectible figurine of the uploaded subject. 1:12 scale, soft plastic texture, highly detailed face that retains the subject likeness, standing on a clear acrylic base labeled with small nameplate, studio white background, rim and key light, shallow DOF, product photography, 4k. - Action figure prompt:
Make an action figure version of the subject with articulated joints, removable accessories, heroic pose, matte resin texture, packaging box included with window and logo, product shot. - Packaging mockup prompt:
Create a boxed packaging mockup for the figurine: cardboard box with clear window, logo on top, barcode and sticker, placed on a reflective surface, 35mm product shot.
Quick editing & design tips I use
- If details look too “smiley” or cartoony, add “realistic facial proportions” to the prompt.
- Swap texture words to change material: vinyl, resin, wooden carved, metallic.
- For consistent sets, reuse the same lighting + base prompt across multiple images.
- Use multiple reference images to minimize likeness drift.
Short legal / ethics note (what I always say)
- Respect likeness rights: don’t create or monetize images of other people (especially public figures) without permission.
- Don’t try to remove SynthID or misrepresent AI-made content. Google’s tools embed provenance for a reason.
If the article was helpful, consider sharing it with your tech-geek friends. Join my FREE Telegram for faster notifications to stay updated with the latest tech news. Thank you for reading this further.
FAQ – Frequently Asked Questions
What is the Nano Banana trend and how does it work?
Nano Banana is the nickname for Google’s viral image editing feature inside Gemini 2.5 Flash Image that lets users upload a photo and reimagine it as a photorealistic collectible figurine (toy-style images). The tool fuses uploaded images and text prompts to produce detailed outputs.
Is Nano Banana free to use?
Many users have access to the tool for free through the Gemini app or Google AI Studio, though availability and quotas depend on region and Google’s current rollout. Always check the app for current access details.
Can I 3D print a figurine directly from the generated image?
Not directly. Gemini outputs 2D images — you need a 3D modelling/retopology step (or a 3D artist) to produce a printable mesh from the AI image references. Treat the images as design references.
Does Gemini mark AI-generated images?
Yes — images created or edited with Gemini 2.5 Flash Image include an invisible SynthID watermark that identifies them as AI-generated per Google’s model policies.
Best prompts to get the figurine look?
Use texture words (vinyl, resin), camera terms (50mm, f/2.8, studio lighting), and product words (acrylic base, packaging mockup, white studio background). Start generic, iterate, and add reference images for consistent likeness.
