Nano Banana Figurine Prompt: Copy, Customize, and Fix Your Image

A Nano Banana figurine prompt turns a reference photo into an image of a small collectible figure. Start with a clearly visible subject and a simple display scene; add packaging or a dramatic pose after the figure looks right. You can use the result as a concept image, profile picture, or gift illustration.

The prompts below are ready to copy into Gemini after attaching your photo. They are suggested starting points, not tested guarantees of likeness. The output is a flat image with a 3D appearance. If you need an actual file for Blender or a printer, see the Gemini image-to-3D guide for the separate workflows and their limits.

Copy this photo-to-figurine prompt

Use this with a photo of one person. It reads the appearance from the attached image, so there are no blanks to fill in.

Create an image of a painted resin collectible figurine based on the person in my attached photo. Use the photo as the reference for the face shape, hairstyle, expression, clothing colors, and distinctive accessories.

Show a single full-body figure with gently stylized proportions: a slightly larger head, a compact body, sculpted hair, and clothing that looks molded and hand-painted. Keep the person recognizable rather than turning them into a different character.

Give the figure a relaxed standing pose, with both hands resting at its sides and both feet supported by a small ivory oval display stand. If the photo does not show the lower body, use plain trousers and simple shoes that coordinate with the visible outfit.

Photograph the collectible from a three-quarter front view against a pale gray studio backdrop. Include the entire figure and stand, with a little space around them. Use soft light from the upper left, subtle reflections on the painted resin, and a contact shadow beneath the stand. Keep the face in sharp focus.

The scene contains only the collectible and its stand. Leave the background and stand blank, without writing or packaging.

The last sentence keeps the first attempt focused on the figure. You can introduce a box later without asking the model to solve face resemblance, toy material, composition, and package lettering at once. The suggested lower-body details are invented styling; use a full-body reference if those details must match the person.

Google's image prompting guide recommends describing the subject, medium, setting, lighting, and camera view, then refining the result conversationally. Those are useful directions for a figurine image, but they do not establish that any particular wording will always work.

Make and edit the image in Gemini

For a photo transformation, use a Gemini account with image editing available. Google's Gemini image help requires users with personal accounts to be 18 or older for image editing; availability also depends on supported countries and languages. Work and school accounts can have different restrictions. Choose a photo you own or have permission to transform.

1. Open Gemini. Google's desktop instructions put image generation under Open Sidebar → Images. You can also upload a photo in a chat and ask Gemini to edit it.
2. Attach the reference photo, paste the starter prompt, and submit it. For the first attempt, use one clear subject rather than a collage or group photo.
3. Inspect the generated figure before requesting accessories. Compare its face, hair, clothing, body outline, and stand with what you intended.
4. Continue in the same chat with a specific repair prompt from the section below. For an image you saved earlier, Google's documented route is Open Sidebar → Library → select the image → Chat; you can also upload the image you want to edit.
5. When you have a result you want to keep, hover over it and choose Download full size. Inspect that downloaded file too, especially the face, hands, and any lettering.

Which Nano Banana model? As of October 9, 2026, Google's image-generation overview lists Nano Banana 2 for free access and Nano Banana 2.1 with a Google AI plan. Its operational help still describes Nano Banana 2 and Lite for initial creation, plus More → Redo with Pro for paid subscribers. The names you see can differ during rollout; use the image controls actually available in your account. A plan or model name is not a guarantee that a particular face or pose will render correctly.

If Redo with Pro is available, it is an option for an image that needs additional detail. Google also says reaching the daily Nano Banana 2 quota prevents further Pro redos. Check your account's displayed allowance rather than assuming a fixed number of free attempts. You do not need to set up an API to use this photo-to-figurine workflow.

Choose a reference that shows what must stay recognizable

A sharp portrait is useful for a face, but it cannot tell the model what unseen shoes or a jacket's back look like. Decide which details matter before you upload.

• For a whole-person figure: use a clear full-body photograph with the outfit and footwear you want. A simple pose makes it easier to compare the result with the reference.
• For a head-and-shoulders collectible: use a portrait and request a bust. This avoids inventing a lower body you did not supply.
• For a pet: choose a photo that shows the face, ears, coat pattern, and tail. If an important marking is hidden, describe it or supply a suitable additional reference where your interface permits it.
• For an original illustrated character: use your own drawing or authorized artwork. Describe whether you want a faithful sculpture or a more stylized interpretation.

When using more than one reference, assign each a job: “Use the first photo for the face and the second for the outfit.” Conflicting photos leave the model to choose between different hairstyles, clothing, or ages. Additional references are a way to clarify a request, not a promise of exact reconstruction.

Variations you can use without rebuilding the prompt

These examples change the deliverable rather than just adding more adjectives. Use the matching reference and paste one complete prompt. For a result you already like, a short follow-up can be enough.

A portrait bust when your photo does not show the whole body

Turn the person in the attached portrait into a painted resin collectible bust. Show the head, shoulders, and upper chest on a short, plain charcoal pedestal. Use the photo for the person's face shape, hair, glasses if present, expression, and visible clothing. Give the hair and fabric sculpted detail, with a satin painted finish. Photograph the bust at eye level against a warm cream background, with soft light and a subtle shadow beneath the pedestal. Keep the whole bust and pedestal in frame. The pedestal is blank and the scene contains no other objects.

A bust makes the reference's visible details the whole subject. If you want a closer likeness, keep the proportions restrained; an oversized cartoon head changes the very shapes you are trying to preserve.

A pet collectible with its own markings

Create an image of a small painted vinyl collectible of the pet in my attached photo. Use the reference for the animal's ear shape, muzzle, coat colors, and distinctive markings. Show the pet sitting naturally, with its front paws visible and its tail resting beside it. Make the fur look like sculpted, painted toy detail rather than real strands. Place the collectible on a low sage-green circular stand, against a plain off-white background. Use a three-quarter view and soft studio lighting, showing the entire animal and stand. Keep the animal's normal anatomy and leave the scene free of writing and accessories.

Check markings individually. A figure can look convincingly toy-like while putting a white patch on the wrong side or changing the tail shape.

A box for a figure you already like

Use this as a follow-up to the accepted figure image:

Keep the existing figurine, its pose, face, colors, stand, and lighting. Add one simple collector box behind it on the right, slightly out of focus and smaller than the figure in the composition. Use a muted teal cardboard design with a clear display window and abstract cream shapes. The only lettering is "MY MINI" in large dark uppercase letters across the top. Keep the complete figurine visible in front of the box.

This gives the box one short label and a supporting role. If the label is essential for a gift or printed graphic, check every letter in the downloaded image. You can leave the box blank and add the exact wording in an image editor if generation keeps changing it.

An original character with a different finish

For your own character artwork, the starter can become a sculpture brief: “Use my attached drawing for the costume silhouette and color palette. Render it as a matte ceramic collectible with simplified sculpted folds.” Keep the desired prop visible, and say where it belongs—for example, “a closed umbrella held vertically in the right hand.” A detailed source drawing does not mean every tiny feature will survive the change of medium.

Fix the part that failed, then compare again

Use the result you are editing as the starting image and keep the reference available in the conversation. Name the visible mismatch rather than guessing why the model made it. Google's semantic editing guidance describes asking for a specific element to change while requesting that the rest stay the same. Treat that as an editing instruction, and inspect what actually changed.

A simpler pose or shorter label reduces what you are asking the image to contain. It does not diagnose every failure or guarantee the next attempt. If the face still differs after a focused correction, return to the reference choice: a sharper or less obstructed photo may communicate the missing feature better than another paragraph of instructions.

Decide when the image is ready to save or share

Judge the result against its purpose. A playful avatar can tolerate exaggerated proportions; a client concept needs the requested outfit and accessories to remain identifiable. Inspect these details at a useful viewing size:

• Identity: the key facial features, hair, or pet markings still identify your subject.
• Figure appearance: the material and sculpted shapes read as a collectible, rather than a flat cutout.
• Anatomy and support: limbs connect sensibly, and the figure sits or stands on its base.
• Presentation: nothing important is cropped or hidden behind packaging; any required text is correct.

Save an image that meets the purpose before experimenting with a new style. If a detail is essential and remains wrong, simplify that detail, change the reference, or finish it in an image editor. A realistic result alone does not prove it accurately reproduces a person or depicts an existing product.

For sharing, make clear that it is an AI-generated concept when the context could imply that a real collectible exists. Google's usage policy prohibits infringement of privacy and intellectual-property rights and deceptive impersonation. Using someone's picture or a recognizable brand on a box may require permissions beyond the image-generation tool's access terms.

FAQ

What is the prompt for the Nano Banana figurine trend?

It is a request to turn an attached reference photo into an image of a sculpted collectible. The complete starter above specifies a resin finish, recognizable appearance, a simple pose, a stand, and studio lighting. A desk, modeling screen, package box, or “1/7 scale” wording is optional; none is required to communicate the basic figurine look.

Can I make a Nano Banana figurine image for free?

Google's current image overview lists free access to Nano Banana 2, with limits, and Nano Banana 2.1 through a Google AI plan. Photo editing requires an eligible account and supported availability. Free access does not mean unlimited attempts, and consumer access is separate from Gemini API billing.

Does “1/7 scale” make the image ready for 3D printing?

No. It suggests a miniature's appearance in the scene; it does not establish dimensions or create geometry. A figurine image needs a separate modeling or image-to-3D workflow to become a mesh, and a printable part needs further checks. The image-to-3D guide explains those distinctions.

Can I use a selfie instead of a full-body photograph?

Yes. Use the bust prompt if you want only what the portrait actually shows. For a full-body figure, specify the unseen clothing yourself or accept that it will be invented. A full-body reference is the better input when the entire outfit needs to match.

How do I keep the face the same while changing the toy style?

Ask the model to use the reference for specific identifying features and change only the material or proportions you intend to change. Inspect the result against the original photo. A request to preserve identity is useful direction, not a guarantee of an exact face match.