How to Create AI Cartoon Figurine Images: Prompts and Reference Images for 3D Modelling

Illustration: the conceptual progression from a portrait to a figurine design and a three-dimensional appearance. Image generation and 3D modelling are separate steps.

To turn a portrait into a cartoon figurine image with AI, start with a clear photo, then use a prompt to specify the character’s proportions, clothing, pose, viewing angle and background. After generation, check that the whole body is visible, the arms and legs make sense, and accessories have not merged into the body.

An AI-generated figurine image is a 2D design reference. To make a physical 3D figurine, you still need to build or generate a 3D model, check its structure and arrange printing. Even an image that looks like a three-dimensional toy does not contain the complete 3D geometry a printer needs.

This guide explains how to prepare a useful reference image and what still needs to happen before it becomes a physical figurine.

What is the difference between an AI figurine image, a 3D model and a physical figurine?

• AI figurine image: An image used to decide the appearance, colours and pose, usually saved as a PNG or JPG.

• 3D model: Three-dimensional data that can be rotated to view the front, back and sides. A model can be built manually or generated with an image-to-3D tool, then checked and refined.

• Physical figurine: An object made from a prepared 3D model. Its appearance also depends on its size, material, printing method and colouring process.

“3D figurine-style image” describes a visual effect. It belongs to a different stage from a “3D model ready for printing”.

Step 1: Prepare a clear portrait photo or character description

If you want to retain your hairstyle, glasses or facial features, use a portrait photo as a reference. Choose one with even lighting, a clear face and few obstructions. For a group photo, make clear which person you want to generate. Starting with a single character makes the result easier to check.

Before uploading, obtain the person’s consent and confirm that you have the right to use the photo. Crop out unnecessary bystanders, identity documents, addresses and other private details. If you prefer not to upload a real person’s photo, describe an original character in words instead. AI tools have different policies for storing images, displaying them publicly and using data, so check your chosen platform’s guidance and settings first.

Step 2: Describe a figurine design with a clear structure

A figurine image needs to do more than look cute: it should show where each part of the character belongs. These are practical ways to prepare a reference image, rather than guarantees of a successful print:

• Show the whole body: Keep the top of the head, both hands and the soles of the shoes within the frame, with a little space around them.

• Make outlines clear: Use a plain background so the hair, clothing and accessories are easy to distinguish from it.

• Choose a readable angle: A front view helps you check left and right placement; a three-quarter view makes volume easier to see. Avoid strong perspective that distorts the proportions of the head, hands or feet.

• Keep the pose simple: Start with a standing pose, both feet on the ground and arms close to the body, reducing the number of unsupported parts that need separate attention.

• Specify accessory positions: State which hand holds what and whether an object is separate, helping avoid merged hands, props and clothing.

• Choose details carefully: Thin strands of hair, glasses frames, fingers and flowing ribbons may need thickening or simplification when reduced to figurine size.

Chibi proportions—a large head and small body—are a style choice, rather than a requirement for 3D printing. Realistic characters can also serve as modelling references, with different details to preserve and process.

Step 3: Generate an image with a reusable prompt

Build your prompt around “character source + style and proportions + clothing + pose + viewing angle + background + problems to avoid”. Paste the following template into an AI tool that supports image generation, replacing the text in square brackets with your choices:

General prompt template

Generate a cartoon figurine design reference image [based on the uploaded portrait / using the following original character description]. Retain [hairstyle, glasses, facial features, etc.]. Use [chibi proportions with a large head and small body / other specified proportions], wearing [clothing and colours], in [a specific pose]. Show the character from a [front / three-quarter] view. Include the entire body, from the top of the head to both hands and feet. Use a plain white background, even lighting and clear outlines. Place [accessory] at [specific position] without covering the face. Avoid cropped body parts, extra limbs, merged hands and props, text, watermarks and complex scenery.

Original character example: a little explorer holding a book

Generate an original cartoon figurine design reference image: a short-haired explorer with chibi proportions, wearing a yellow jacket, dark blue trousers and white shoes. The character stands facing forward with both feet on the ground. Their right hand holds a closed red book below chest level, and their left arm hangs naturally at their side. Show the whole body against a plain white background, with soft, even lighting. Keep the outlines of the book and hand clear, without covering the face. Avoid extra fingers, floating accessories, text and brand logos.

These prompts are starting points for creation, rather than tested formulas that always succeed. Review the result, then refine it using the checks in the next section.

How can you use these prompts across different AI tools?

First, check which features the tool supports. With a text-to-image tool, enter the prompt directly. If reference images are supported, add a portrait or character image you have the right to use. If the tool supports image editing, describe the changes you want. If it does not, adjust the prompt and generate a new image.

If your visual concept is still unclear, you can use a text AI assistant to help organise the character description before sending it to an image generation tool. This is optional. Input methods and features vary, so follow your chosen platform’s instructions.

Step 4: Check the result and fix one issue at a time

Choose the version whose proportions, clothing and pose are closest to your idea, then check the following:

• Recognisable features: Have the hairstyle, glasses and main facial features been retained?

• Body structure: Are the number of arms and legs, joint directions and left/right positions sensible?

• Outlines: Can you distinguish the boundaries between hands, props, clothing and hair?

• Completeness: Has the top of the head, the bottom of the shoes or an accessory been cropped?

• Readable design: When you reduce the image size, can you still recognise the main clothing and props?

When asking for a change, state what should stay the same, then identify one specific issue. For example: “Keep the current face, hairstyle, clothing and colours. Change only the right hand so it holds the red book naturally, without the book merging into the arm.”

After every edit or new generation, check the whole image again. Even if you requested a change only to the right hand, compare the face, clothing, pose and accessories to make sure the parts you wanted to keep still match the original design.

Should you generate front, side and back views?

Multiple views can help explain a backpack, the back of a hairstyle or the rear of an outfit. Use your selected character image as a reference, request the same proportions, clothing and colours from other angles, then compare each image.

AI may change details between views: a pocket shown at the front might disappear in the side view, for example. If the images conflict, settle on one design and label which image should guide each part. Avoid treating contradictory images as accurate views of the same model. If a tool supports multiple input views, check that the character’s proportions, clothing and accessories are consistent before preparing the images to its requirements.

Step 5: What else is needed to go from an AI reference image to 3D printing?

Once your images are ready, you can build the character in 3D modelling software or try an image-to-3D tool. The latter can generate a model from an image, but it still has to infer the unseen back and obscured parts. Rotate the generated model and inspect it from every angle. Tools differ in the number of images they accept, background requirements and output formats, so check these before use.

Before printing, you also need to check issues that a picture alone cannot resolve:

• Model surface: Are there holes, abnormal intersections or non-manifold structures?

• Physical size: Once scaled to the intended height, are the fingers, glasses and other details thick enough?

• Pose and supports: Do overhanging arms or accessories need supports, separate parts or a revised pose?

• Balance and base: Can the feet or base support the character?

• Colour and production method: Will the model’s colours be reproduced through full-colour printing, multi-material printing or painting afterwards?

Prusa’s guide to modelling for 3D printing explains requirements such as closed models, thin walls and overhangs. Suitable thickness and support arrangements depend on size, material and printing technology. No single value suits every figurine.

A clear image helps communicate the design. Whether it can be printed depends on the final 3D model and production settings. Renaming a PNG or JPG file with an STL extension does not create 3D geometry.

Frequently asked questions

Can you 3D print a figurine directly from an AI-generated image?

A typical character image first needs to be converted into a 3D model, followed by model preparation and print settings. Some software can turn images into reliefs or lithophanes, but those are different from complete three-dimensional figurines that can be viewed from every direction.

Does the background have to be transparent?

No. Plain white or another solid colour can also work for a design reference, provided the character’s outline is clear. If you use an image-to-3D tool, follow its background and file requirements.

Must you write “suitable for 3D printing” to get a good reference image?

That phrase alone does not specify the actual structure. Describing the whole body, a simple pose, accessory positions and clear outlines makes your design requirements more specific. Printability still needs to be checked at the 3D model stage.

Will the generated figurine look exactly like the real person?

Cartoon styling changes proportions and details, and AI may also alter facial features. It can express a stylised character, but the result should not be treated as an accurate reconstruction from a real-person 3D scan.

Can you use an existing anime character or branded design?

Before preparing images for personal use, public display or commercial use, check the permissions for the characters, photos and brand assets involved. AI generation does not mean permission has been obtained. To rely less on existing characters, describe original people, clothing and colours instead.

Prefer not to generate the images yourself? Another option

If you do not have a suitable AI tool, or prefer not to write prompts and repeatedly revise images, explore the Beets AI 3D Figurine Vending Machine. The service starts with a photo you upload. The system generates figurine designs for you to choose from, then we arrange production of a physical full-colour 3D figurine.

This is an alternative workflow that starts with a photo. Before ordering, check the service page for design choices, sizes and terms of use to decide whether it suits your needs.