This project elegantly demonstrates how AI can turn personal memories into tangible objects, moving 3D printing beyond generic models toward truly bespoke creation. It marks a significant shift where technology finally makes high-end personalization accessible to the everyday maker.
Deep Dive
Prerequisite Knowledge
- No data available.
Where to go next
- No data available.
Deep Dive
I Turned My Son Into a LEGO Figure With Hi3D
Added:Every 3D printer owner eventually has the same thought. What if I could turn something personal into a real object?
Not just download another model from the internet, but create something that is actually unique. A small toy based on a real person, a custom character, or even a gift that nobody else has.
So today, I want to test a new high 3D workflow. I'll start with a simple photo of my son and try to turn him into a cute Lego-style minifigure, prepare it as separate printable parts, and see how close we can get to a real physical minifigure.
The interesting part is not just generating a 3D model. The real question is whether the model can be split, printed, assembled, and displayed like a small custom toy.
Normally, if I want to print a small character, I would probably search for a ready-made STL online. That works when you want a generic figure, but it does not help when the idea is personal. In this case, I don't want a random toy character. I want to create a figure inspired by a specific photo with a similar hairstyle, clothing colors, and overall personality.
That is exactly where AI 3D generation becomes useful because the starting point is not an existing file. The starting point is a personal image.
The goal is to use high 3D as a shortcut between the photo and the first printable version of the character.
Before generating the model, the input needs to be clear. If the image has too much background, strange lighting, or unclear body shape, the AI has to guess too much. For this test, I'll use the image editing step to guide the result before creating the 3D model. The prompt is important here because I don't just want a realistic scan of a child.
I don't want a realistic scan of my son.
I want a cute Lego-style character that still keeps the recognizable details.
So the prompt describes the figure as a small toy character with the same hairstyle, clothing colors, and general facial feeling, while also making it suitable for 3D printing. After the image and prompt are ready, high 3D generates the character model. This is the first important checkpoint because I want to see whether it understood the idea correctly. The model does not need to be perfect at this stage. What I'm looking for is the overall character direction. Does it look like a toy? Does it have a clear head, body, arms, and legs? Is the silhouette simple enough for printing? And does it still feel connected to the original photo? This is where the workflow starts to become practical. Instead of sculpting the entire character manually from scratch, I already have a first version that I can evaluate and adjust. A complete character model can look good on screen, but it is not always ideal for 3D printing. If everything is one solid piece, the print can become more difficult, supports can be messy, and small areas such as arms or joints may not come out cleanly. This is where Hai3D's split to print disassembly function becomes useful. Instead of manually cutting the model in Blender or designing connectors from scratch, the system can separate the character into printable parts. For this type of toy figure, that means the head, torso, arms, and legs can become separate components. The workflow is much closer to making a physical toy because each part can be printed more cleanly and then assembled afterwards.
This part is important because a printable toy is not only about the outside shape. The connections between the parts matter as well. If the joints are too loose, the figure will not hold together properly. If they are too tight, assembly becomes difficult or the parts may break.
Hai3D gives control over the split parts and connection structure so the model can be prepared more like a real assemblable object. For small figure like this, I would keep the design simple.
Once the parts are ready, they still need to be arranged on the print bed.
Normally, I would have to position every piece manually, rotate them, and think about supports and print orientation.
But, Hai3D's smart arrangement does this automatically. It places all the parts efficiently, suggests better printing angles, and tries to minimize supports.
More importantly, it avoids putting joints and connection areas in places that could affect the final assembly.
Now the digital model becomes a real object. Each part can be sent to the printing workflow, printed separately, cleaned up, and assembled.
This is the stage where the whole idea either works or fails.
A model can look impressive on a website, but for makers, the real test is whether it can become something you can actually hold.
After printing the parts, the figure starts to feel much more like a custom toy rather than just an AI experiment.
It began as a personal photo, went through image editing, became a 3D character, was separated into printable parts, and now it can be assembled into a small physical model. There is another feature I want to show quickly because it has a different use case. Instead of creating a full 3D character, Hai 3D can also turn an image into a relief style design.
For example, you can take a photo and create a World Cup style player card with raised details, a custom name, colors, and a collectible card look.
For this video, I would treat it more as a website feature demonstration rather than a normal 3D printing test because this type of result depends heavily on the printer, material, and finishing process. But, as a creative tool, it is interesting because it turns a flat personal image into something with depth and a more physical presentation.
What impressed me here was not just that Hai 3D can generate a 3D model. Many AI tools can create something that looks good on screen. The more useful part is the workflow around it. Starting from a personal photo, I can generate a toy style character, prepare it as separate printable parts, adjust the connections, and move much closer to something that can actually become a physical object.
For makers, parents, collectors, or anyone interested in personalized gifts, this is a much more practical use of AI 3D generation. You are not just downloading a model someone else already made. You are turning your own idea or memory into something printable.
Starting from June 24th, Hi3D is launching its biggest promotion ever with discounts of up to 80% off. If you want to try it yourself, use the tracking link and promo code in the description. You will also receive an extra 500 credits on top of the promotion, which is a good opportunity to test your own ideas and prototypes.
Related Videos

Setting up a curved screen with Immersive Calibration Pro 4 and multiple cameras (P3D v4)
FlyerOneZero
23K views•2019-07-21

Robot Learning with Sparsity and Scarcity
allenai
379 views•2025-10-14

Jorge Mendez-Mendez: Unlocking Lifelong Robot Learning With Modularity (2023-10-05)
umassmlfl
237 views•2024-01-06

Northwestern’s MS in Robotics: Student Robotics Projects, 2023
NorthwesternEngineering
1K views•2024-05-31

"Perfect" Turns: Turning by the Gyro - FIRST LEGO League (FLL) SPIKE Prime + EV3 RePlay Programming
ZacharyTrautwein
94K views•2020-10-02

Gorkem Secer: TSLIP-based Deadbeat Running Control of Bipedal Robot ATRIAS
DynamicWalking-wv6qm
298 views•2018-06-22

Self-Driving Cars Need Lessons On Human Drivers | Maddie About Science
skunkbear
26K views•2018-08-21

Milrem Robotics’ THeMIS UGVs used in a live-fire manned-unmanned teaming exercise
MilremRobotics
99K views•2021-05-20
Trending

2.4 BILLION Records Got Leaked...
DeepHumor
15K views•2026-07-22

Playstation NO DISC/NO BUY Fight Is Over...
DavidJaffeGames
4K views•2026-07-23

Should I buy a Sawmill?
essentialcraftsman
29K views•2026-07-22

Americans Confused in Australia for 17 Minutes Straight
IWrocker
17K views•2026-07-23