Week 05 — 3D Scanning and Printing

This week focused on understanding additive manufacturing through printer rule testing, designing an additive-only object, and comparing different digital scanning workflows.

Tools used

Key outputs


Overview

This week was about understanding what 3D printing makes possible and where its limits appear in practice. Instead of only reading generic rules, I tested two printers directly and looked at how real prints behave under unsupported curved geometry.

I also explored the opposite direction of the workflow: converting physical or visual information back into 3D form. For this part, I compared mobile reconstruction with Luma AI and image-based generation with HunYuan 3D. These tools are fast and accessible, but they do not produce the same kind of geometric truth as fabrication-oriented workflows.


Group Assignment — Testing Printer Design Rules

For the group assignment, I compared two Bambu printers with different motion systems: the A1 Mini and the P1C. The objective was not simply to decide which machine is better, but to understand how printer structure influences the result when geometry becomes difficult.

Both tests were printed in PLA using the same benchmark model. The key area of observation was the large curved overhang, because it quickly reveals where a printer begins to lose control without support.

Printers Used

Test Setup

Benchmark Model

Benchmark overhang print model
Benchmark print used to compare unsupported curved overhang behavior on both printers.

A1 Mini Result (Red Print)

Red benchmark print produced on Bambu Lab A1 Mini
The red print was produced on the Bambu Lab A1 Mini. Most features printed clearly, but visible deformation appeared when the curved overhang approached 50 degrees.

P1C Result (Yellow Print)

Yellow benchmark print produced on Bambu Lab P1C
The yellow print was produced on the Bambu Lab P1C. The print remained stable overall, but the same curved area began to fail at around 50 degrees as well.

Close-up Observation

Close-up of curved overhang failure
Close-up of the unsupported curved section. Both printers show that the geometry becomes unreliable around the same threshold.

Critical Observation

The key result of this test is that both printers began to struggle at approximately 50° in the curved unsupported section. This is useful because it turns a generic design rule into an observed value under my own printing conditions.

In practice, this means that if I want a reliable print without support, designing for 45° or lower is a safer strategy.

Printing Summary

Test A1 Mini P1C
Overhang limit ≈ 50° ≈ 50°
Surface quality Good Slightly smoother
General stability Reliable Reliable

Under these standard PLA settings, both machines performed well. The difference was subtle rather than dramatic. The P1C felt slightly more stable, but the A1 Mini was also fully capable for this type of benchmark print.


Individual Assignment — Additive-Only Object

For the individual assignment, I designed a hollow LEGO-inspired block with enclosed internal geometry. The exterior is simple and familiar, while the interior demonstrates the core logic of additive manufacturing.

This object cannot be made subtractively because the internal curved geometry is fully enclosed. CNC tools cannot access the inside without destroying the outer shell. The structure can only be produced layer-by-layer using 3D printing.

Step 1 — Base Model

LEGO base model in Fusion 360
Initial LEGO-inspired block model created in Fusion 360.

Step 2 — Hollow Structure

Hollow LEGO model with openings
Internal cavities and side openings were added to create enclosed geometry.

Step 3 — Slicing

Slicer preview with supports
Slicer preview showing internal support structures required for printing.

Step 4 — Printed Result

Final 3D printed LEGO object
Final printed object. The outer shell is clean, while internal support traces remain visible inside the enclosed structure.

What I Learned from the Object

This design made one important point very clear: additive manufacturing does not only allow complex geometry, it also introduces new constraints after printing. Once the internal volume is enclosed, post-processing becomes much harder.


3D Scanning Workflows

For the scanning part of this week, I focused on two lightweight workflows: Luma AI and HunYuan 3D. I wanted to compare reconstruction from captured data with AI-generated 3D from image input.

Luma AI

Luma AI scanning workflow screenshot
Luma AI is fast and accessible. It works well for quick visual capture, but the mesh usually needs repair before it becomes fabrication-ready.

HunYuan 3D

HunYuan 3D reconstructed model
3D model generated from real images using HunYuan 3D.

In this workflow, HunYuan 3D was used to convert real-world photographs into a 3D model. Unlike pure generative AI, this process is based on image input, but the geometry is still partially inferred rather than directly measured.

The result captures the overall shape, but fine details and structural accuracy are limited. This makes the model suitable for conceptual exploration, but less reliable for precise fabrication.

3D printed result from the HunYuan-generated model.

Comparison

Method Input Type Geometry Source Accuracy Fabrication Ready
Luma AI Captured images / video Reconstructed Medium Needs repair
HunYuan 3D Real images Reconstructed + inferred Low–Medium No


The key difference is that Luma AI reconstructs geometry from spatial data, while HunYuan 3D reconstructs from images but still relies on AI inference to complete missing information.

As a result, HunYuan 3D is useful for quickly generating 3D content from real-world references, but the output is not yet reliable for fabrication without further processing.

The important distinction is that Luma AI tries to reconstruct an existing object, while HunYuan 3D generates a plausible form. Both are useful for early exploration, but neither replaces precise fabrication-oriented geometry.


Reflection

This week made additive manufacturing much more concrete for me. It is one thing to understand overhangs or supports as abstract design rules, and another to see them appear directly in a print result.

The printer test was useful because it gave me a measured reference point: under my current conditions, unsupported curved geometry starts to fail at around 50°. That makes future design decisions much more grounded.

The hollow LEGO-inspired object helped me understand that additive-only design is not just about complexity for its own sake. The real issue is whether a tool can physically access the geometry. Once I started thinking in terms of access, the distinction between subtractive and additive fabrication became much clearer.

The scanning comparison also clarified the difference between captured geometry and generated geometry. Fast AI tools are useful, but if the goal is fabrication, the quality and trustworthiness of the mesh still matter a lot.


Design Files