3d-printed-fu

Week 5: 3D Printing and Scanning

This week felt like a small journey. I wanted to turn a traditional "Fu" (福) pattern into a real 3D object, and at the same time test scanning tools with real objects in daily lab conditions.

My long-term goal is still my final-project idea, a "smart Fu", so every test this week became useful practice. I was not only trying to "get a print," but also trying to understand where and why each step succeeds or fails.

Tools and materials

• Printer: Bambu Lab A1 mini
• Material: PLA filament
• Slicer: Bambu Studio
• 2D vector editing: Inkscape
• 3D CAD: Onshape
• Quick image conversion: ImageToSTL
• 3D scanning apps: Polycam, Luma AI

Group assignment

For group work, we tested our printer design rules: overhangs, clearances, and wall thickness. This helped me understand where the machine is reliable and where designs start to fail.

Detailed group documentation: Week 5 group assignment

Testing and recording

The printing process went smoothly:

Angle test showed some stringing and uneven surfaces at angles below 30 degrees. The 20, 10, and 0 degree overhang surfaces collapsed and could not maintain flatness.

Thickness test had three thin sections completely missing from the print, all the slots were present, but the thinnest walls simply did not appear.

Overhang test generally looks great, but the last one overhangs 10mm looks very rough.

Overhang and bridging

We checked the quality of different overhang angles on Angle test.

Overhang Angle (degrees)	Overhang Surface Quality
90	Perfect
80	Perfect
70	Perfect
60	Perfect
50	Perfect
30	Good, slight protrusion at edges
20	Good, slight protrusion at edges
10	Noticeable stringing
0	Accumulation and uneven surface

Wall thickness

We checked the completeness of different wall thicknesses:

Design Thickness (mm)	Actual Thickness (mm)	Deviation (mm)
0.1	Not printed	—
0.2	Not printed	—
0.3	Not printed	—
0.4	0.5	0
0.5	0.5	0
0.6	0.6	0
0.7	0.7	0
0.8	0.8	0
0.9	0.9	0
1	1	0
2	2	0
3	3	0

The 0.1–0.3 mm walls were ignored entirely during slicing and never appeared on the print. From 0.4 mm upward, the results were very close to the design, with only a small 0.1 mm deviation at 0.4 mm.

Overhang distance test

We checked how the printer handles increasing overhang distances — how far a section can extend horizontally without support before the quality breaks down.

Overhang Distance (mm)	Surface Quality
1	Fine
2	Fine
3	Fine
4	Fine
5	Fine
6	Fine
7	Slightly rough
8	Slightly rough
9	Very rough
10	Very rough

Up to 1 mm, the overhang still looked clean. From 2–5 mm, the surface got a bit rough but held its shape. From 6 mm onward, the print couldn't bridge the gap properly, lots of drooping and stringing.

What we learned

Through this test, we got a clearer picture of the Bambu Lab A1 mini's real-world limits with PLA:

• Overhang angles of 30° and above print reliably. Below 30°, expect stringing or surface collapse.
• Wall thicknesses below 0.4 mm are too thin to print — the slicer skips them entirely.
• Overhang distances up to 1 mm print cleanly. From 2–5 mm, expect some surface roughness. Beyond 6 mm, stringing and spaghetti become a real problem without supports.
• First layer adhesion was solid for both test models.

These numbers give me a practical reference for my own designs. If I'm designing something with fine details, I now know to keep walls above 0.4 mm and avoid unsupported overhangs longer than 5 mm.

Individual Assignment: from 2D Fu to 3D print

I decided to try two workflows and three print iterations:

1. A quick "just print it first" workflow.
2. A more controlled workflow with vector cleanup and CAD modeling.

First attempt: quick test with ImageToSTL

I began with a high-contrast Fu image from the web and uploaded it to ImageToSTL.

• The platform generated an STL automatically from image contrast.
• In the converter, I set the thickness around 20 mm in the Z direction, but this first run was still a practical thin test after slicing and scaling.

• I registered and prepared the file in Bambu Studio.

• The initial scale was too large for the machine, so I resized it to fit the A1 mini's build area.

• I connected to the A1 mini over Wi-Fi (IP + pairing code) and sent the sliced file wirelessly.

• I kept this run simple with basic PLA settings, no advanced supports, and no multi-color features, because my goal was to test behavior quickly.

I am not suppose to design a sticker, but it looks good, thin and flexible and light, almost feel like a paper.

That first print still helped a lot: it quickly told me the visual direction works, and it looks very smooth, the material looks good.

Second attempt: Inkscape + Onshape for full control

After that, I switched to a cleaner workflow so I could control the structure.

Step 1: prepare vector in Inkscape

• I imported the Fu image and used Trace Bitmap.
• I cleaned the nodes and connected anchor points.
• I used Union to make sure the design became one closed path.
• Then I exported the vector for CAD use (SVG/DXF workflow).

Step 2: build 3D model in Onshape

• I imported the vector onto a sketch plane in Onshape.
• I scaled and positioned the shape.
• I extruded 3mm to give it thickness and generate a solid body.
• I chose this method because a parametric model is easier to adjust later for my "smart Fu" version (for example: thickness updates, back plate, electronics-related structure).

Step 3: the failure I almost missed

In this version, I increased thickness but forgot to create one continuous backing layer behind the strokes.

The print came out as separated parts, not one integrated decoration.
When I looked back, the issue was already visible in preview, but I did not catch it before printing.

Step 4: redesign and final successful print

I redesigned the pattern again and added a solid base plate under the Fu character.

I also tested orange filament to make the pattern more vivid and easier to read visually.

This time, the print came out as one complete piece, with much better strength and a cleaner finish. Looks very nice:D

What changed between attempts

• Attempt 1 (quick STL conversion): visually clear, but too thin and flexible.
• Attempt 2 (parametric model without full backing): thicker, but disconnected geometry and failed as a single rigid object.
• Attempt 3 (parametric model + solid base layer): structurally connected, successful one-piece print.

Print notes

• Build volume used: 180 x 180 x 180 mm
• Nozzle: 0.4 mm stainless steel nozzle
• Material tested this week: PLA (single-color print)
• Transfer method: Bambu Studio + Wi-Fi
• Process strategy: quick prototype first, then parametric redesign
• Printer capability note: the A1 mini can also support materials like PETG, TPU, and PVA, and supports AMS Lite for multi-color, but I did not use multi-color this week.

Plus. I deisgned some decorations for my final project, I am quite satisfied with the result:

3D scanning

I also tested two scanning tools and compared how they behaved in real use.

1) Polycam

I scanned a strawberry porcelain ornament and a Seeed mug using Polycam.

I tried Polycam several times (four attempts total).
My early scans failed when photo coverage was incomplete.
In the successful run, I took almost 60 photos around the object; around 30 photos was not enough in my test.

2) Luma AI

I also tried Luma AI with both photo upload and video upload.
In my case, the upload stayed at 5% for around two hours and failed.

Reflection

This week reminded me that a "nice shape" is not the same as a "printable structure." The key turning point was adding one continuous base layer under the Fu strokes. I also learned that fast tools are great for first experiments, but for real iteration I need parametric CAD. For scanning, coverage and patience matter much more than speed.

I like this result because it keeps the traditional feeling of a window-flower Fu, but now I understand the engineering side much better: when lines look connected on screen, they still need true structural connection for real printing.