6. Computational Couture

References & Inspiration

My biggest inspiration this week is still Behnaz Farahi’s “Caress of the Gaze.” I keep coming back to her work because of the way she combines digital fabrication, material design, and electronics to question ideas around intimacy, gender, and personal identity. It’s not just about the tech — it’s how she makes the materials feel alive and aware.

For this week’s assignment, I want to explore how 3D printing can become part of the fabric itself. I’m curious about using parametric design to create dynamic surface effects — something that feels like it’s moving or breathing. I also want to experiment with stretchy fabrics to see how 3D printing reacts to tension and flexibility, and whether I can create something like a smocking effect, but generated digitally through form and material behavior.

Caress of the Gaze by Behnaz Farahi — a kinetic, 3D-printed wearable that responds to the gaze of others, exploring the boundaries between body, technology, and emotion.

Making Fabric with 3D Printing

I started this week’s assignment by trying to make a simple 3D-printed fabric. I followed a YouTube tutorial by Sara Alvarez, which was super easy to understand and fun to follow.

Tool

Tool Type	Name / Specification
Modeling Software	Rhino
Slicer	PrusaSlicer
3D Printer	Original Prusa i3 MK3
Filament	Flex TPU 85A (White)

Steps

1. Model the base: I created a small solid fabric block sized 100mm x 100mm x 0.3mm in my 3D software.

2. Export as STL: Once the model was ready, I exported it as an STL file for slicing.

3. Set infill pattern: In PrusaSlicer, I selected a 3D honeycomb infill to give it flexibility and texture.

4. Choose material: I printed with TPU 85, which was recommended during Kadian Gosler Inspirational Talk.

5. Print and test: The final print came out soft, slightly stretchy, and fabric-like—though still a bit fragile. It was exciting to see how a printed pattern could start to behave like cloth.

Flexible 3D-printed fabric with honeycomb infill — modeled in Rhino and sliced in PrusaSlicer — by Pattaraporn (Porpla) Kittisapkajon, following Sara Alvarez's Youtube Tutorial

Alternating Loop Pattern in Blender (Geometry Nodes)

I’m not sure if Blender is just an incredibly well-designed software, or if Rico Kanthatham is simply a brilliant instructor — probably both! Following his tutorial, I found that parametric design in Blender feels surprisingly intuitive and playful once you get the hang of it.

You can find Rico’s tutorial here — it’s super clear, beginner-friendly, and fun to follow along. His way of explaining each step made the process feel less technical and more like sculpting with logic.

Once again, I found myself struggling with wanting to do so many things for this assignment — I had a whole list of ideas. But with time constraints and other responsibilities, I decided to narrow my exploration to just one focus: creating an alternating loop effect, inspired by Cristopher Kanne’s 3D-printed sweater. It felt like a manageable but still exciting challenge to experiment with how digital patterning could mimic the softness and rhythm of knitted loops.

Goal

Make a small “loop” shape, tile it on a grid, then make every other row behave differently (scale/offset) so it feels like woven/knitted loops.

Step 1 — Start a Geometry Nodes setup

1.1 Open Blender → add a Mesh > Plane (any mesh is fine).

1.2 With the mesh selected, click Modifiers → Add → Geometry Nodes → New.

Step 2 — Build one loop (the “unit”)

Node Setup for Generating a Loop Structure in Blender Geometry Nodes — by Pattaraporn (Porpla) Kittisapkajon

Step 3 — Prepare a grid of “points” to place loops on

3.1 Add Grid (Mesh Primitives → Grid).

Size X/Y: e.g., 120 mm x 120 mm

Vertices X/Y: e.g., 8 x 8 (more = denser fabric)

3.2 Add Instance on Points:

Points = Grid

Instance = your loop mesh (from the Boolean)

3.3 Why the Group Input Node Matters

The Group Input node lets you expose controls (like size and vertex count) in the modifier panel on the right side of Blender.

This means you can later adjust the grid size or density without reopening the node editor, making your setup flexible and parametric.

Think of it as creating “sliders” for your design.

You should now see a field of loops.

Node Setup for Creating a Grid of Points and Placing the Loop Instances — by Pattaraporn (Porpla) Kittisapkajon

Step 4 — Select every other row (the trick for the alternating effect)

We’ll compute the row index from the point Index, then keep only even/odd rows.

4.1 Add nodes:

Index (Utilities → Index)

Divide (Math): set A = Index, B = number of columns (your Grid Vertices X).

This converts the flat index to a row number.

Floor (Math) after Divide → gives a clean row index.

Modulo (Math) with Value = 2 → outputs 0,1,0,1… (even/odd rows).

Greater Than (Math) with A = Modulo output, B = 0 → Boolean mask (true on odd rows).

4.2 This Boolean becomes our Selection for adjustments.

Node Setup for Selecting Every Other Row — by Pattaraporn (Porpla) Kittisapkajon

Step 5 — Make alternating rows behave differently

5.1 Add Scale Instances:

Selection = the Boolean from Greater Than

Scale: try Y = 2.0 (stretches every other row).

5.2 Add Translate Instances:

Selection = same Boolean

Translation X: offset by half the grid cell width (e.g., +25 mm if your cell is ~50 mm wide).

This creates that woven/staggered rhythm you can see in your screenshot.

Node Setup for Making Every Other Row Behave Differently — by Pattaraporn (Porpla) Kittisapkajon

Step 6 — Tune + preview

6.1 Adjust Grid sizes, Vertices X/Y, the loop Arc radius, and Curve Circle radius until it looks right.

6.2 If loops intersect, lower the Scale or increase the Grid size.

Possible Alternating Loop Pattern Variations — by Pattaraporn (Porpla) Kittisapkajon

Step 7 — Bake for Export

7.1 Add Realize Instances after your adjustments and before Group Output.

7.2 Back in the Modifiers panel, click the ▼ on the Geo Nodes modifier → Apply (or Object → Convert → Mesh).

7.3 File → Export → STL. (Units in mm recommended.)

Step 8 — 3D Printing

8.1 Import the STL file

Open your slicer software of choice — here I’m using PrusaSlicer — and import your .stl file.

8.2 Adjust the print settings

Material: TPU 85A
Layer height: 0.2 – 0.25 mm
Print speed: slow (≈ 20 – 30 mm/s)
Retraction: off or very low (for TPU, this helps prevent jams)

8.3 Generate and preview G-code

Slice your model to create G-code, preview the toolpath, and adjust settings if needed. Re-slice until the layers and infill look correct.

8.4 Export G-code to SD card

Once satisfied with the preview, export your G-code file and save it to the SD card for your 3D printer.

8.5 Prepare for printing

8.5.1 Unload any existing filament.

8.5.2 Load the TPU 85A filament.

8.5.3 Prepare the print bed — if printing on fabric, secure it with tape to prevent shifting.

8.5.4 Insert the SD card, select your file on the printer, and start the print.

3D-printed alternating loop pattern — flexibility and deformation test on mesh fabric — by Pattaraporn (Porpla) Kittisapkajon

Creating Smocking effect with 3D Printing and Strechy Fabrics

Following my earlier experiment with self-smocking fabric from Week 3: Circular Fashion, I became curious about whether a similar smocking effect could be created using 3D printing on stretchy fabrics.

This idea was further inspired by Anastasia’s reference to Alek Bursac’s “Unfolding the Fold” — a project that explores how traditional pleating techniques can inform new digital fabrication methods, allowing folds to form directly on fabrics that are difficult to pleat by hand.

Reference — Alex Bursac's Unfolding the Fold

Experimentation #1 — Printing a Pre-Designed Smocking Pattern

For my first trial, I traced an existing smocking pattern found online and recreated it in 3D modeling software. The goal was to see if a traditional sewing-based pattern could be translated directly into a 3D-printed texture when printed onto stretchable mesh fabric. Once modeled, I exported the design as an .stl file and printed it using TPU 85A filament.

Results

The overall pattern transferred well onto the fabric surface, with clear and consistent line quality. The printed “V” shapes adhered smoothly to the power mesh and maintained their form after printing. However, unlike traditional smocking, the fabric did not contract significantly or create deep folds.

Instead, the result produced a slight surface gathering effect — more like a subtle relief texture than a true pleated structure. This suggests that while the visual geometry of the smocking was captured, the mechanical tension (which usually comes from thread pulling fabric together) wasn’t fully replicated in this first attempt.

Possible causes include:

• The TPU lines were not connecting between tension points, so no opposing forces were created.

• The fabric was not fully stretched during printing, reducing potential contraction after release.

-The pattern spacing may be too large to trigger the self-gathering effect.

Overall, this experiment successfully demonstrated adhesion and print precision on mesh, but revealed that further structural adjustment is needed to achieve the 3D folding behavior of real smocking.

Canadian smocking pattern directly 3D-printed onto stretch fabric.

Experimentation #2 — Modifying the Pattern for Tension and Movement

After observing that the first pattern mainly produced surface texture rather than structural folds, I focused on redesigning the geometry to introduce tension-based interaction between the printed lines and the stretch fabric.

Design Approach

Instead of using open “V” shapes, I explored patterns where each printed element connects two or more anchor points across the fabric — allowing the TPU to pull areas together once the fabric relaxes. The idea was to replicate how traditional smocking stitches physically draw fabric into gathered clusters, but through material behavior rather than thread.

I also adjusted the bar length and spacing (from 15 mm spacing and 8 mm bars to slightly shorter connections) to encourage stronger contraction without overstretching or detachment. Since fully stretching the fabric on the print bed proved difficult, I designed this version to work with only partial tension (about 20–30%), making it more practical for real testing conditions.

Reference Smocking Technique and Corresponding TPU Actuation Pattern

Results

Test 1

Mesh Preparation

4-Way Stretch Setup on Prusa i3 MK3

The mesh was stretched evenly in all four directions using an embroidery hoop to maintain tension during the 3D-printing process.

Fabric Pre-Stretched for Over a Week

Reference Smocking Technique and Corresponding TPU Actuation Pattern

In this test, the mesh had been stretched and left under tension for more than a week before printing. Because the fibers relaxed over time, the fabric no longer contracted after printing. As a result, the TPU pattern remained mostly flat, producing little to no 3D deformation.

Fabric Stretched Immediately Before Printing

Reference Smocking Technique and Corresponding TPU Actuation Pattern

Here, the mesh was freshly stretched immediately before printing. The fibers still retained active tension, so once the TPU cooled, the fabric contracted around it. This produced a more dynamic, arched surface as the TPU locked the stretched mesh into a curved, smocking-inspired formation. However, the contraction was still not strong enough to generate pronounced creases like traditional smocking.

Test 2

Since the previous test did not produce strong enough contraction to form pronounced creases, I questioned whether stretching the mesh in four directions was preventing the fabric from folding. In this test, I stretched the fabric in only two directions and adjusted the TPU pattern by reducing the strip length and doubling their thickness. The goal was to increase shrinkage and encourage more defined crease formation.

2-Way Stretch Setup on Bambu Ps1

Effect of 2-Directional Tension on TPU Contraction (Printed on Bambu P1S)

The TPU did not contract significantly in this test, likely because the fabric was not stretched tightly enough during setup. After switching from the Prusa i3 MK3 to the Bambu P1S, it became more difficult to fully tension the mesh inside the enclosed printing chamber. Without strong, even pre-stretching, the fabric cannot release enough stored tension for the TPU to create pronounced deformation.

For this test, I also increased the thickness of the TPU elements in hopes of generating stronger contraction. However, it is unclear whether the added thickness improved the effect, or if the limited fabric tension prevented any noticeable response. My hypothesis is that the results would be more successful if the mesh could be fully and consistently stretched before printing.

Experimentation #3 — Modifying the Pattern into a Connected Grid

Connected Grid TPU Pattern — Experiment 3

After the previous tests, I realized that the most effective way to achieve stronger and more dynamic deformation is to print the TPU pattern as a fully connected grid, meaning all elements are linked with no gaps between them. When the TPU is continuous, the contraction forces distribute more evenly across the fabric, creating larger and more visible 3D formations.

For this test, my stretching setup was still not optimal, so the mesh was not fully tensioned during printing. Despite this limitation, the connected grid already begins to lift and form three-dimensional structures. With a properly pre-stretched setup, I believe this pattern would produce even more dramatic deformation.

Fabrication files

File: Vexelized Mesh