I run a creative coding club called TypeLab. While running it, I noticed a recurring frustration: code-generated work often felt fleeting—more like a temporary rendering than something creators could hold, share, or claim as their own. This raised a broader question about what kind of material presence digital work can have, and how creators might regain a sense of ownership over what they make.

Seoul-based creative coding club, TypeLab

We shared a recurring feeling that code art can seem distant. We started asking what kind of materiality code art can have, and what realistic ways there are to distribute it. I wanted to unpack the questions we raised together that day.

To explore the problem space more deeply, I conducted desk research to gather actionable insights. The research revealed that as digital creative tools become faster and more automated, creators experience increased efficiency—but at the cost of authorship and ownership. Prior studies show that automation often hides the visible effort behind creative work, weakening psychological ownership without actually improving the quality of ideas.

1. Brainstorming feels more effective with conversational AI

Participants reported that back-and-forth interaction with AI supported ideation by helping them explore ideas faster and iterate more fluidly during brainstorming.

2. Quality of ideas remains largely unchanged

Self-reported idea quality showed little difference between no-AI and human-AI collaboration, despite noticeable changes in workflow and speed.

3. Automation reduces authorship and ownership

On a 1-7 scale measuring perceived authorship, creators reported a sharp drop in ownership as AI took over more of the writing.

Key Insight

Automation can increase efficiency and perceived usefulness without improving idea quality—often at the cost of creator ownership, making it essential for tools to intentionally preserve agency, effort, and authorship through interaction.

From observing how automation improved efficiency yet weakened creators’ sense of ownership, I was motivated to look deeper into the IKEA Effect as a way to understand how effort contributes to value and authorship.

What creates ownership?

Effortful contribution | Physical assembly, Visible mistakes and corrections, Time and cognitive investment

What often disappears in digital tools?

Effort and visibility | One-click results, Hidden intermediate steps, Minimal personal trace

Design implication

Ownership emerges when tools make effort, decisions, and corrections visible—suggesting that automation should support making, not replace it.

Synthesizing insights from desk research and first-hand observations within a creative coding practice, I reframed digital making around effort and assembly—shifting away from one-click generation toward hands-on construction.

Key Question

Reframing digital making through effort and assembly

"How might a creative coding tool preserve effort, decision-making, and authorship instead of collapsing them into one-click outputs?"

One of the representative creative coding platform, p5.js editor

Ideation Leap

From Generation to Assembly

Shifting digital creation toward hands-on construction

I translated ownership-through-effort into a computational model by treating digital outputs as assemblable structures—drawing from physical assembly (IKEA), hands-on craft (origami), and printable nets.

Building on research findings and synthesis, TypoFold is guided by three principles: making structure visible, treating effort as meaningful rather than friction, and reframing automation as material that supports—not replaces—user participation.

TypoFold supports an iterative making flow where users move between writing code, inspecting form, and assembling physical outputs—using each step to refine the next.

To support the iterative making flow shown in the previous scenario, TypoFold translates code-generated 3D forms into foldable paper nets through a structured 3D-to-2D pipeline.

While triangulation is efficient for rendering, it became a critical obstacle when building TypoFold—fragmenting perceived surfaces and preventing reliable generation of foldable nets for physical assembly.

Problem

Perceptual Surfaces Were Fragmented by Triangulation

When TypoFold imports 3D letterforms, each surface is automatically triangulated for rendering.While efficient computationally, this breaks a single perceived surface into many small faces—making it difficult to generate paper nets that align with how users cut, fold, and assemble physical forms.

Solution

To address this, I grouped adjacent triangles that (1) share edges and (2) have aligned surface normals,allowing the system to reconstruct perceptually meaningful faces.

This enables the unfolding process to operate on surfaces as users perceive them—rather than on low-level geometric primitives.

After reconstructing perceptual faces, a another issue emerged during net generation. When unfolding these faces into 2D layouts, internal surfaces were still included—causing overlaps that broke physical assemblability and required a more selective unfolding strategy.

Problem

Unfolding Internal Faces Breaks Physical Assembly

When unfolding 3D letterforms into paper nets, including all faces caused internal surfaces to overlap in the 2D layout. These internal faces do not contribute to physical assembly and instead produce nets that cannot be cut, folded, or assembled correctly.

Solution

To address this, I ordered faces by orientation and applied a DFS-based unfolding strategy that skips internal faces. This preserves physical assemblability while generating valid, non-overlapping paper nets.

After resolving structural constraints in net generation, I evaluated how different unfolding strategies affected real-world cutting and folding. I conducted a usability preference test to identify which unfolding method best supports physical assembly.

Key Question

"Which unfolding strategy makes paper craft templates easier to cut, fold, and assemble?"

Participants consistently preferred side-priority unfolding, reporting that it was easier to follow, cut, and assemble. Based on this feedback, TypoFold adopts side-priority unfolding as the default method.

User Test

User Preference Survey by Unfolding Method

The final web interface of TypoFold, showcasing its core features and user flow.

Examples of paper craft outputs created by generating, printing, cutting, folding, and assembling 3D letterforms from A to Z using TypoFold.

After the solution was developed, I ran a hands-on workshop during HCI Korea 2025 to evaluate how TypoFold’s design translated into real user behavior. The workshop focused on whether engaging in the full process—generating, assembling, and finishing a single letter—would reinforce users’ sense of ownership through effort and physical making.

Workshop results show that hands-on assembly increased both enjoyment and perceived ownership, supporting TypoFold’s goal of reinforcing authorship through effort.

*18 participants responded to the post-workshop survey

In the next phase, I extended TypoFold beyond letterforms to test how the unfolding pipeline performs across more complex 3D geometries.

By applying the system to varied shapes with different topology and surface structures, I evaluated whether the workflow could generalize beyond typography while maintaining physical assemblability and visual coherence.

This shift reframes TypoFold from a typography-specific prototype into a broader method for translating digital 3D forms into foldable, physical structures—while surfacing new challenges around face segmentation, overlap handling, and consistency at scale.