Author: Professor Chang Meijun
Abstract
Canvas 152 is a hand-drawn embroidery structure created on 6 mm grid paper as part of the Yunbroidery AI Embroidery Structure Project. The diagram records a sequence of sixteen embroidery paths that describe the execution order of stitches. The drawing was originally developed through the repetition and extension of a small motif without intentionally designing a central point.
After the structure was translated into analyzable data, an AI-based structural reconstruction was performed. The system successfully generated multiple possible path interpretations (A, B, and C) based on stitch entry–exit relationships, path continuity, spatial distribution, and tension stability.
The analysis also revealed an emergent radial organization in which the structural center appeared naturally within the composition, despite not being intentionally designed during the drawing process.
This case demonstrates that embroidery structures can be interpreted as logical systems that can be analyzed and reconstructed by artificial intelligence. Canvas 152 therefore represents an early experimental example of AI beginning to understand embroidery structure.
Introduction
During the development of the Yunbroidery AI Embroidery Structure Project, a remarkable moment occurred in the study of Canvas 152. What originally began as a hand-drawn embroidery structure diagram unexpectedly revealed the possibility that artificial intelligence could begin to interpret the structural language of embroidery.
This case has become an important research milestone within the Yunbroidery project.
The Origin of the Hand-Drawn Diagram
Canvas 152 was originally drawn on 6 mm grid paper. The numbers 1–16 marked on the right side indicate the execution order of the embroidery path, representing the actual movement of the thread across the fabric surface.
When this diagram was created, there was no predetermined center and no intention to design a symmetrical composition. The structure began with a small motif and gradually expanded through repetition and extension.
However, after the drawing was completed, careful observation revealed that the structural center naturally appeared at the exact middle of the composition.
AI Structural Reconstruction
Once the hand-drawn diagram of Canvas 152 was translated into analyzable structural data, AI began to reconstruct the embroidery path structure.
The analysis was based primarily on the following principles:
• relationships between stitch entry and exit
• continuity of the path
• stability of thread tension
• spatial distribution of the pattern
Under these conditions, the AI generated three possible path structures:
A — primary structural path
B — balanced alternative path
C — rotational path structure
These paths were not manually designed but generated through AI interpretation of embroidery structural logic.
The Hidden Center
Another remarkable phenomenon discovered in Canvas 152 is the presence of a “hidden center”.
Although no central point was intentionally designed, the AI structural diagram reveals a radial distribution of the entire path system.
This suggests that embroidery structures may achieve global balance through local repetition.
Implications for the Embroidery Structure Database
The case of Canvas 152 demonstrates that embroidery is not only visual decoration but also a structured system governed by internal logic.
When stitch points, paths, and structural relationships are recorded as data, AI can begin to analyze and reconstruct embroidery structures.
The long-term goal of the Yunbroidery project is to establish a global embroidery structure database.
Conclusion
Canvas 152 began as a hand-drawn embroidery path diagram, yet through AI analysis it became an important experimental case.
In this study, AI not only analyzed embroidery paths but also generated multiple possible structural paths.
Canvas 152 therefore records an important moment:
When artificial intelligence begins to understand embroidery structure.
