Yunbroidery White Paper 1
Embroidery as a Computational Language System
Dataset v3, Execution System, and Grammar Framework
Abstract
This white paper presents Yunbroidery, a novel framework that transforms embroidery from a traditional craft into a structured computational language.
By integrating dual-surface representation (front/back), temporal execution modeling, and grammar-based annotation, embroidery is redefined as a sequence of structured operations.
Dataset v3 introduces three major innovations: cross detection, region classification, and role assignment. These enable embroidery to be analyzed, modeled, and learned by artificial intelligence systems.
1. introduction
Embroidery has historically been understood as a visual and material craft. However, its underlying logic—stitch sequences, path decisions, and tension control—reveals a structured system comparable to language.
Yunbroidery proposes a shift:Embroidery is not only an image.
It is a language composed of operations.
2. System Overview
2.1 Dual-Surface Model
Embroidery is represented as two synchronized layers:
✦Front (Surface): visible structure
✦Back (Path): execution logic
These layers form a cause–effect system.
2.2 Execution Sequence
Each embroidery process is modeled as:Step 1 → Step 2 → ..... → Step N
Each step is a discrete computational unit.
2.3 Grammar Layer
Each stitch is annotated with:
✦Direction (dx, dy)
✦ Distance
✦Region
✦Role
✦Event
This transforms execution into grammar.
3. Diagram System
Figure 1 — Dual-Surface Execution Diagram
[[Front Structure] [Back Path]
Visible Result Hidden Logic
Description:The front shows structural outcomes, while the back reveals procedural generation. Synchronization allows mapping between cause and effect.
Figure 2 — Embroidery Grammar Diagram
[Stitch] → [Vector] → [Region] → [Role] → [Event]
Description:Each stitch is progressively enriched from geometry to semantic meaning, forming a grammar system.
Figure 3 — Dataset v3 Structure
✦ Token Level:Stitch = (from, to)
✦ Feature Level:(dx, dy, distance, direction)
✦ Semantic Level:(region, role, event)
4. Dataset v3 Specification
Each frame contains:
✦ step, layer, type
✦ from, to
✦ dx, dy, distance, direction
✦ is_return, is_loop, is_cross
✦ region_from, region_to
✦ tension_level
✦Role
✦Event
5. Core Innovations
5.1 Cross Detection
Identifies structural intersections between stitches.
Meaning:Represents tension, constraint, and structural complexity.
5.2 Region Classification
Defines spatial hierarchy:
✦ center
✦ inner
✦ boundary
✦ outer
Meaning:Controls expansion and structural organization.
5.3 Role Assignment
Defines functional behavior:
✦ anchor
✦ connector
✦ tension_line
✦ loop
✦ return
Meaning:Transforms stitches into grammatical units.
6. Case Study
Case: Canvas 206 — Multi-Layer Structure
Characteristics:
✦ 5-layer system
✦ alternating front/back execution
✦ multiple return and cross events
Structural Observations
1. Repeated return patterns→ Indicates controlled path recycling
2. Cross structures→ Reveal tension distribution points
3. Region transitions→ Center → boundary expansion
Grammar Interpretation
Canvas 206 can be described as:
A multi-layer recursive system with tension-controlled crossings and distributed anchoring.
7. From Embroidery to Language
Mapping:
✦ Stitch → Token
✦ Path → Sequence
✦ Structure → Grammar
This enables:
✦ Pattern recognition
✦ Generative modeling
✦ AI training
8. Applications
✦ AI training datasets
✦ Computational design systems
✦ Structural analysis
✦ Generative embroidery models
9. Future Work
Dataset v4
✦ Path grouping (sentence-level)
✦ Grammar patterns (zigzag, radial)
✦ Tension flow fields
System Expansion
✦ Real-time grammar visualization
✦ AI-assisted pattern generation
✦ Cross-domain application (architecture, materials)
10. Conclusion
Yunbroidery establishes embroidery as a computational language system.
By formalizing structure, path, and tension into a unified framework, it bridges traditional craft and artificial intelligence.
This work lays the foundation for a new field:Embroidery as Computation.