Fashion Technology · Research & Prototyping
MODA
GPT
Measurement Optimized Design AI
Where the dressmaker’s geometric intuition meets the computational precision of artificial intelligence.
The Name · The Meaning · The Mission
“Every garment begins with the body. Measurement is not preparation — it is the first act of design.”
ModaGPT’s foundation is the validated, sub-millimeter-accurate extraction of body landmarks from any human geometry — digital avatar, 3D scan, or multi-view photograph. No pattern can fit what has not been precisely understood.
MODA·SCAN“Not approximated. Not estimated. Optimized — computed from the geometry of the body itself, within the tolerances production demands.”
Where traditional CAD applies fixed grade rules, ModaGPT applies differential geometry — geodesic paths, Gaussian curvature, surface parameterization — so that every measurement and every pattern reflects how bodies actually are, not how tables assume them to be.
MODA·PATTERN“Design is intent made physical. ModaGPT reads that intent — from a sketch, a reference, a mood — and translates it into the language of construction.”
From image or description, ModaGPT identifies garment components, classifies fabric and fabric behavior, specifies notions, and produces a technical specification ready for the pattern table — bridging the gap between creative vision and production reality.
MODA·DESIGN“Not AI as spectacle. AI as the formalization of knowledge that has lived in human hands for centuries — finally given a language precise enough to be computed.”
ModaGPT trains on the physical reality of garments sewn and scanned — every pattern cut, every fitting, every correction becomes data. The system learns from every garment it sees, growing more accurate with each physical validation cycle.
MODA·VALIDATEMeasurement · Optimized · Design · Artificial Intelligence
The computational formalization of what master dressmakers have always known.
What is ModaGPT
The computational formalization of what master dressmakers have always known.
ModaGPT is a fashion-technology platform and research-driven initiative investigating how artificial intelligence can support apparel creation workflows — from accurate body measurement extraction to geometry-driven pattern generation and design interpretation. It is built on the premise that fit is a geometric problem, and that solving it requires the relationship between body surface and fabric plane to be understood with mathematical precision.
Commercial 3D garment platforms digitized the ruler — they did not replace the dressmaker’s geometric intuition. Pattern making knowledge still lives entirely in human experts. The software shows what happens after a pattern is constructed, not how to construct it correctly. ModaGPT is designed to close that gap.
Every fitting tradition — from Savile Row to Haute Couture — begins with the body. ModaGPT’s first principle is that accurate, validated body measurements are the non-negotiable foundation of any garment that fits. The platform establishes measurement accuracy at sub-millimeter precision before any pattern is generated.
Traditional CAD grading applies fixed incremental rules per size step. ModaGPT replaces this with differential geometry computed from the actual body surface — geodesic paths, Gaussian curvature mapping, and surface parameterization — producing patterns that reflect how bodies actually change across sizes.
A first pattern that fits collapses the sample loop. ModaGPT’s long-term goal is made-to-measure at scale — each garment cut to the wearer’s actual measurements, eliminating the size-run overproduction responsible for an estimated 30% of fashion industry waste. AI not as decoration, but as transformation.
Core Pillars
Three systems. One coherent pipeline.
01 · MODA·SCAN
Human Geometry
Body measurements, proportions, and anatomical landmarks as structured, validated data. From multi-view images or 3D scans to a measurement set within production tolerance.
02 · MODA·DESIGN
Design Intelligence
Interpretation of design intent from text and visual references within an apparel context. From sketch or image to material specification, notion list, and construction method.
03 · MODA·PATTERN
Garment Geometry
Parametric pattern logic and garment representations prepared for digital workflows. Geometry-driven pattern generation — darts, seams, ease, grain — derived from body surface mathematics.
The Pipeline
From body to pattern, without the guesswork.
1 Input
Body Capture
Multi-view photographs, 3D scan, or Exported Avatar Geometry (EAG) file. Any representation of the human form becomes the starting point.
2 MODA·SCAN
Measurement Extraction
SMPL-X fitting and geodesic computation produce a validated measurement set. Every landmark verified against production tolerance.
3 MODA·DESIGN + PATTERN
Pattern Generation
Design intent interpreted from image or description. Geometry-driven pattern pieces generated from body surface mathematics, not inherited rules.
4 MODA·VALIDATE
Physical Validation
Sewn garment scanned on instrumented mannequin. Prediction vs. reality feeds back into the model — the system learns from every garment made.
Research Foundation
Validated. Quantified. Production-honest.
IAM ↔ SME Agreement · Male Avatar
All 25 measurements match to <0.1%. The Standard Measurement Export (SME) is a direct rounded output of the platform’s Internal Avatar Measurements (IAM) — the validated reference standard is accessible and reliable.
Mean IAM Precision
Internal Avatar Measurements (IAM) confirmed as validated reference standard. Production-grade precision established before any pattern is drafted.
Female EAG Mean Geometric Error
Female Exported Avatar Geometry (EAG) validated: Hip ±0.76%, Calf ±0.14%, Bust ±1.47%, Wrist ±1.23% — all within production tolerance.
EAG Neck Geometry Limitation · Cross-Platform
Hair mesh contamination above 150 cm makes convex hull neck measurement unreliable. Identified as a platform-wide limitation requiring mesh segmentation.
Female EAG Waist Overestimation
Exceeds the ±1 cm finished garment tolerance. Fitted bodice patterns require physical validation before grading — a tiered trust framework is established.
Presented at FAIREE 2026 · AI in Fashion
This research was developed for the FAIREE 2026 Symposium on AI in Fashion, establishing the first quantified validation framework for Internal Avatar Measurement (IAM) accuracy and Exported Avatar Geometry (EAG) fidelity across commercial 3D garment platforms.
No commercial 3D garment platform derives the pattern from the body geometry. The designer still applies traditional pattern making knowledge manually. The simulation only shows what happens after the pattern is constructed — not how to construct it correctly. ModaGPT is designed to close that gap.
— ModaGPT Research Foundation
Tiered Trust Framework
Tier 1 — Trust Digitally
Height, Hip, Waist (Adam), Calf, Thigh. All within ±1 cm. Pattern directly to first physical sample.
Tier 2 — Verify with Toile
Bust (Tina), Underbust, Knee, Ankle. One physical check at base size before grading.
Tier 3 — Always Physical
Neck / collar. Geometry unreliable in both platforms. Physical fitting mandatory regardless of simulation quality.
Scope of Work
Research, prototyping, and educational applications.
ModaGPT is developed to support measurement-informed garment development, digital pattern and 3D garment workflows, data structuring for apparel design systems, and preparation of geometry outputs for downstream visualization and fabrication tools.
Research
Doctoral-level investigation into the measurement accuracy of commercial 3D fashion platforms, the reliability of digital-to-physical pattern transfer, and the computational formalization of garment geometry.
Prototyping
Active development of MODA·SCAN, MODA·PATTERN, MODA·DESIGN, and MODA·VALIDATE as interconnected system prototypes. Each component independently deployable and collectively forming the full pipeline.
Education
Curriculum and methodology development for fashion technology programs. Bridging the gap between traditional pattern making pedagogy and computational design methods for the next generation of fashion technologists.