GarmentX: Autoregressive Parametric Representations for High-Fidelity 3D Garment Generation

2025-04-29

AI Summary

Traditional single-image garment reconstruction methods directly predict 2D pattern outlines and connectivity, often yielding self-intersecting or physically infeasible structures. To address this, we propose the first parametric framework for wearable 3D garment generation. Our method (1) introduces a GarmentCode-compatible, editable parametric representation encoding topology, geometry, and sewing constraints; (2) employs masked autoregressive modeling to explicitly enforce topological validity and geometric consistency across pattern pieces; and (3) constructs GarmentX—a large-scale parametric-image paired dataset (378K samples)—alongside a scalable synthetic data generation pipeline. Experiments demonstrate state-of-the-art performance in geometric fidelity and image alignment. Generated garments are guaranteed self-intersection-free, compatible with physics-based simulation, and support intuitive parametric editing. Our approach significantly improves diversity, physical plausibility, and editability of 3D garments.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Web Mining and Content Analysis: Web data generation and simulation

Abstract

This work presents GarmentX, a novel framework for generating diverse, high-fidelity, and wearable 3D garments from a single input image. Traditional garment reconstruction methods directly predict 2D pattern edges and their connectivity, an overly unconstrained approach that often leads to severe self-intersections and physically implausible garment structures. In contrast, GarmentX introduces a structured and editable parametric representation compatible with GarmentCode, ensuring that the decoded sewing patterns always form valid, simulation-ready 3D garments while allowing for intuitive modifications of garment shape and style. To achieve this, we employ a masked autoregressive model that sequentially predicts garment parameters, leveraging autoregressive modeling for structured generation while mitigating inconsistencies in direct pattern prediction. Additionally, we introduce GarmentX dataset, a large-scale dataset of 378,682 garment parameter-image pairs, constructed through an automatic data generation pipeline that synthesizes diverse and high-quality garment images conditioned on parametric garment representations. Through integrating our method with GarmentX dataset, we achieve state-of-the-art performance in geometric fidelity and input image alignment, significantly outperforming prior approaches. We will release GarmentX dataset upon publication.

Problem

• Generating high-fidelity 3D garments from single images
• Overcoming self-intersections in traditional garment reconstruction
• Creating editable parametric representations for valid sewing patterns

Innovation

• Autoregressive model for sequential garment parameter prediction
• Editable parametric representation ensuring valid garment structures
• Large-scale dataset with automatic garment-image synthesis