Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GO-MLVTON: Garment Occlusion-Aware Multi-Layer Virtual Try-On with Diffusion Models

Yang Yu, Yunze Deng, Yige Zhang, Yanjie Xiao, Youkun Ou, Wenhao Hu, Mingchao Li, Bin Feng, Wenyu Liu, Dandan Zheng, Jingdong Chen

TL;DR

GO-MLVTON addresses multi-layer garment try-on by explicitly modeling occlusion between inner and outer garments using a Garment Occlusion Learning module and a StableDiffusion-based Garment Morphing & Fitting module. It introduces the MLG dataset and the Layered Appearance Coherence Difference (LACD) metric to evaluate layering realism, and demonstrates state-of-the-art performance on ML-VTON tasks. The approach combines occlusion-aware feature refinement with diffusion-based garment fitting to produce high-fidelity, realistically layered try-ons. This work advances practical multi-layer virtual try-on by enabling end-to-end, occlusion-aware dressing with diffusion-based fitting and provides a valuable dataset and metric for the ML-VTON community.

Abstract

Existing Image-based virtual try-on (VTON) methods primarily focus on single-layer or multi-garment VTON, neglecting multi-layer VTON (ML-VTON), which involves dressing multiple layers of garments onto the human body with realistic deformation and layering to generate visually plausible outcomes. The main challenge lies in accurately modeling occlusion relationships between inner and outer garments to reduce interference from redundant inner garment features. To address this, we propose GO-MLVTON, the first multi-layer VTON method, introducing the Garment Occlusion Learning module to learn occlusion relationships and the StableDiffusion-based Garment Morphing & Fitting module to deform and fit garments onto the human body, producing high-quality multi-layer try-on results. Additionally, we present the MLG dataset for this task and propose a new metric named Layered Appearance Coherence Difference (LACD) for evaluation. Extensive experiments demonstrate the state-of-the-art performance of GO-MLVTON. Project page: https://upyuyang.github.io/go-mlvton/.

GO-MLVTON: Garment Occlusion-Aware Multi-Layer Virtual Try-On with Diffusion Models

TL;DR

GO-MLVTON addresses multi-layer garment try-on by explicitly modeling occlusion between inner and outer garments using a Garment Occlusion Learning module and a StableDiffusion-based Garment Morphing & Fitting module. It introduces the MLG dataset and the Layered Appearance Coherence Difference (LACD) metric to evaluate layering realism, and demonstrates state-of-the-art performance on ML-VTON tasks. The approach combines occlusion-aware feature refinement with diffusion-based garment fitting to produce high-fidelity, realistically layered try-ons. This work advances practical multi-layer virtual try-on by enabling end-to-end, occlusion-aware dressing with diffusion-based fitting and provides a valuable dataset and metric for the ML-VTON community.

Abstract

Existing Image-based virtual try-on (VTON) methods primarily focus on single-layer or multi-garment VTON, neglecting multi-layer VTON (ML-VTON), which involves dressing multiple layers of garments onto the human body with realistic deformation and layering to generate visually plausible outcomes. The main challenge lies in accurately modeling occlusion relationships between inner and outer garments to reduce interference from redundant inner garment features. To address this, we propose GO-MLVTON, the first multi-layer VTON method, introducing the Garment Occlusion Learning module to learn occlusion relationships and the StableDiffusion-based Garment Morphing & Fitting module to deform and fit garments onto the human body, producing high-quality multi-layer try-on results. Additionally, we present the MLG dataset for this task and propose a new metric named Layered Appearance Coherence Difference (LACD) for evaluation. Extensive experiments demonstrate the state-of-the-art performance of GO-MLVTON. Project page: https://upyuyang.github.io/go-mlvton/.
Paper Structure (14 sections, 4 equations, 3 figures, 3 tables)

This paper contains 14 sections, 4 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. MLG Dataset And Layered Appearance Coherence Difference
  4. Method
  5. Garment Occlusion Learning Module
  6. Garment Morphing & Fitting Module
  7. Experiments
  8. Datasets and Implementation Details
  9. Evaluation Metrics and Comparing Settings
  10. Qualitative Results
  11. Quantitative Results
  12. Ablation Studies
  13. Conclusion
  14. Acknowledgement

Figures (3)

  • Figure 1: Examples from the MLG dataset.
  • Figure 2: (a) Pipeline of GO-MLVTON. (b) Details of the Garment Occlusion Learning module.
  • Figure 3: Qualitative Results. Red boxes indicate areas that exhibit visually unrealistic effects or details misalignment.