Layering Virtual Try-On

2026-07-24

AI Summary

Existing virtual try-on methods struggle to handle multi-layered clothing and dressing/undressing sequences in real-world scenarios, typically supporting only single-garment replacement. This work proposes the Layering Virtual Try-On (LVTON) framework, which for the first time formulates and establishes a benchmark for multi-layer virtual try-on. LVTON decouples general try-on priors from layering logic, enabling sequential addition or removal of garments while preserving existing attire. The method leverages an automated data generation pipeline to synthesize training samples from fashion videos and employs image segmentation and inpainting techniques for pretraining, followed by fine-tuning on a small-scale LVTON dataset to learn layering order and occlusion relationships. Experiments demonstrate that LVTON achieves state-of-the-art performance on the new benchmark, exhibits strong generalization on conventional VTON tasks, and shows promising zero-shot transfer capability.

Abstract

In the real world, fashion is about layering: adding a jacket over a shirt, or a sequence of adding and removing layers, rather than just a single-layer swap. This fundamental real-world task remains a challenge in existing Virtual Try-On (VTON) methods, which excel at single-layer replacement but are not designed to layer or de-layer an existing outfit. This paper proposes Layering Virtual Try-On (LVTON), a layering benchmark and method that preserves an existing outfit while enabling sequential layering. We find that current VTON paradigms are fundamentally ill-equipped for LVTON, as their reliance on cloth-agnostic representations and single-item datasets discards essential layering context. Our key insight is that the LVTON challenge must be disentangled into two distinct competencies: (1) General VTON Priors (e.g., deformation, identity preservation) and (2) Specific Layering Knowledge (e.g., layering order and occlusion reasoning). First, our model obtains general VTON priors by being trained on data produced by an automatic data generation pipeline that synthesizes samples from fashion videos via segmentation and inpainting. Second, the model is fine-tuned on a small, dedicated LVTON dataset to learn the layering logic. Our method achieves state-of-the-art results on our LVTON benchmark and demonstrates superior generalizability on traditional VTON benchmarks, setting new state-of-the-art results when fine-tuned and exhibiting zero-shot capabilities.

Problem

• Virtual Try-On
• Layering
• Clothing Occlusion
• Outfit Composition
• Fashion Video

Innovation

• Layering Virtual Try-On
• Clothing Layering
• Occlusion Reasoning
• Data Synthesis
• Disentangled Representation