ITVTON: Virtual Try-On Diffusion Transformer Based on Integrated Image and Text

TL;DR

ITVTON introduces an efficient diffusion transformer framework for virtual try-on that combines image and text inputs, achieving high realism with reduced computational complexity and superior performance over existing methods.

Contribution

The paper presents ITVTON, a novel diffusion transformer-based virtual try-on model that simplifies architecture and enhances realism by integrating image and text features within a single generator.

Findings

Outperforms baseline methods in quality and realism.

Reduces computational cost by focusing on attention parameters.

Demonstrates robustness on large real-world dataset.

Abstract

Virtual try-on, which aims to seamlessly fit garments onto person images, has recently seen significant progress with diffusion-based models. However, existing methods commonly resort to duplicated backbones or additional image encoders to extract garment features, which increases computational overhead and network complexity. In this paper, we propose ITVTON, an efficient framework that leverages the Diffusion Transformer (DiT) as its single generator to improve image fidelity. By concatenating garment and person images along the width dimension and incorporating textual descriptions from both, ITVTON effectively captures garment-person interactions while preserving realism. To further reduce computational cost, we restrict training to the attention parameters within a single Diffusion Transformer (Single-DiT) block. Extensive experiments demonstrate that ITVTON surpasses baseline methods both qualitatively and quantitatively, setting a new standard for virtual try-on. Moreover, experiments on 10,257 image pairs from IGPair confirm its robustness in real-world scenarios.