FDM-MFVT: Few-step Sampling Diffusion Model for Mask-Free Virtual Try-On

2026-06-28

AI Summary

Existing virtual try-on methods rely on costly segmentation masks and auxiliary networks, involve complex multi-step sampling procedures, and suffer from a lack of large-scale paired data without masks, which hinders the advancement of mask-free try-on. This work proposes an efficient few-step sampling diffusion model that leverages a garment-aware noise optimization module to initialize a well-aligned latent space and integrates an instruction-driven dressing mechanism to generate high-fidelity try-on results in just six sampling steps. We introduce MFVT, the first mask-free image-based virtual try-on (IVTON) dataset comprising 30,000 paired samples. Without using any masks, our method significantly outperforms both existing mask-based and mask-free approaches, achieving simultaneous improvements in computational efficiency and image quality.

Abstract

Image-based Virtual Try-On (IVTON) has greatly advanced through diffusion models, yet existing methods require many sampling steps and depend on masks with costly auxiliary networks. In addition, the absence of large-scale mask-free paired datasets further limits the development of mask-free IVTON. We propose FDM-MFVT, a few-step diffusion model for mask-free IVTON, integrating an Outfit-aware Noise Optimization Module (OANO) and an Instruction-driven Try-on Module (IDT) to enhance efficiency and flexibility.The OANO module initializes the alignment space with noise using the input image and only needs 6 steps to generate a higher-fidelity try-on image compared to 30 steps.The IDT module uses virtual try-on prompts and efficient adaptation to generate high-quality results from garment and person images alone. We further introduce MFVT, a 30,000-pair mask-free IVTON dataset. Experiments show that FDM-MFVT achieves superior quantitative and qualitative results with fewer inference steps than mask-based and mask-free baseline methods.

Problem

• Virtual Try-On
• Diffusion Model
• Mask-Free
• Few-step Sampling
• Paired Dataset

Innovation

• Few-step diffusion
• Mask-free virtual try-on
• Noise optimization
• Instruction-driven generation
• Outfit-aware alignment