Skip to content

Fashion-VDM

Generate virtual try-on video from a garment image and a person video

Visit official site

What Fashion-VDM is

The model is described as a video diffusion model that generates a try-on video from a garment image and a person video while keeping the person identity and motion. The material names clothing detail and temporal consistency as strengths compared with image-based virtual try-on.

What you can do with it

  • Generate a high-quality try-on video from a garment image and a person video
  • Keep the original identity and motion of the person during try-on
  • Apply classifier-free guidance to strengthen control over conditioning inputs
  • Maintain temporal coherence with 3D-Conv and temporal attention blocks

Application scenarios

  • A fashion designer prepares personalised try-on videos for clients
  • An e-commerce platform offers an online fitting room service
  • A clothing brand shows try-on results of new garments on social media

Getting started

  • Prepare a garment image and a person video as inputs
  • Upload the inputs to the model
  • Choose the try-on effect and parameters you want
  • Start the model to generate the virtual try-on video

Pros & cons

✓ What we like

  • The person identity and motion are kept during try-on
  • Clothing detail and temporal consistency are named as strengths over image-based try-on

! What to watch out for

  • Generation depends on both a garment image and a person video as inputs
  • Progressive temporal training is described for a single pass of 64 frames at 512 px

FAQ

What are the inputs?

The material states one garment image and a video of a person.

How many frames are generated in one pass?

The progressive temporal training strategy is described as a single pass of 64 frames at 512 px.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review