Skip to content

TryOnDiffusion

Virtual try-on that keeps garment detail across poses

Visit official site

What TryOnDiffusion is

The technique is a diffusion-based try-on method that keeps garment detail while adapting to pose and shape changes.

What you can do with it

  • Warp a garment in one unified process
  • Process images at two listed resolutions

Getting started

  • Prepare a person image and a garment image
  • Segment the person image, then feed the 128x128 Parallel-UNet

Pros & cons

✓ What we like

  • Warping and blending share one process
  • Processing runs at 128x128 and 256x256

! What to watch out for

  • It starts from a person and a garment image
  • Pose embeddings are fused by linear layers

FAQ

Which network?

Two UNets combined as Parallel-UNet in one network.

Which resolutions?

The feature list names 128x128 and 256x256.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review