Skip to content

Trajectory Consistency Distillation (TCD)

A consistency distillation method for text to image

Visit official site

What TCD is

TCD is described as a consistency distillation technique for text to image synthesis that cuts errors during generation with TCF and SSS.

What you can do with it

  • Apply a trajectory consistency function
  • Use strategic stochastic sampling
  • Keep image quality at different NFE settings

Application scenarios

  • An artist turns a description into artwork
  • A game developer draws characters and settings
  • A researcher compares models at different NFE

Pros & cons

✓ What we like

  • Adapts to various SDXL base models and community plugins
  • No extra discriminator or LPIPS supervision is stated as needed

! What to watch out for

  • The technique is framed around SDXL base models
  • Quality is described across NFE settings

FAQ

What are the two components?

The material names a trajectory consistency function and strategic stochastic sampling.

Which models are named?

The material names SDXL base models and community plugins.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review