Skip to content

ComfyUI-TCD-scheduler

Add trajectory consistency distillation sampling to ComfyUI

Visit official site

What ComfyUI-TCD-scheduler is

The plugin is described as a custom sampler node set for ComfyUI that implements the trajectory consistency distillation sampling method proposed by Zheng and others. It adds TCDScheduler and SamplerTCD nodes under the Custom Sampler category, and is enabled by cloning it into the custom_nodes folder and restarting ComfyUI.

What you can do with it

  • Implement the trajectory consistency distillation sampling algorithm
  • Add custom sampler nodes to ComfyUI
  • Control sampling randomness with the eta parameter
  • Produce smoother and more consistent output from AI models

Sampler parameters

  • eta controls randomness at each sampling step
  • eta set to 0 gives deterministic sampling
  • eta set to 1 gives fully random sampling
  • The default value is 0.3, and higher eta is suggested when inference steps increase

Example uses

  • Use the plugin during text generation in ComfyUI for smoother sentences
  • Combine it with Stable Diffusion and similar models for more coherent images
  • Adjust eta to balance diversity and consistency of output

Pros & cons

✓ What we like

  • Custom sampler nodes are added under the Custom Sampler category
  • The eta value lets sampling randomness be tuned step by step

! What to watch out for

  • The plugin works as custom nodes inside ComfyUI
  • Sampling randomness is controlled through the eta value of TCDScheduler

FAQ

Which nodes does it add?

The material names TCDScheduler and SamplerTCD under the Custom Sampler category.

What does eta do?

It controls randomness at each step: 0 is deterministic, 1 is fully random, and the default is 0.3.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review