ComfyUI > Nodes > ComfyUI-HunyuanVideoWrapper > HunyuanVideo TeaCache

ComfyUI Node: HunyuanVideo TeaCache

Class Name

HyVideoTeaCache

Category
HunyuanVideoWrapper
Author
kijai (Account age: 2506days)
Extension
ComfyUI-HunyuanVideoWrapper
Latest Updated
2025-05-12
Github Stars
2.4K

How to Install ComfyUI-HunyuanVideoWrapper

Install this extension via the ComfyUI Manager by searching for ComfyUI-HunyuanVideoWrapper
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter ComfyUI-HunyuanVideoWrapper in the search bar
After installation, click the Restart button to restart ComfyUI. Then, manually refresh your browser to clear the cache and access the updated list of nodes.

Visit ComfyUI Online for ready-to-use ComfyUI environment

  • Free trial available
  • 16GB VRAM to 80GB VRAM GPU machines
  • 400+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 200+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

HunyuanVideo TeaCache Description

Optimize HunyuanVideo framework performance with TeaCache for faster video processing.

HunyuanVideo TeaCache:

The HyVideoTeaCache node is designed to optimize the performance of the HunyuanVideo framework by implementing a caching mechanism known as TeaCache. This node is particularly beneficial for speeding up the inference process in video processing tasks. By intelligently caching intermediate data, it reduces the computational load and accelerates the processing time, making it an essential tool for users looking to enhance the efficiency of their video workflows. The node allows for fine-tuning of caching parameters, enabling users to balance between speed and potential artifacts, thus providing a customizable approach to video processing optimization.

HunyuanVideo TeaCache Input Parameters:

rel_l1_thresh

The rel_l1_thresh parameter is a floating-point value that determines the aggressiveness of the TeaCache mechanism. A higher value makes the caching more aggressive, which can significantly speed up the processing but may introduce artifacts in the video output. The parameter ranges from 0.0 to 1.0, with a default value of 0.15. Adjusting this threshold allows you to control the trade-off between speed and quality, making it a crucial setting for optimizing performance based on your specific needs.

cache_device

The cache_device parameter specifies the device where the cache will be stored. It offers two options: main_device and offload_device, with the default being offload_device. This setting is important for managing resource allocation, as it allows you to choose whether to use the primary device for caching or to offload the cache to a secondary device, potentially freeing up resources on the main device for other tasks.

start_step

The start_step parameter is an integer that defines the step at which the TeaCache mechanism begins to apply. It ranges from 0 to 100, with a default value of 0. This parameter is useful for controlling when the caching process should start during the video processing workflow, allowing for strategic application of caching to optimize performance at specific stages.

end_step

The end_step parameter is an integer that indicates the step at which the TeaCache mechanism stops applying. It ranges from -1 to 100, with a default value of -1, which typically means that the caching will continue until the end of the process. This parameter provides flexibility in determining the duration of caching, enabling you to fine-tune the caching period to suit your processing requirements.

HunyuanVideo TeaCache Output Parameters:

teacache_args

The teacache_args output parameter is a dictionary containing the configured settings for the TeaCache mechanism. This includes the relative L1 threshold, the selected cache device, and the start and end steps for caching. This output is crucial as it encapsulates all the necessary parameters that define how the caching will be executed, allowing for easy integration and application within the HunyuanVideo framework.

HunyuanVideo TeaCache Usage Tips:

  • To maximize performance, consider setting the rel_l1_thresh to a higher value if speed is a priority and minor artifacts are acceptable. Conversely, lower the threshold if maintaining video quality is more important.
  • Use the cache_device setting to offload caching to a secondary device if your main device is under heavy load, ensuring smoother operation and resource management.

HunyuanVideo TeaCache Common Errors and Solutions:

"TeaCache skipped steps"

  • Explanation: This message indicates that certain steps were skipped by the TeaCache mechanism, which might be due to the configuration of start and end steps or the aggressiveness of the caching.
  • Solution: Review the start_step and end_step parameters to ensure they are set correctly for your workflow. Adjust the rel_l1_thresh if necessary to modify the caching behavior.

"Device not found for caching"

  • Explanation: This error occurs when the specified cache_device is not available or incorrectly configured.
  • Solution: Verify that the device specified in the cache_device parameter is correctly set up and available for use. Ensure that the device names match those recognized by your system.

HunyuanVideo TeaCache Related Nodes

Go back to the extension to check out more related nodes.
ComfyUI-HunyuanVideoWrapper
RunComfy
Copyright 2025 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Playground, enabling artists to harness the latest AI tools to create incredible art.