Save 4 hours! We auto-setup your workflow! Free!

Drop your workflow.json — we handle every dependency, custom node, and model. Just open the link and run.

Auto-Setup Workflow Json (Free) Now!
ComfyUI > Nodes > sweet-tea-nodes

ComfyUI Extension: sweet-tea-nodes

Repo Name

sweet-tea-nodes

Author
tea-time-labs (Account age: 335 days)
Nodes
View all nodes(1)
Latest Updated
2026-11-04
Github Stars
0.0K

How to Install sweet-tea-nodes

Install this extension via the ComfyUI Manager by searching for sweet-tea-nodes
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter sweet-tea-nodes 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

sweet-tea-nodes Description

sweet-tea-nodes adds ComfyUI nodes for seamless tiled image upscaling, video transitions between known start and end frames, guided subject swaps, temporary previews, and workflow metadata.

sweet-tea-nodes Introduction

sweet-tea-nodes is a collection of custom nodes for ComfyUI. It focuses on two creative tasks: generating video with more control over transitions and subjects, and enlarging images in tiles while reducing visible joins between those tiles. It also includes preview and metadata nodes that help you inspect results and keep track of how they were made.

A node is a step in a ComfyUI workflow. You connect nodes to describe what should happen to your images or video. With sweet-tea-nodes, you can build workflows for tasks such as reconstructing the middle of a clip between known beginning and ending frames, changing a subject while retaining structural guidance, or upscaling a large image without obvious tile boundaries.

How sweet-tea-nodes Works

The nodes each handle a particular part of a workflow:

  • Tiled upscaling divides an image into smaller areas, enlarges them, and blends them back together. Think of restoring a large mural one section at a time: the sections need to line up, and their edges should not reveal where one section ends and another begins.
  • Video cap stitching uses known content at the start and end of a clip as anchors. An inpaint mask identifies the middle region to generate, so the workflow concentrates on reconstructing the transition between those anchors.
  • Hybrid subject swapping combines structural guidance, a character reference, and optional information from the source video. These inputs give the workflow different kinds of guidance about what to change and what visual qualities to retain.
  • Preview nodes make a video available for viewing without creating a permanent ComfyUI output copy. A separate preview node can first turn a sequence of images into a temporary video.

These nodes provide controls and outputs for ComfyUI workflows; the result still depends on the other components and inputs you connect to them.

sweet-tea-nodes Features

High-fidelity tiled upscaling

Ultimate SD Upscale (Pose Tiled ControlNet) (UltimateSDUpscalePoseTiled) is designed for upscaling in tiles. It includes lockable padding, improvements to ControlNet tile alignment, normalized overlap blending, and seam-fix passes.

These controls address artifacts that can appear when separately processed tiles are reassembled, including visible seams, color or detail bleeding across tile edges, and offset or ghosted details. Padding gives the node surrounding image context while it works on a tile; overlap blending and seam fixes help neighboring tiles meet more naturally.

Video generation between known start and end caps

Wan VACE -> Video (caps inpaint) (WanVaceToVideoCapsInpaint) supports clip-stitching and transition-reconstruction workflows. It treats the start and end caps as known regions and marks the middle as the region to generate. The node supports variable clip lengths and cap sizes.

For example, if you have a beginning shot and an ending shot, you can use them as anchors for a generated transition. Changing the cap sizes changes how much of the clip is treated as known content at either end; changing the overall length changes the span of video the workflow must cover.

Hybrid subject swapping

Wan VACE -> Video (hybrid subject swap) (WanVaceToVideoHybridSubjectSwap) is a depth-first subject-swap node for Wan VACE workflows. It combines structural control with a character reference. You can also supply optional source-video context for brightness, detail, and background.

This is useful when you want to guide a new subject’s appearance while retaining aspects of the original scene. The optional contexts let you decide which source qualities to provide as guidance, including when working across illustrated and live-action looks.

Temporary video previews

Sweet Tea Preview Video (SweetTeaPreviewVideo) accepts ComfyUI’s native VIDEO type and acts as a preview output without making a permanent ComfyUI video copy. With Auto selected for both container and codec, it preserves an existing encoded stream. Selecting a specific container or codec instead creates a temporary output in the requested format.

Sweet Tea Preview Video From Images (SweetTeaPreviewVideoFromImages) turns an IMAGE sequence into one temporary video. You can specify its frame rate, container, codec, quality, bit depth and color space, and optionally include audio. These choices affect how the image sequence is presented and encoded. The temporary video lets Sweet Tea Studio retain the visible result without requiring a duplicate ComfyUI output file.

Image-preview metadata

SweetTeaPreviewImage supports the sweet_tea_output_metadata_v1 client feature. For identified frames, it can provide the prompt, node, list invocation, item position and count, and a stable artifact ID. This helps a compatible client associate a preview with the workflow result that produced it.

Older clients still receive the standard ComfyUI image preview. This preview does not create a permanent ComfyUI image copy.

Execution receipts

Sweet Tea Execution Receipt (SweetTeaExecutionReceipt) is a no-save metadata output for supported external API execution facts. It can publish the provider, request, endpoint, operation, and an optional estimated cost for Sweet Tea Studio provenance. It is designed to report these details without exposing API credentials.

Troubleshooting sweet-tea-nodes

Why can’t I find a permanent output file after using a Sweet Tea preview node?
The preview nodes are designed to avoid a permanent ComfyUI output copy. If you need a lasting file, use an appropriate save step elsewhere in your workflow rather than relying on a preview output.

Why does the video preview create a temporary file when I choose a container or codec?
An explicit format choice asks the node to produce a temporary output in that format. If you want to preserve an already encoded stream, use Auto for both container and codec.

Why is image-preview metadata missing in my client?
The additional frame details depend on support for sweet_tea_output_metadata_v1. Older clients continue to receive the normal ComfyUI preview, but may not show those extra details.

Can the caps-inpaint node work with different clip lengths?
Yes. WanVaceToVideoCapsInpaint supports variable lengths and cap sizes. Check that the start and end caps represent the content you want to keep, and that the middle is the region you intend to generate.

Learn More about sweet-tea-nodes

The sweet-tea-nodes repository is the source for the node pack. For a closer look at how a feature is implemented, its code is organized into areas for tiled upscaling, Wan VACE caps inpainting, and Wan VACE hybrid subject swapping.

sweet-tea-nodes Related Nodes

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 Models, enabling artists to harness the latest AI tools to create incredible art.

sweet-tea-nodes detailed guide | ComfyUI