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 > ComfyUI_Costume-Gallery_ZH

ComfyUI Extension: ComfyUI_Costume-Gallery_ZH

Repo Name

ComfyUI_Costume-Gallery_ZH

Author
Stone-dielianhua (Account age: 2555 days)
Nodes
View all nodes(2)
Latest Updated
2026-11-05
Github Stars
0.08K

How to Install ComfyUI_Costume-Gallery_ZH

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

ComfyUI_Costume-Gallery_ZH Description

ComfyUI_Costume-Gallery_ZH provides a searchable offline gallery of outfit references in ComfyUI, with visual browsing, favorites, descriptive text, and single-image or batch output.

ComfyUI_Costume-Gallery_ZH brings a collection of 1,407 outfit reference images into ComfyUI. Instead of searching through folders and manually loading references, you can browse a visual gallery, filter outfits, and send one or several images directly to your workflow.

The extension appears in the Chinese interface as 服装图库 (“Costume Gallery”). It provides both images and descriptive text, making it useful when you want to:

  • Find clothing references by season, occasion, garment type, or color combination.
  • Explore several outfit ideas before choosing a direction for your artwork.
  • Use a single outfit as a visual reference in an image-editing or generation workflow.
  • Prepare a batch of outfit references for comparison or further processing.
  • Retrieve the original English keywords and Chinese descriptions associated with an outfit.
  • Keep frequently used outfits in favorites and revisit recently used references. The gallery is a reference-selection tool, not an automatic outfit-changing model. It supplies images and accompanying information. Any character generation, clothing replacement, or consistency between images depends on the models and workflow you connect afterward.

Everyday browsing and image output use local files and work offline once the assets are available. No model download is required to use the gallery itself.

The extension is independent of ComfyUI_Pose-Gallery_ZH. You can use both together; their nodes, sidebar entries, favorites, recent items, and asset settings are stored separately.

Expanded Costume Gallery window in ComfyUI

Think of the extension as a local wardrobe catalog connected to your ComfyUI canvas.

Each outfit has a catalog entry containing an ID, image paths, keywords, a Chinese description, and tags. The gallery uses this information to help you search and browse. When you choose an outfit, a node loads its local image and makes it available to the rest of your workflow.

1. Find an outfit

Open Costume Gallery (服装图库) in the sidebar. Narrow the collection with filters or search terms.

For example, you can search for:

  • SPR-0001 — a specific outfit ID.
  • 白色 连衣裙 — Chinese terms for “white” and “dress.”
  • English clothing terms from the original keywords.
  • Multiple keywords to refine your search. Color matching considers colors present anywhere in the complete outfit, rather than only one garment.

2. Send your selection to a node

You can send:

  • One outfit to a single-image node.
  • Several outfits to a batch node. The image output can then connect to a preview node or a workflow that accepts reference images.

3. Use the image and text separately

The single-image node also provides the outfit ID, original English keywords, and Chinese description.

For example, you might connect:

  • The image to your reference-image workflow.
  • The English keywords to a text-encoding node.
  • The Chinese description to a text-preview node for easier review. If a text field does not have an input connection, use ComfyUI’s input-conversion feature to turn it into a connectable input where supported.

The keywords come from the original index. You may need to edit or reorganize them for the model you use; the gallery does not automatically turn them into a model-specific prompt.

4. Let your downstream workflow do the creative work

The extension does not generate a person, apply clothing to a character, or guarantee that an outfit is reproduced exactly. It supplies the reference material. The connected model and workflow determine how that reference influences your final artwork.

Searchable, filterable outfit library

You can browse the built-in collection using:

| Control | What it helps you find | | --- | --- | | Season / occasion filters | Outfits suited to a particular seasonal or situational theme | | Clothing-type filters | Particular garment types | | Color-combination filters | Outfits containing the colors you want | | Search field | Chinese terms, English terms, outfit IDs, or multiple keywords |

For example, if you are planning a character with a white-dress theme, searching 白色 连衣裙 gives you a more focused starting point than browsing the full collection.

Large-image preview

Click an image to inspect it more closely. The preview supports:

  • Left and right arrow-key navigation.
  • Swipe navigation.
  • Adding outfits to favorites.
  • Copying English keywords, Chinese descriptions, or outfit IDs.
  • Downloading the original-size image. Use the upper-right Expand (展开) control to open a larger browsing window. This is useful when you want to compare references without working in the narrow sidebar. Batch selections synchronize when you return to the workflow.

Single-image selection

Open an outfit preview and choose Use this outfit (使用此服装) to create a single-image node.

The node appears as 服装图库 · 单图 and provides:

| Output | Contents | Typical use | | --- | --- | --- | | IMAGE | One outfit reference image | Previewing, image editing, or reference-based generation | | Outfit ID | The outfit’s identifier | Finding the same reference again | | English keywords | Original English keyword text | Preparing prompts or text inputs | | Chinese description | Chinese reference text | Reviewing the outfit’s description |

Single-image output preserves the original image dimensions by default.

Batch selection

Select several images, then choose Send selected images (发送所选图片) at the bottom of the gallery.

The node appears as 服装图库 · 批量 and provides:

| Output | Contents | Typical use | | --- | --- | --- | | IMAGE batch | Multiple outfit images | Comparing references or supplying a batch to another workflow | | Information-list JSON | Structured information corresponding to each image | Matching each output image with its ID, keywords, and processing details |

Your selection order is the default output order. Use View selected (查看已选) to check the selection and remove unwanted images before sending.

Batch output supports:

  • Up to 32 images.
  • A maximum of 32 million output pixels in total. Larger output dimensions leave room for fewer images within that limit. If a batch fails because it is too large, reduce the image count or output dimensions.

Replace images without rebuilding your workflow

The bottom Send to (发送到) control lets you target an existing gallery node instead of creating another one.

You can also click Open Costume Gallery (打开服装图库) on a node to choose a replacement image.

Replacing a selection preserves:

  • Size settings.
  • Horizontal-flip settings.
  • Ordering settings.
  • Existing connections. Choose New gallery node (新建图库节点) when you want a separate node instead.

Important limitations:

  • A single-image node cannot receive a multi-image selection.
  • If a selection input is controlled by a connection, the gallery prompts you to address that connection before changing the selection.

Image sizing and processing

Batch output defaults to 768 × 1152, fitting each image proportionally and adding padding as needed.

| Mode | What it does | When it is useful | | --- | --- | --- | | Proportional fit with padding | Preserves the image’s proportions and adds space to reach the target size | When keeping the full outfit visible matters | | Center crop | Crops around the center to reach the target dimensions | When filling the frame matters more than preserving the edges | | Stretch | Resizes directly to the requested width and height | When exact dimensions matter and distortion is acceptable | | Original size | Keeps the source image dimensions | When you want to retain the original resolution |

For example, fitting a wide reference into a tall canvas preserves the complete image but adds padding. Center cropping may remove content near the sides. Stretching fills the canvas but can change the apparent proportions of clothing.

Additional behavior:

  • Width and height apply only to proportional fit, center crop, and stretch.
  • Original-size mode ignores the width and height settings.
  • Original-size batches require all selected images to have matching dimensions.
  • Horizontal flipping is supported.
  • Transparent areas are composited onto white.
  • Image output is RGB.

Batch ordering and corresponding information

You can arrange a batch by:

  • Selection order

  • ID order

  • Category order The information-list JSON is an array that corresponds item by item to the image batch. Each entry records:

  • A zero-based index: the first image is 0, the second is 1, and so on.

  • Outfit id.

  • Category.

  • Original English and Chinese keywords.

  • Tags.

  • Asset library.

  • Actual output width and height.

  • Processing parameters. This is useful when you need to keep a particular image paired with its description.

A batch does not automatically run text encoding or image generation once per outfit. If each image needs its own prompt or generation step, your downstream workflow must handle the image-to-text correspondence.

Favorites, recent items, and saved workflow settings

Favorites and recent items are stored in the current browser, separately for the built-in and custom libraries.

The workflow retains the selected outfit IDs, asset-library selection, dimensions, and flip settings. These are separate from your browser’s favorites.

This means:

  • Switching browsers does not transfer favorites.
  • Clearing browser site data removes locally stored favorites and recent items.
  • A saved workflow can still retain its outfit selections and node settings.

The author provides two example workflows. Drag a JSON file into ComfyUI and run it. Neither example requires a generation model.

| Example | What it demonstrates | | --- | --- | | 01_single_image.json | Sends one outfit to an image preview and displays its English keywords | | 02_batch_images.json | Sends three outfits to a batch preview and displays their corresponding information JSON |

These examples are a useful first step before adding the gallery to a larger generation workflow. They let you check image loading, text output, and batch behavior separately from any model-related settings.

Text display uses ComfyUI’s built-in PreviewAny, shown as Preview as Text. If your ComfyUI version does not include this node, replace it with a text-display node you already use, or keep only the image preview.

You can point the gallery to your own compatible outfit collection through Asset settings (素材设置).

Choose a library location

Enter an absolute path on the computer running ComfyUI. An absolute path is the full location of a folder, rather than a location relative to another folder.

If you access ComfyUI through a browser on a different computer, the path must still be readable by the computer running ComfyUI—not merely by your browser’s computer.

The directory must contain:

  • data/costume-zh.json
  • Every original image referenced by that file Thumbnail and large-preview files are optional. If they are missing, the gallery falls back to the original image.

Describe your images

The library uses the project’s existing field names, including the Chinese keys:

json [ { "id": "MY-0001", "分类": "Custom", "关键词": "white dress", "关键词中文": "白色 · 连衣裙", "标签": ["白色", "连衣裙"], "原图": "png/MY-0001.png", "缩略图": "thumbs/MY-0001.webp", "大图": "img/MY-0001.webp" } ]

| Field | Meaning | | --- | --- | | id | Unique outfit identifier | | 分类 | Category | | 关键词 | English keywords | | 关键词中文 | Chinese keywords or description | | 标签 | Searchable tags | | 原图 | Original-image path | | 缩略图 | Optional thumbnail path | | 大图 | Optional large-preview path |

Each ID must be unique. Image paths must be relative to the library directory and use / separators.

You can extend search synonyms through the 同义词 field in data/meta-zh.json.

Save and switch libraries

When you save the directory setting, the extension checks that the index and referenced original images exist. If validation fails, it keeps the previous setting.

After saving, switch to Custom asset directory (自定义素材目录) at the top of the gallery.

Understand what is shared

The custom-library setting is saved in the ComfyUI user directory at:

text costume_gallery/settings.json

It is a setting for that ComfyUI instance. Changing the custom directory also changes where existing custom-library nodes look for their images. Before switching directories, check that the outfit IDs still refer to the intended images.

When sharing a workflow:

  • The recipient needs the extension and matching assets.
  • The workflow does not embed the complete asset library.
  • Your local custom-directory path is not included in the workflow.

The sidebar entry or gallery nodes are missing

  1. Restart ComfyUI.
  2. Refresh the browser page.
  3. Look for the sidebar label 服装图库 and node names 服装图库 · 单图 and 服装图库 · 批量.
  4. Check the ComfyUI startup log for whether ComfyUI_Costume-Gallery_ZH loaded successfully.
  5. Check that the extension folder is not nested inside another folder with the same name.

An outfit image is missing or will not load

  1. Confirm whether you are using the built-in or custom library.
  2. Check that the library contains data/costume-zh.json.
  3. Check that the original image referenced by the affected entry exists at the listed relative path.
  4. For a custom library, confirm that the computer running ComfyUI can read the directory. For a full check of the project’s included assets, the repository provides:

sh python scripts/verify.py

Missing thumbnails or large previews should fall back to original images; missing original images need to be resolved.

A batch fails in original-size mode

  1. Check whether all selected images have identical dimensions.
  2. If they differ, switch to proportional fit, center crop, or stretch.
  3. If the batch still exceeds the limits, reduce the number of images or output dimensions.
  4. Keep the batch within 32 images and 32 million total output pixels.

Width and height settings have no effect

  1. Check the size-processing mode.
  2. If it is Original size, the width and height controls are intentionally ignored.
  3. Choose proportional fit, center crop, or stretch to apply a target size.

I cannot replace a node’s selection

  1. Check whether you selected multiple images for a single-image node.
  2. Use a batch node for multiple references.
  3. If the gallery reports that a selection input is controlled by a connection, address that connection before replacing the selection.
  4. If you want to keep the current node unchanged, choose New gallery node.

Favorites appear empty

  1. Confirm that you are viewing the correct asset library.
  2. Clear restrictive search terms and filters.
  3. Check whether you switched browsers or cleared site data. Favorites are browser-local and are not shared between the built-in and custom libraries.

A custom path works on one computer but not another

  1. Identify the computer actually running ComfyUI.
  2. Make sure the asset directory is accessible from that computer.
  3. Enter its full path in Asset settings.
  4. Confirm that the index and original images exist before saving.

The example workflow has a missing text-preview node

  1. Check whether your ComfyUI version includes PreviewAny.
  2. Replace it with a text-display node available in your setup.
  3. Alternatively, remove the text-preview portion and use the image preview alone.

Frequently asked questions

Does the gallery change a character’s clothes automatically? No. It supplies outfit references and descriptive text. Clothing changes require an appropriate downstream model and workflow.

Does it need an internet connection for everyday use? No. Browsing and node execution use local assets and work offline once those assets are available.

Does the extension include different AI models? The gallery itself does not require a model. The models used later in your workflow determine how the references affect generated or edited images.

Will every image in a batch automatically use its matching keywords? No. The batch provides corresponding information in JSON, but your downstream workflow must use that information to pair images with text.

Can I use it alongside the pose gallery? Yes. ComfyUI_Costume-Gallery_ZH and ComfyUI_Pose-Gallery_ZH are independent and keep their respective nodes and saved gallery data separate.

Will sharing my workflow also share the outfit images? No. The recipient needs access to the corresponding assets as well as the extension.

Practical resources

Asset sources and usage rights

The images and original data belong to ギア(giaaibusi). The Chinese index comes from the original Chinese outfit website, and the interface draws inspiration from Anhao’s Chinese pose gallery.

The author states that the added plugin code follows the repository’s MIT license. The contents of data/ and the image assets are not covered by that MIT license. Review LICENSE and the original creator’s terms before using or redistributing those materials.

ComfyUI_Costume-Gallery_ZH 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.

ComfyUI_Costume-Gallery_ZH detailed guide | ComfyUI