Chinese Tutorial for the AI Image-Making Asset Library: Quickly Build Prompts with Face, Pose, and Clothing References

Project Overview

AI Image-Making Asset Library · Chinese Edition is a Chinese search interface that brings together three sets of AI image-making reference assets publicly released by the author あぎ / ギア in one place. The interface provides Chinese categories and tags, bilingual Chinese and English search, and one-click copying of image-making instructions.

The project is primarily intended for asset search. The repository contains thumbnails, while original images must be obtained from the configured local directory or the author's source site.

Three Asset Libraries

• Face Library: Provides front-facing head references with solid-color backgrounds, including 893 close-up face images and 606 styling images, for a total of 1,499 images. The source site is AI Portrait Gallery.
• Pose Library: Provides faceless gray mannequin line drawings. Each pose includes 5 camera angles, for a total of 288 poses and 1,439 images. The source site is AI Pose Gallery.
• Clothing Library: Provides reference images showing clothing without people, divided into 8 categories, with a total of 1,407 outfits. The source site is AI Costume Gallery.

Installation and Startup

Method 1: Open index.html Directly

This is the simplest way to use the project. After downloading or cloning the repository, double-click index.html in your file manager. The browser will open the page using file://, with no dependencies or server required.

1. Obtain and extract the project files.
2. Open the project directory.
3. Double-click index.html.
4. Select an asset library in the Chinese interface and start searching.

Method 2: Start a Local Static Server

If the browser restricts local scripts or some features do not work correctly, run a Python static server in the project directory:

python3 -m http.server 8080

After starting the server, visit http://localhost:8080. This method still provides only local static access and does not require any additional backend service.

Basic Workflow

1. Select an Asset Library

After opening the page, enter the face, pose, or clothing library according to your needs. Use the face library when you need to control facial features, the pose library when you need to arrange actions and camera angles, and the clothing library when you need to find clothing directions.

2. Search in Chinese and English

Enter Chinese or English keywords in the search box and use categories and tags to filter assets. Chinese tags are useful for quick browsing, while English search can support reference searches within an existing image-making workflow.

3. View Reference Images and Copy Instructions

After selecting a suitable asset, view its thumbnail and related information. Use the copy function provided by the interface to copy the corresponding image-making instructions with one click, then paste them into your AI image-making tool for further creation.

4. Download Original Images

The images included in the project are primarily thumbnails. When you click “Download Original,” the program first checks for a local original image. If the local original is unavailable, it redirects to the corresponding file on the author's source site. Access to the source site may be slow in mainland China, so preparing the original images in advance is recommended when you need high-resolution references.

Configure Local Original Images

If you want to view or download high-resolution images faster in an offline environment, prepare the files according to the local directory configuration in config.js.

1. Open config.js in the project and check the configured local original-image directory.
2. Download the corresponding assets from the cloud storage addresses provided in the README.
3. Place the original images in the parent directory of the project, and ensure that the directory structure matches the configuration in config.js.
4. Open index.html using file://, then try downloading the original images again.

The program uses local original images directly only when file:// is used to open the page, a local directory is configured, and the target files actually exist. Otherwise, it redirects to the author's source site.

Original Image Download Addresses

The following addresses provide the high-resolution original images required for local previews:

• AI Image-Making Face Library: Baidu Netdisk, Quark Netdisk.
• AI Clothing Library: Baidu Netdisk, Quark Netdisk.
• AI Image-Making Pose Library: Quark Netdisk, Baidu Netdisk.

Advanced Usage Tips

Combine All Three Reference Types

First determine the facial direction in the face library, then select actions and camera angles from the pose library, and finally add clothing references from the clothing library. Combining image-making instructions from different asset libraries as needed can help create more complete creative references.

Prioritize Local Original Images

If you search assets frequently, place the required original images in a local directory that matches the configuration. This reduces waiting time when accessing the source site and enables the use of local resources offline.

Distinguish Thumbnails from Original Images

The thumbnails included in the repository are intended for searching and previewing and do not mean that the original images are included in the repository. When you need high-resolution versions, obtain them through the download addresses provided in the README or from the author's source site. Do not mistake the repository thumbnails for complete original-image resources.

Licensing and Usage Restrictions

Before using the assets, be sure to distinguish the licensing terms of the three libraries:

• Face Library: Uses CC0. The README states that the images may be redistributed.
• Pose Library: May be used commercially, modified, and used as reference for free, but redistribution of the images themselves, including modified versions, is prohibited.
• Clothing Library: May be used commercially, modified, and used as reference for free, but redistribution of the images themselves, including modified versions, is prohibited.

None of the three asset libraries involves real people; all images are AI-generated. The clothing library author has conducted a brand-similarity check but does not guarantee that there are no infringement risks. For commercial projects, you must still make your own assessment and assume the corresponding responsibility.

This repository provides only a Chinese search interface for personal use. Obtain original images through the author's source site or the methods provided in the README, and do not package and redistribute the original images.

Summary

With this Chinese interface, you can quickly search local face, pose, and clothing references, filter assets using Chinese and English keywords, and copy image-making instructions with one click. Double-click index.html to get started. If the browser imposes restrictions, run python3 -m http.server 8080 to start a local static service. When using original images or creating commercial work, always follow the licensing requirements of the corresponding asset library.