CyberHarem/chihiro_bluearchive

---
license: mit
task_categories:
- text-to-image
tags:
- art
- not-for-all-audiences
size_categories:
- n<1K
---

# Dataset of chihiro/各務チヒロ/千寻 (Blue Archive)

This is the dataset of chihiro/各務チヒロ/千寻 (Blue Archive), containing 348 images and their tags.

The core tags of this character are `short_hair, glasses, halo, black_hair, hair_ornament, breasts, semi-rimless_eyewear, large_breasts, green_eyes, rabbit_hair_ornament, hair_between_eyes, blue_hair, blue-framed_eyewear`, which are pruned in this dataset.

Images are crawled from many sites (e.g. danbooru, pixiv, zerochan ...), the auto-crawling system is powered by [DeepGHS Team](https://github.com/deepghs)([huggingface organization](https://huggingface.co/deepghs)).

## List of Packages

| Name | Images | Size | Download | Type | Description |
|:-----------------|---------:|:-----------|:---------------------------------------------------------------------------------------------------------------------|:-----------|:---------------------------------------------------------------------|
| raw | 348 | 609.03 MiB | [Download](https://huggingface.co/datasets/CyberHarem/chihiro_bluearchive/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). |
| 1200 | 348 | 507.19 MiB | [Download](https://huggingface.co/datasets/CyberHarem/chihiro_bluearchive/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. |
| stage3-p480-1200 | 884 | 1.04 GiB | [Download](https://huggingface.co/datasets/CyberHarem/chihiro_bluearchive/resolve/main/dataset-stage3-p480-1200.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. |

### Load Raw Dataset with Waifuc

We provide raw dataset (including tagged images) for [waifuc](https://deepghs.github.io/waifuc/main/tutorials/installation/index.html) loading. If you need this, just run the following code

```python
import os
import zipfile

from huggingface_hub import hf_hub_download
from waifuc.source import LocalSource

# download raw archive file
zip_file = hf_hub_download(
repo_id='CyberHarem/chihiro_bluearchive',
repo_type='dataset',
filename='dataset-raw.zip',
)

# extract files to your directory
dataset_dir = 'dataset_dir'
os.makedirs(dataset_dir, exist_ok=True)
with zipfile.ZipFile(zip_file, 'r') as zf:
zf.extractall(dataset_dir)

# load the dataset with waifuc
source = LocalSource(dataset_dir)
for item in source:
print(item.image, item.meta['filename'], item.meta['tags'])
```

## List of Clusters

List of tag clustering result, maybe some outfits can be mined here.

### Raw Text Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | Tags |
|----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| 0 | 10 |  |  |  |  |  | 1girl, blue_necktie, collared_shirt, simple_background, solo, upper_body, white_shirt, looking_at_viewer, white_background, closed_mouth, blush, long_sleeves, id_card, open_jacket, under-rim_eyewear, two-tone_jacket |
| 1 | 11 |  |  |  |  |  | 1girl, blue_cardigan, blue_necktie, long_sleeves, solo, blush, closed_mouth, collared_shirt, looking_at_viewer, open_jacket, two-tone_jacket, white_shirt, black_skirt, canned_coffee, holding_can, pleated_skirt, id_card, wristwatch, cowboy_shot, drink_can, outdoors, smile |
| 2 | 8 |  |  |  |  |  | 1girl, black_skirt, blue_cardigan, blue_necktie, collared_shirt, id_card, long_sleeves, open_jacket, pleated_skirt, solo, white_shirt, looking_at_viewer, cowboy_shot, two-tone_jacket, wristwatch, closed_mouth, blue_sweater_vest, simple_background, white_background |
| 3 | 5 |  |  |  |  |  | 1girl, alternate_costume, black_shirt, blue_jacket, simple_background, solo, track_jacket, white_background, gym_uniform, long_sleeves, looking_at_viewer, blush, closed_mouth, black_buruma, blue_buruma, cowboy_shot, open_clothes, partially_unzipped, short_sleeves, sweat, thighs |
| 4 | 13 |  |  |  |  |  | 1girl, cleavage, millennium_cheerleader_outfit_(blue_archive), white_skirt, navel, pleated_skirt, blush, cosplay, midriff, solo, holding_pom_poms, miniskirt, bare_shoulders, closed_mouth, looking_at_viewer, sweat, simple_background, stomach, crop_top, detached_collar, white_background |
| 5 | 19 |  |  |  |  |  | 1girl, solo, collarbone, blush, closed_mouth, looking_at_viewer, navel, cleavage, stomach, bare_shoulders, bikini, simple_background, alternate_costume, white_background |
| 6 | 14 |  |  |  |  |  | 1girl, blush, hetero, 1boy, nipples, sex, open_mouth, solo_focus, vaginal, mosaic_censoring, navel, penis, blue_necktie, open_clothes, white_shirt, female_pubic_hair, pussy, collarbone, long_sleeves, sweat, completely_nude, cowgirl_position, girl_on_top |
| 7 | 9 |  |  |  |  |  | long_sleeves, 1girl, black_gloves, blue_headwear, closed_mouth, hat, looking_at_viewer, black_pantyhose, smile, solo, hood, multicolored_jacket, backpack, blush, official_alternate_costume, black_footwear, black_skirt, holding, shoes, simple_background |
| 8 | 15 |  |  |  |  |  | 1girl, alternate_costume, cleavage, detached_collar, fake_animal_ears, looking_at_viewer, playboy_bunny, rabbit_ears, strapless_leotard, solo, bare_shoulders, blush, simple_background, white_background, wrist_cuffs, covered_navel, highleg_leotard, blue_leotard, closed_mouth, black_leotard, black_pantyhose, blue_bowtie, rabbit_tail, blue_necktie, fake_tail, jacket, open_clothes |

### Table Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | blue_necktie | collared_shirt | simple_background | solo | upper_body | white_shirt | looking_at_viewer | white_background | closed_mouth | blush | long_sleeves | id_card | open_jacket | under-rim_eyewear | two-tone_jacket | blue_cardigan | black_skirt | canned_coffee | holding_can | pleated_skirt | wristwatch | cowboy_shot | drink_can | outdoors | smile | blue_sweater_vest | alternate_costume | black_shirt | blue_jacket | track_jacket | gym_uniform | black_buruma | blue_buruma | open_clothes | partially_unzipped | short_sleeves | sweat | thighs | cleavage | millennium_cheerleader_outfit_(blue_archive) | white_skirt | navel | cosplay | midriff | holding_pom_poms | miniskirt | bare_shoulders | stomach | crop_top | detached_collar | collarbone | bikini | hetero | 1boy | nipples | sex | open_mouth | solo_focus | vaginal | mosaic_censoring | penis | female_pubic_hair | pussy | completely_nude | cowgirl_position | girl_on_top | black_gloves | blue_headwear | hat | black_pantyhose | hood | multicolored_jacket | backpack | official_alternate_costume | black_footwear | holding | shoes | fake_animal_ears | playboy_bunny | rabbit_ears | strapless_leotard | wrist_cuffs | covered_navel | highleg_leotard | blue_leotard | black_leotard | blue_bowtie | rabbit_tail | fake_tail | jacket |
|----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------|:---------------|:-----------------|:--------------------|:-------|:-------------|:--------------|:--------------------|:-------------------|:---------------|:--------|:---------------|:----------|:--------------|:--------------------|:------------------|:----------------|:--------------|:----------------|:--------------|:----------------|:-------------|:--------------|:------------|:-----------|:--------|:--------------------|:--------------------|:--------------|:--------------|:---------------|:--------------|:---------------|:--------------|:---------------|:---------------------|:----------------|:--------|:---------|:-----------|:-----------------------------------------------|:--------------|:--------|:----------|:----------|:-------------------|:------------|:-----------------|:----------|:-----------|:------------------|:-------------|:---------|:---------|:-------|:----------|:------|:-------------|:-------------|:----------|:-------------------|:--------|:--------------------|:--------|:------------------|:-------------------|:--------------|:---------------|:----------------|:------|:------------------|:-------|:----------------------|:-----------|:-----------------------------|:-----------------|:----------|:--------|:-------------------|:----------------|:--------------|:--------------------|:--------------|:----------------|:------------------|:---------------|:----------------|:--------------|:--------------|:------------|:---------|
| 0 | 10 |  |  |  |  |  | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 1 | 11 |  |  |  |  |  | X | X | X | | X | | X | X | | X | X | X | X | X | | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 2 | 8 |  |  |  |  |  | X | X | X | X | X | | X | X | X | X | | X | X | X | | X | X | X | | | X | X | X | | | | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 3 | 5 |  |  |  |  |  | X | | | X | X | | | X | X | X | X | X | | | | | | | | | | | X | | | | | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 4 | 13 |  |  |  |  |  | X | | | X | X | | | X | X | X | X | | | | | | | | | | X | | | | | | | | | | | | | | | | | X | | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 5 | 19 |  |  |  |  |  | X | | | X | X | | | X | X | X | X | | | | | | | | | | | | | | | | | X | | | | | | | | | | | | X | | | X | | | | | X | X | | | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 6 | 14 |  |  |  |  |  | X | X | | | | | X | | | | X | X | | | | | | | | | | | | | | | | | | | | | | | X | | | X | | | | | X | | | | | | | | | X | | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | |
| 7 | 9 |  |  |  |  |  | X | | | X | X | | | X | | X | X | X | | | | | | X | | | | | | | | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | |
| 8 | 15 |  |  |  |  |  | X | X | | X | X | | | X | X | X | X | | | | | | | | | | | | | | | | | X | | | | | | | X | | | | | X | | | | | | | | X | | | X | | | | | | | | | | | | | | | | | | | | X | | | | | | | | X | X | X | X | X | X | X | X | X | X | X | X | X |

原始信息汇总

数据集概述：chihiro/各務チヒロ/千寻 (Blue Archive)

数据集基本信息

• 名称：chihiro/各務チヒロ/千寻 (Blue Archive)
• 包含内容：348张图像及其标签
• 核心标签：short_hair, glasses, halo, black_hair, hair_ornament, breasts, semi-rimless_eyewear, large_breasts, green_eyes, rabbit_hair_ornament, hair_between_eyes, blue_hair, blue-framed_eyewear
• 许可证：MIT
• 任务类别：text-to-image
• 标签：art, not-for-all-audiences
• 大小类别：n<1K

数据集包列表

名称	图像数量	大小	下载链接	类型	描述
raw	348	609.03 MiB	下载	Waifuc-Raw	包含元信息的原始数据（如果较大，则最小边缘对齐到1400）。
1200	348	507.19 MiB	下载	IMG+TXT	短边不超过1200像素的数据集。
stage3-p480-1200	884	1.04 GiB	下载	IMG+TXT	三阶段裁剪数据集，区域不小于480x480像素。

数据集使用示例

• 加载原始数据集：提供了一个Python代码示例，用于从Hugging Face Hub下载并解压数据集，然后使用waifuc加载数据集。

数据集标签集群列表

• 集群列表：提供了多个集群的详细信息，包括每个集群的样本数量、图像示例及其对应的标签。这些集群可能包含特定的服装或场景，适合进一步的数据挖掘和分析。

搜集汇总

数据集介绍

构建方式

该数据集聚焦于《蔚蓝档案》中的角色千寻（各務チヒロ），基于深度图像采集系统，从Danbooru、Pixiv、Zerochan等多个平台自动爬取原始图像，共计包含348张图片及其对应标签。数据经过去重与核心标签精简处理，保留如短发、眼镜、光环、黑发等关键描述。为适应不同训练需求，数据集提供了三种预处理版本：原始元数据包（Waifuc-Raw）、短边不超过1200像素的标准化图像包，以及基于三级裁剪策略、面积不低于480×480像素的增强图像包，后者通过切分扩充至884个样本。

特点

该数据集以角色一致性为设计核心，通过标签聚类分析，自动挖掘出角色在不同着装与场景下的视觉特征，如学院制服、运动服、兔女郎装等变体，并以可视化表格呈现各聚类的样本与标签分布。所有图像均经过质量筛选与尺寸统一，确保数据干净、标注规范。此外，数据集保留了完整的元信息，包括文件名称与标签列表，便于进行细粒度的风格迁移或角色泛化模型训练。

使用方法

用户可通过Hugging Face Hub直接下载各压缩包。对于原始数据，推荐使用Waifuc框架加载，通过LocalSource接口读取解压后的图像与标签，实现高效的数据流处理。标准化与裁剪数据集则直接提供图像与对应TXT标签文件，适用于常见的文本到图像生成管线。用户可根据模型需求选择合适版本，例如直接使用1200像素版本进行训练，或利用stage3版本增强数据多样性。

背景与挑战

背景概述

在生成式人工智能领域，文本到图像（text-to-image）模型的发展日新月异，其中对特定角色或主题的高质量、精细化数据集的需求日益凸显。CyberHarem/chihiro_bluearchive数据集正是为此而生，由DeepGHS团队于近期创建，专注于游戏《蔚蓝档案》中角色“千寻”（Chihiro）的形象数据。该数据集包含348张经过标注的高清图像，核心标签如短发、眼镜、光环等，精准捕捉了角色的外观特征。作为动漫角色数据集构建的典范，它通过自动爬取Danbooru、Pixiv等多源平台，结合Waifuc框架实现了标准化处理，为文本到图像模型在二次元角色生成、风格迁移等任务中提供了坚实的数据基础，显著推动了针对特定虚拟角色的可控生成研究。

当前挑战

该数据集面临的核心挑战在于领域问题的复杂性与构建过程的精细化要求。首先，在解决文本到图像生成任务时，角色数据集需应对细粒度特征一致性难题，例如千寻的眼镜、光环等独特装饰在多种姿态和场景下的准确再现，以及避免生成结果中的特征混淆或缺失。其次，构建过程中遭遇多重困难：多源爬取导致图像分辨率、光照、构图差异显著，需通过边缘对齐和裁剪策略（如1200像素限制和三级裁剪）进行标准化；标签系统虽基于自动标注，但不同来源的元数据质量参差不齐，需人工校验以提升语义精确度；此外，数据量仅348张，规模较小，可能限制模型泛化能力，易引发过拟合风险，这对数据增强技术和标签多样性提出了更高要求。

常用场景

经典使用场景

在文本到图像生成领域，该数据集常被用于训练和微调基于扩散模型的动漫角色生成器。研究者利用其提供的348张高分辨率图像及配套标签，能够针对特定角色（如《蔚蓝档案》中的千寻）进行概念定制化学习，从而在保持角色核心特征（如短发、眼镜、光环）的前提下，生成风格多样且姿态各异的全新图像。该数据集因其精心裁剪与标签聚类，成为评估生成模型在细粒度角色一致性控制方面性能的经典基准。

解决学术问题

该数据集有效解决了动漫角色生成中身份保持与风格泛化之间的矛盾。学术研究中，模型常因训练数据不足而丢失角色标志性元素，或过度拟合导致生成多样性匮乏。该数据集通过提供多角度、多装束的图像与结构化标签，使得研究者能够探索条件生成、概念分解及属性解耦等关键问题，推动了可控图像生成技术在虚拟角色创作领域的理论发展，并显著提升了生成结果在语义准确性上的评估标准。

衍生相关工作

该数据集衍生了一系列经典工作，包括基于LoRA（Low-Rank Adaptation）的角色微调方法，通过将角色特征编码为轻量级权重模块，实现了高效迁移学习；以及利用标签聚类结果构建多概念组合生成框架，能够同时控制角色身份与特定服饰风格。此外，部分研究以此为基础，探索了基于扩散模型的角色姿态与表情编辑技术，显著提升了生成内容的交互性与可控性。这些工作共同构成了动漫角色生成领域从数据到算法的完整研究链条。