CyberHarem/ui_bluearchive

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

# Dataset of ui/古関ウイ/忧 (Blue Archive)

This is the dataset of ui/古関ウイ/忧 (Blue Archive), containing 500 images and their tags.

The core tags of this character are `long_hair, black_hair, hairband, halo, bags_under_eyes, hair_between_eyes, blue_hairband, very_long_hair, blue_eyes, breasts, purple_eyes`, 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 | 500 | 946.38 MiB | [Download](https://huggingface.co/datasets/CyberHarem/ui_bluearchive/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). |
| 1200 | 500 | 779.09 MiB | [Download](https://huggingface.co/datasets/CyberHarem/ui_bluearchive/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. |
| stage3-p480-1200 | 1312 | 1.57 GiB | [Download](https://huggingface.co/datasets/CyberHarem/ui_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/ui_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 | 30 |  |  |  |  |  | 1girl, black_bikini, official_alternate_costume, ponytail, solo, long_sleeves, looking_at_viewer, small_breasts, blush, collarbone, necklace, navel, purple_hairband, open_mouth, grey_cardigan, jacket, stomach, simple_background, white_background, open_cardigan, off_shoulder, sitting |
| 1 | 7 |  |  |  |  |  | 1girl, black_bikini, blush, collarbone, looking_at_viewer, o-ring_bikini, official_alternate_costume, ponytail, purple_hairband, navel, simple_background, small_breasts, solo, stomach, white_background, necklace, open_mouth, medium_breasts, sidelocks |
| 2 | 37 |  |  |  |  |  | 1girl, cardigan, solo, long_sleeves, looking_at_viewer, pleated_skirt, white_skirt, holding_book, white_sailor_collar, blue_neckerchief, closed_mouth, necklace, simple_background, white_background, blush, white_serafuku, long_skirt, sweater, low_twintails, off_shoulder |
| 3 | 7 |  |  |  |  |  | 1girl, cardigan, long_sleeves, solo, blue_neckerchief, looking_at_viewer, necklace, pleated_skirt, white_sailor_collar, white_skirt, blush, closed_mouth, holding_book, long_skirt, low_twintails, simple_background, white_background, white_serafuku, bookshelf, hair_ribbon, indoors, off_shoulder, shirt, sitting, slippers |
| 4 | 5 |  |  |  |  |  | 1girl, blush, female_pubic_hair, nipples, small_breasts, solo, looking_at_viewer, navel, open_mouth, simple_background, completely_nude, twintails, black_pubic_hair, cardigan, crossed_bangs, flying_sweatdrops, grey_background, hair_ribbon, pussy, white_background |
| 5 | 7 |  |  |  |  |  | 1girl, blush, nipples, pussy_juice, solo, female_masturbation, female_pubic_hair, spread_legs, completely_nude, medium_breasts, navel, anus, open_mouth, sitting, sweat, dildo, fingering, mosaic_censoring, small_breasts |
| 6 | 14 |  |  |  |  |  | 1girl, blush, 1boy, hetero, nipples, solo_focus, penis, open_mouth, female_pubic_hair, spread_legs, sweat, navel, sex, small_breasts, black_pantyhose, looking_at_viewer, pov, torn_pantyhose, heart, vaginal, bar_censor, cardigan, completely_nude, cum_in_pussy, low_twintails, medium_breasts, on_back, panties_aside |
| 7 | 5 |  |  |  |  |  | 1girl, bare_shoulders, blush, detached_collar, fake_animal_ears, looking_at_viewer, playboy_bunny, rabbit_ears, simple_background, solo, alternate_costume, black_leotard, black_pantyhose, highleg_leotard, strapless_leotard, white_background, cardigan, covered_navel, fishnet_pantyhose, low_twintails, medium_breasts, open_mouth, bowtie, cleavage, cowboy_shot, heart, long_sleeves, neckerchief, off_shoulder, ribbon, sitting, sleeves_past_wrists |
| 8 | 7 |  |  |  |  |  | 1girl, frilled_apron, looking_at_viewer, simple_background, white_apron, black_dress, enmaided, juliet_sleeves, maid_apron, maid_headdress, solo, white_background, open_mouth, blush, cat_ears, purple_hairband, purple_necktie, ribbon |

### Table Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | black_bikini | official_alternate_costume | ponytail | solo | long_sleeves | looking_at_viewer | small_breasts | blush | collarbone | necklace | navel | purple_hairband | open_mouth | grey_cardigan | jacket | stomach | simple_background | white_background | open_cardigan | off_shoulder | sitting | o-ring_bikini | medium_breasts | sidelocks | cardigan | pleated_skirt | white_skirt | holding_book | white_sailor_collar | blue_neckerchief | closed_mouth | white_serafuku | long_skirt | sweater | low_twintails | bookshelf | hair_ribbon | indoors | shirt | slippers | female_pubic_hair | nipples | completely_nude | twintails | black_pubic_hair | crossed_bangs | flying_sweatdrops | grey_background | pussy | pussy_juice | female_masturbation | spread_legs | anus | sweat | dildo | fingering | mosaic_censoring | 1boy | hetero | solo_focus | penis | sex | black_pantyhose | pov | torn_pantyhose | heart | vaginal | bar_censor | cum_in_pussy | on_back | panties_aside | bare_shoulders | detached_collar | fake_animal_ears | playboy_bunny | rabbit_ears | alternate_costume | black_leotard | highleg_leotard | strapless_leotard | covered_navel | fishnet_pantyhose | bowtie | cleavage | cowboy_shot | neckerchief | ribbon | sleeves_past_wrists | frilled_apron | white_apron | black_dress | enmaided | juliet_sleeves | maid_apron | maid_headdress | cat_ears | purple_necktie |
|----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------|:---------------|:-----------------------------|:-----------|:-------|:---------------|:--------------------|:----------------|:--------|:-------------|:-----------|:--------|:------------------|:-------------|:----------------|:---------|:----------|:--------------------|:-------------------|:----------------|:---------------|:----------|:----------------|:-----------------|:------------|:-----------|:----------------|:--------------|:---------------|:----------------------|:-------------------|:---------------|:-----------------|:-------------|:----------|:----------------|:------------|:--------------|:----------|:--------|:-----------|:--------------------|:----------|:------------------|:------------|:-------------------|:----------------|:--------------------|:------------------|:--------|:--------------|:----------------------|:--------------|:-------|:--------|:--------|:------------|:-------------------|:-------|:---------|:-------------|:--------|:------|:------------------|:------|:-----------------|:--------|:----------|:-------------|:---------------|:----------|:----------------|:-----------------|:------------------|:-------------------|:----------------|:--------------|:--------------------|:----------------|:------------------|:--------------------|:----------------|:--------------------|:---------|:-----------|:--------------|:--------------|:---------|:----------------------|:----------------|:--------------|:--------------|:-----------|:-----------------|:-------------|:-----------------|:-----------|:-----------------|
| 0 | 30 |  |  |  |  |  | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 1 | 7 |  |  |  |  |  | X | X | X | X | X | | X | X | X | X | X | X | X | X | | | X | X | X | | | | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 2 | 37 |  |  |  |  |  | X | | | | X | X | X | | X | | X | | | | | | | X | X | | X | | | | | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 3 | 7 |  |  |  |  |  | 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 | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 4 | 5 |  |  |  |  |  | X | | | | X | | X | X | X | | | X | | X | | | | X | X | | | | | | | X | | | | | | | | | | | | X | | | | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 5 | 7 |  |  |  |  |  | X | | | | X | | | X | X | | | X | | 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 | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 7 | 5 |  |  |  |  |  | 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 | X | X | X | X | X | | | | | | | | | |
| 8 | 7 |  |  |  |  |  | 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：ui/古関ウイ/忧 (Blue Archive)

数据集基本信息

• 名称：ui/古関ウイ/忧 (Blue Archive)
• 包含内容：500张图像及其标签
• 核心标签：long_hair, black_hair, hairband, halo, bags_under_eyes, hair_between_eyes, blue_hairband, very_long_hair, blue_eyes, breasts, purple_eyes

数据集版本与下载

版本详情

包名	图像数量	大小	下载链接	类型	描述
raw	500	946.38 MiB	下载	Waifuc-Raw	包含元信息的原始数据（最小边长对齐至1400像素）
1200	500	779.09 MiB	下载	IMG+TXT	短边不超过1200像素的数据集
stage3-p480-1200	1312	1.57 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

• 加载工具：waifuc

• 加载示例代码： python import os import zipfile

  from huggingface_hub import hf_hub_download from waifuc.source import LocalSource

  下载原始存档文件

  zip_file = hf_hub_download( repo_id=CyberHarem/ui_bluearchive, repo_type=dataset, filename=dataset-raw.zip, )

  提取文件至目录

  dataset_dir = dataset_dir os.makedirs(dataset_dir, exist_ok=True) with zipfile.ZipFile(zip_file, r) as zf: zf.extractall(dataset_dir)

  使用waifuc加载数据集

  source = LocalSource(dataset_dir) for item in source: print(item.image, item.meta[filename], item.meta[tags])

数据集标签集群

• 集群列表：提供了标签集群的结果，可能包含可挖掘的服装信息。
• 集群示例：
  • 集群0：包含30个样本，主要标签包括1girl, black_bikini, official_alternate_costume等。
  • 集群1：包含7个样本，主要标签包括1girl, black_bikini, blush等。
  • 集群2：包含37个样本，主要标签包括1girl, cardigan, solo等。
  • 集群3：包含7个样本，主要标签包括1girl, cardigan, long_sleeves等。
  • 集群4：包含5个样本，主要标签包括1girl, blush, female_pubic_hair等。
  • 集群5：包含7个样本，主要标签包括1girl, blush, nipples等。
  • 集群6：包含14个样本，主要标签包括1girl, blush, 1boy等。
  • 集群7：包含5个样本，主要标签包括1girl, bare_shoulders, fake_animal_ears等。
  • 集群8：包含7个样本，主要标签包括1girl, frilled_apron, looking_at_viewer等。

搜集汇总

数据集介绍

构建方式

在数字内容创作领域，高质量的角色图像数据集对于文本到图像生成模型的训练至关重要。该数据集聚焦于《蔚蓝档案》中的角色“忧”（ui/古関ウイ），通过自动化爬虫系统从Danbooru、Pixiv、Zerochan等多个主流图像平台采集原始图像，并辅以精细的标签体系。数据集共包含500张图像，核心标签如长发、黑发、光环、眼袋等已进行精简处理。构建过程中，团队采用多尺度版本策略：原始版本保留元信息并最小边对齐至1400像素；1200版本将短边限制在1200像素以内；而stage3-p480-1200版本则通过三阶段裁剪，确保图像区域不小于480×480像素，从而适配不同训练需求。

使用方法

使用该数据集时，研究者可依据项目需求选择适当的版本。对于需要原始元信息的场景，建议通过waifuc库加载raw数据集：首先从HuggingFace Hub下载dataset-raw.zip文件，解压至本地目录，随后利用LocalSource接口读取图像与标签。此方法支持逐项迭代，便于提取文件名、标签等元数据。若追求训练效率，可直接使用1200或stage3-p480-1200版本的IMG+TXT格式压缩包，这些文件已对齐至统一尺寸，无需额外预处理。此外，数据集提供的标签聚类表格可作为先验知识，辅助设计提示词或评估模型在不同风格下的生成质量。

背景与挑战

背景概述

在生成式人工智能与二次元文化交汇的浪潮中，针对特定虚拟角色的高质量图像数据集成为推动文本到图像生成模型精细化发展的关键资源。CyberHarem团队于近期构建了ui_bluearchive数据集，聚焦于游戏《碧蓝档案》中的人气角色“古関ウイ”（忧）。该数据集由DeepGHS团队主导开发，依托其自主研发的自动化爬取系统，从Danbooru、Pixiv、Zerochan等多个知名二次元图像平台采集并整理了500张高分辨率图像及其对应的标签元数据。数据集的核心研究问题在于为特定动漫角色提供标准化、多模态的训练素材，以提升生成模型对该角色特征（如长发、黑色发色、光环、眼袋等）的还原度与泛化能力。这一数据集的发布，不仅丰富了二次元垂直领域的数据生态，也为角色定制化生成任务提供了可复用的基准。

当前挑战

该数据集所面临的挑战首先体现在领域问题的复杂性上：文本到图像生成任务要求模型能够精准理解并再现角色在多种服饰、姿态与背景下的细粒度视觉特征，而500张图像的数据规模相对有限，难以覆盖角色所有可能的变体，易导致生成结果出现特征混淆或模式塌缩。其次，在数据集构建过程中，自动化爬取虽提升了效率，却引入了标签噪声与不一致性问题——不同来源的图像标注风格各异，且部分标签（如“completely_nude”等成人内容）涉及敏感类别，增加了数据清洗与伦理审查的难度。此外，图像分辨率与裁剪策略的多样性（如raw、1200、stage3-p480-1200三种规格）虽提供了灵活性，但也对模型训练的标准化流程提出了额外要求。

常用场景

经典使用场景

在二次元角色生成与风格迁移领域，CyberHarem/ui_bluearchive数据集为文本到图像模型提供了精细化的角色模板。该数据集包含500张来自《蔚蓝档案》中角色“忧”的高质量图像及其对应的标签，涵盖从原始采集到多尺度裁剪的多种格式，尤其适合用于训练基于扩散模型的角色一致性与细节生成任务。通过整合Danbooru、Pixiv等多源平台的数据，该数据集确保了角色特征的多样性，成为研究可控图像生成和角色属性解耦的经典基准。

解决学术问题

该数据集有效解决了二次元角色生成中特征一致性与标签稀疏性的学术难题。通过提供经过人工筛选的核心标签（如发型、瞳色、服饰）和聚类后的多模态标签体系，它支持研究者探索细粒度属性对齐与跨模态表示学习。这一资源促进了文本到图像模型在动漫领域中的泛化能力研究，并为评估生成图像与原始角色特征的语义匹配度提供了标准化测试集，推动了领域内可解释性与鲁棒性分析的进展。

实际应用

在实际应用中，该数据集被广泛用于个性化角色生成工具的开发，例如游戏角色立绘的自动创作、虚拟主播形象定制以及动漫风格插画的批量生产。开发者可基于该数据集训练轻量级模型，实现从文本描述到符合原设的角色图像的高效转换。此外，其多尺度裁剪版本（如stage3-p480-1200）可直接用于移动端或Web端的实时生成管线，降低了二次元内容创作的技术门槛，加速了数字娱乐产业的工业化流程。

数据集最近研究

最新研究方向

在二次元角色定制化生成领域，CyberHarem/ui_bluearchive数据集聚焦于《蔚蓝档案》角色“忧”的高质量图像与标签采集，为文本到图像模型提供了精细化的训练资源。当前前沿研究方向集中于利用此类角色专属数据集驱动扩散模型（如Stable Diffusion）进行风格迁移与角色一致性保持，尤其是在LoRA微调技术中实现特定人物特征的精准复现。该数据集通过多源爬取（Danbooru、Pixiv等）和聚类标注（涵盖服装、姿势、场景等属性），助力解决动漫角色生成中的身份保持难题，其影响在于推动虚拟偶像、游戏美术自动化及个性化创作工具的发展，成为连接二次元文化生态与生成式AI技术的重要桥梁。