CyberHarem/virtuosa_arknights

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

# Dataset of virtuosa/ヴィルトゥオーサ/塑心 (Arknights)

This is the dataset of virtuosa/ヴィルトゥオーサ/塑心 (Arknights), containing 321 images and their tags.

The core tags of this character are `black_hair, long_hair, blunt_bangs, sidelocks, straight_hair, halo, hime_cut, mole, mole_under_eye, black_halo, broken_halo, very_long_hair, dark_halo, black_eyes, wings, black_wings, pale_skin, detached_wings, energy_wings, breasts, colored_inner_hair, multicolored_hair, two-tone_hair, eyelashes, white_pupils, bright_pupils, medium_breasts, grey_hair`, 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 | Images-others | Images-head |
|:-----------------|---------:|:------------|:--------------------------------------------------------------------------------------------------------------------|:-----------|:---------------------------------------------------------------------|:----------------|:--------------|
| raw | 321 | 667.10 MiB | [Download](https://huggingface.co/datasets/CyberHarem/virtuosa_arknights/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). | -- | -- |
| stage3-p480-1200 | 818 | 1013.68 MiB | [Download](https://huggingface.co/datasets/CyberHarem/virtuosa_arknights/resolve/main/dataset-stage3-p480-1200.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. | 528 | 290 |

### 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/virtuosa_arknights',
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 | 5 |  |  |  |  |  | 1girl, belt_buckle, black_ascot, black_belt, black_garter_straps, black_gloves, black_skirt, black_thighhighs, bow_(music), bustier, buttons, cello, closed_mouth, collared_jacket, cowboy_shot, dress_shirt, floating_hair, grey_shirt, hands_up, holding_instrument, holding_violin, light_smile, miniskirt, playing_instrument, pleated_skirt, short_over_long_sleeves, solo, standing, white_jacket, wide_sleeves, wing_collar, zettai_ryouiki, antenna_hair, artist_name, belt_pouch, light_particles, lips, ahoge, backlighting, blurry, looking_at_viewer, musical_note, staff_(music), closed_eyes, facing_viewer, feet_out_of_frame, from_side, simple_background |
| 1 | 8 |  |  |  |  |  | 1girl, belt_buckle, black_ascot, black_belt, black_footwear, black_garter_straps, black_gloves, black_skirt, black_thighhighs, bow_(music), bustier, cello, closed_mouth, collared_jacket, dress_shirt, full_body, grey_shirt, holding_instrument, holding_violin, miniskirt, pleated_skirt, short_over_long_sleeves, solo, standing, white_jacket, wide_sleeves, wing_collar, zettai_ryouiki, absurdly_long_hair, looking_at_viewer, simple_background, strappy_heels, buttons, light_smile, hands_up, white_background, belt_pouch, blush, chibi, high_heels, yellow_pupils, artist_name, shadow |
| 2 | 7 |  |  |  |  |  | 1girl, black_ascot, black_gloves, bow_(music), cello, closed_mouth, collared_jacket, holding_instrument, holding_violin, light_smile, playing_instrument, short_over_long_sleeves, solo, strap, upper_body, white_jacket, wide_sleeves, wing_collar, blurry, looking_at_viewer, dress_shirt, floating_hair, from_side, hands_up, lips, looking_to_the_side, simple_background, white_background, backlighting, buttons, grey_shirt, hand_up, standing |
| 3 | 16 |  |  |  |  |  | 1girl, collared_jacket, looking_at_viewer, simple_background, solo, upper_body, white_jacket, black_ascot, white_background, wide_sleeves, wing_collar, strap, closed_mouth, short_over_long_sleeves, light_smile, ahoge, black_gloves, from_side, hand_up, looking_to_the_side, lips, dress_shirt |
| 4 | 10 |  |  |  |  |  | 1girl, collarbone, solo, closed_mouth, navel, stomach, looking_at_viewer, nude, simple_background, smile, white_background, hair_flowing_over, thighs, black_thighhighs, shadow, hair_censor, hair_over_breasts, lying, nipples, armpits, black_panties, cowboy_shot, groin, hair_spread_out, topless |

### Table Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | belt_buckle | black_ascot | black_belt | black_garter_straps | black_gloves | black_skirt | black_thighhighs | bow_(music) | bustier | buttons | cello | closed_mouth | collared_jacket | cowboy_shot | dress_shirt | floating_hair | grey_shirt | hands_up | holding_instrument | holding_violin | light_smile | miniskirt | playing_instrument | pleated_skirt | short_over_long_sleeves | solo | standing | white_jacket | wide_sleeves | wing_collar | zettai_ryouiki | antenna_hair | artist_name | belt_pouch | light_particles | lips | ahoge | backlighting | blurry | looking_at_viewer | musical_note | staff_(music) | closed_eyes | facing_viewer | feet_out_of_frame | from_side | simple_background | black_footwear | full_body | absurdly_long_hair | strappy_heels | white_background | blush | chibi | high_heels | yellow_pupils | shadow | strap | upper_body | looking_to_the_side | hand_up | collarbone | navel | stomach | nude | smile | hair_flowing_over | thighs | hair_censor | hair_over_breasts | lying | nipples | armpits | black_panties | groin | hair_spread_out | topless |
|----:|----------:|:-----------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------|:--------|:--------------|:--------------|:-------------|:----------------------|:---------------|:--------------|:-------------------|:--------------|:----------|:----------|:--------|:---------------|:------------------|:--------------|:--------------|:----------------|:-------------|:-----------|:---------------------|:-----------------|:--------------|:------------|:---------------------|:----------------|:--------------------------|:-------|:-----------|:---------------|:---------------|:--------------|:-----------------|:---------------|:--------------|:-------------|:------------------|:-------|:--------|:---------------|:---------|:--------------------|:---------------|:----------------|:--------------|:----------------|:--------------------|:------------|:--------------------|:-----------------|:------------|:---------------------|:----------------|:-------------------|:--------|:--------|:-------------|:----------------|:---------|:--------|:-------------|:----------------------|:----------|:-------------|:--------|:----------|:-------|:--------|:--------------------|:---------|:--------------|:--------------------|:--------|:----------|:----------|:----------------|:--------|:------------------|:----------|
| 0 | 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 | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 1 | 8 |  |  |  |  |  | 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 | | | | | | | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | |
| 2 | 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 | X | | | | | | X | X | | | | | X | | | | | | X | X | X | X | | | | | | | | | | | | | | | | |
| 3 | 16 |  |  |  |  |  | X | | X | | | X | | | | | | | X | X | | X | | | | | | X | | | | X | X | | X | X | X | | | | | | X | X | | | X | | | | | | X | X | | | | | X | | | | | | X | X | X | X | | | | | | | | | | | | | | | | |
| 4 | 10 |  |  |  |  |  | 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

数据集名称

Dataset of virtuosa/ヴィルトゥオーサ/塑心 (Arknights)

数据集描述

该数据集包含321张图片及其标签，主要描述《明日方舟》中的角色virtuosa/ヴィルトゥオーサ/塑心。

核心标签

• black_hair
• long_hair
• blunt_bangs
• sidelocks
• straight_hair
• halo
• hime_cut
• mole
• mole_under_eye
• black_halo
• broken_halo
• very_long_hair
• dark_halo
• black_eyes
• wings
• black_wings
• pale_skin
• detached_wings
• energy_wings
• breasts
• colored_inner_hair
• multicolored_hair
• two-tone_hair
• eyelashes
• white_pupils
• bright_pupils
• medium_breasts
• grey_hair

数据集包列表

名称	图片数量	大小	下载链接	类型	描述	其他图片数量	头部图片数量
raw	321	667.10 MiB	Download	Waifuc-Raw	包含元信息的原始数据（最小边对齐到1400像素，如果更大）。	--	--
stage3-p480-1200	818	1013.68 MiB	Download	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素。	528	290

标签聚类结果

原始文本版本

#	样本数量	图片1	图片2	图片3	图片4	图片5	标签
0	5	Img-1	Img-2	Img-3	Img-4	Img-5	1girl, belt_buckle, black_ascot, black_belt, black_garter_straps, black_gloves, black_skirt, black_thighhighs, bow_(music), bustier, buttons, cello, closed_mouth, collared_jacket, cowboy_shot, dress_shirt, floating_hair, grey_shirt, hands_up, holding_instrument, holding_violin, light_smile, miniskirt, playing_instrument, pleated_skirt, short_over_long_sleeves, solo, standing, white_jacket, wide_sleeves, wing_collar, zettai_ryouiki, antenna_hair, artist_name, belt_pouch, light_particles, lips, ahoge, backlighting, blurry, looking_at_viewer, musical_note, staff_(music), closed_eyes, facing_viewer, feet_out_of_frame, from_side, simple_background
1	8	Img-1	Img-2	Img-3	Img-4	Img-5	1girl, belt_buckle, black_ascot, black_belt, black_footwear, black_garter_straps, black_gloves, black_skirt, black_thighhighs, bow_(music), bustier, cello, closed_mouth, collared_jacket, dress_shirt, full_body, grey_shirt, holding_instrument, holding_violin, miniskirt, pleated_skirt, short_over_long_sleeves, solo, standing, white_jacket, wide_sleeves, wing_collar, zettai_ryouiki, absurdly_long_hair, looking_at_viewer, simple_background, strappy_heels, buttons, light_smile, hands_up, white_background, belt_pouch, blush, chibi, high_heels, yellow_pupils, artist_name, shadow
2	7	Img-1	Img-2	Img-3	Img-4	Img-5	1girl, black_ascot, black_gloves, bow_(music), cello, closed_mouth, collared_jacket, holding_instrument, holding_violin, light_smile, playing_instrument, short_over_long_sleeves, solo, strap, upper_body, white_jacket, wide_sleeves, wing_collar, blurry, looking_at_viewer, dress_shirt, floating_hair, from_side, hands_up, lips, looking_to_the_side, simple_background, white_background, backlighting, buttons, grey_shirt, hand_up, standing
3	16	Img-1	Img-2	Img-3	Img-4	Img-5	1girl, collared_jacket, looking_at_viewer, simple_background, solo, upper_body, white_jacket, black_ascot, white_background, wide_sleeves, wing_collar, strap, closed_mouth, short_over_long_sleeves, light_smile, ahoge, black_gloves, from_side, hand_up, looking_to_the_side, lips, dress_shirt
4	10	Img-1	Img-2	Img-3	Img-4	Img-5	1girl, collarbone, solo, closed_mouth, navel, stomach, looking_at_viewer, nude, simple_background, smile, white_background, hair_flowing_over, thighs, black_thighhighs, shadow, hair_censor, hair_over_breasts, lying, nipples, armpits, black_panties, cowboy_shot, groin, hair_spread_out, topless

表格版本

#	样本数量	图片1	图片2	图片3	图片4	图片5	1girl	belt_buckle	black_ascot	black_belt	black_garter_straps	black_gloves	black_skirt	black_thighhighs	bow_(music)	bustier	buttons	cello	closed_mouth	collared_jacket	cowboy_shot	dress_shirt	floating_hair	grey_shirt	hands_up	holding_instrument	holding_violin	light_smile	miniskirt	playing_instrument	pleated_skirt	short_over_long_sleeves	solo	standing	white_jacket	wide_sleeves	wing_collar	zettai_ryouiki	antenna_hair	artist_name	belt_pouch	light_particles	lips	ahoge	backlighting	blurry	looking_at_viewer	musical_note	staff_(music)	closed_eyes	facing_viewer	feet_out_of_frame	from_side	simple_background	black_footwear	full_body	absurdly_long_hair	strappy_heels	white_background	blush	chibi	high_heels	yellow_pupils	shadow	strap	upper_body	looking_to_the_side	hand_up	collarbone	navel	stomach	nude	smile	hair_flowing_over	thighs	hair_censor	hair_over_breasts	lying	nipples	armpits	black_panties	groin	hair_spread_out	topless
0	5	Img-1	Img-2	Img-3	Img-4	Img-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	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
1	8	Img-1	Img-2	Img-3	Img-4	Img-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					X		X	X	X	X	X	X	X	X	X	X	X										X	X	X	X	X	X	X	X	X	X	X
2	7	Img-1	Img-2	[Img-3](https://huggingface.co/datasets/CyberHarem/

搜集汇总

数据集介绍

构建方式

在二次元角色数据集构建领域，针对《明日方舟》角色塑心（Virtuosa），该数据集通过自动化爬虫系统从Danbooru、Pixiv、Zerochan等多个知名图站采集了321张原始图像。爬取系统由DeepGHS团队开发，确保了数据来源的多样性和广度。原始数据经过精细的标签修剪，保留了诸如黑发、长发、光环、翅膀、异色瞳等29个核心特征标签，并提供了两种规格的封装包：原始压缩包包含最小边对齐至1400像素的元数据信息，而经过三阶段裁剪后的数据集则提供了面积不小于480x480像素的图像，并配套了文本标签文件。

特点

该数据集的核心特色在于其精细化的标签体系和结构化的聚类分析。除了提供基础的原始图像与标签映射外，数据集还通过标签聚类技术，将图像依据服装、姿态、场景等语义特征自动划分为若干簇群，例如演奏大提琴的全身像、半身特写、甚至特定角度的裸露构图，每个簇群均附带详细的标签列表和样本预览。这种层次化的组织方式使得研究者能够直观地挖掘角色在不同情境下的视觉模式，为精细化图像生成与控制提供了坚实的数据基础。

使用方法

数据集的使用路径灵活多样。对于需要原始数据的用户，可通过Hugging Face Hub下载压缩包，并借助Waifuc库的LocalSource加载器快速读取图像及其对应的标签元数据；对于追求高效训练的场景，可直接使用三阶段裁剪后的图像-文本配对数据，免去了预处理步骤。此外，数据集提供的聚类结果表格可直接作为先验知识用于模型的条件控制或数据增强，用户仅需根据簇号索引即可定向筛选具有特定视觉特征的子集，从而提升模型对角色细节的还原能力。

背景与挑战

背景概述

在文本到图像生成领域，高质量、精细标注的动漫角色数据集是驱动模型学习特定角色特征与风格的关键资源。CyberHarem/virtuosa_arknights数据集由DeepGHS团队创建，专注于收录手游《明日方舟》中的角色塑心（Virtuosa），包含321张经过严格筛选与标注的图像。该数据集构建于2023年至2024年间，核心研究问题在于如何通过自动化爬取与多阶段裁剪技术，为角色定制化生成任务提供纯净、多样化的训练样本。其影响力体现在为LoRA微调等高效生成方法奠定了数据基础，推动了二次元角色个性化生成的发展。

当前挑战

该数据集面临的挑战主要来自两个方面。在领域问题层面，动漫角色图像生成需应对复杂多变的人物姿态、服饰细节及背景干扰，数据集需确保角色核心特征的准确性与一致性，避免生成结果出现语义混淆或特征丢失。在构建过程中，自动爬取自多个来源（如Danbooru、Pixiv）的图像质量参差不齐，需解决版权合规、标签噪声及图像分辨率差异等问题。此外，对角色关键属性（如发型、瞳孔颜色、翅膀形态）的精细标注与聚类分析，要求数据处理流程具备高度自动化与准确性，以支撑后续的模型训练与评估。

常用场景

经典使用场景

在文本到图像生成领域，CyberHarem/virtuosa_arknights 数据集被广泛用于训练和微调面向动漫角色定制化生成的扩散模型。该数据集包含321张来自丹博鲁、Pixiv等平台的高质量图像及其精细化标签，覆盖了角色从整体造型到局部特征的丰富细节，如发型、瞳孔颜色、服装配饰等。研究者常将其作为少样本学习的经典素材，借助 DreamBooth、LoRA 等参数高效微调技术，使预训练模型能够精准复现“塑心”这一角色的独特美学风格，从而在保留背景泛化能力的同时，实现对特定角色外观的忠实还原。

解决学术问题

该数据集有效缓解了动漫角色个性化图像生成中高质量标注数据匮乏的困境。学术研究中，如何让生成模型在极少量样本下学会某一角色的稳定表征，是一个关键挑战。CyberHarem/virtuosa_arknights 通过提供经过严格裁剪和标签聚类处理的图像集合，为探索角色身份保持（identity preservation）与风格迁移的平衡提供了基准。其意义在于推动了少样本条件生成的理论发展，使研究者能够更系统地分析标签粒度、样本多样性对角色概念解耦与泛化性能的影响，并为对比不同微调策略提供了可复现的实验平台。

衍生相关工作

基于 CyberHarem/virtuosa_arknights 数据集，学术界衍生了一系列经典工作。最典型的是利用其标签聚类结果进行角色概念解耦的研究，通过分析不同簇内的视觉共性（如演奏大提琴的特定姿势或服饰搭配），探索如何将复杂角色分解为可独立编辑的语义子空间。此外，该数据集还被用于验证多模态对齐技术，例如对比语言-图像预训练模型（如 CLIP）在动漫领域细粒度标签下的表征能力。部分工作进一步将其整合到自动化数据流水线中，结合 waifuc 工具链，为大规模动漫角色数据集构建提供了可复现的范式参考。