CyberHarem/ningguang_genshin

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

# Dataset of ningguang/凝光/凝光 (Genshin Impact)

This is the dataset of ningguang/凝光/凝光 (Genshin Impact), containing 500 images and their tags.

The core tags of this character are `long_hair, white_hair, hair_ornament, red_eyes, parted_bangs, breasts, tassel, very_long_hair, sidelocks, large_breasts`, 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 | 1.05 GiB | [Download](https://huggingface.co/datasets/CyberHarem/ningguang_genshin/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). |
| 1200 | 500 | 878.04 MiB | [Download](https://huggingface.co/datasets/CyberHarem/ningguang_genshin/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. |
| stage3-p480-1200 | 1315 | 1.63 GiB | [Download](https://huggingface.co/datasets/CyberHarem/ningguang_genshin/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/ningguang_genshin',
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 | 8 |  |  |  |  |  | 1girl, black_gloves, china_dress, crossed_legs, elbow_gloves, hair_stick, hairpin, holding_smoking_pipe, looking_at_viewer, sitting, solo, thighs, bare_legs, bare_shoulders, claw_ring, smile, white_dress, detached_sleeves, high_heels, fur_collar, closed_mouth, black_footwear, kiseru, leg_tattoo, yellow_dress |
| 1 | 17 |  |  |  |  |  | 1girl, bare_shoulders, black_gloves, china_dress, fur_collar, hairpin, solo, claw_ring, detached_sleeves, elbow_gloves, holding_smoking_pipe, looking_at_viewer, white_dress, fur_trim, smile, hair_stick, simple_background, white_background |
| 2 | 12 |  |  |  |  |  | 1girl, black_gloves, china_dress, detached_sleeves, elbow_gloves, fur_collar, hairpin, looking_at_viewer, smile, solo, white_dress, bare_shoulders, hair_stick, white_background, claw_ring, simple_background, closed_mouth, fur_trim, medium_breasts |
| 3 | 6 |  |  |  |  |  | 1girl, bare_shoulders, black_gloves, china_dress, elbow_gloves, fur_collar, hair_stick, hairpin, looking_at_viewer, sitting, solo, white_dress, claw_ring, crossed_legs, thighs, vision_(genshin_impact), detached_sleeves, smile, closed_mouth |
| 4 | 25 |  |  |  |  |  | 1girl, bare_shoulders, blue_dress, official_alternate_costume, solo, tassel_hair_ornament, looking_at_viewer, cleavage, smile, blue_collar, bracelet, vision_(genshin_impact), fur_trim, detached_sleeves, closed_mouth, thighs, white_background |
| 5 | 7 |  |  |  |  |  | 1girl, bare_shoulders, hairpin, navel, solo, cleavage, collarbone, hair_stick, looking_at_viewer, smile, tassel_hair_ornament, blush, stomach, thighs, armpits, arms_up, black_bra, parted_lips, black_panties, closed_mouth, curtains, underwear_only |
| 6 | 7 |  |  |  |  |  | 1girl, black_gloves, looking_at_viewer, smile, solo, white_shirt, alternate_costume, black_skirt, high-waist_skirt, elbow_gloves, hair_stick, hairpin, closed_mouth, teacup, hand_up, holding_cup, long_sleeves, necklace, sitting, tassel_hair_ornament |

### Table Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | black_gloves | china_dress | crossed_legs | elbow_gloves | hair_stick | hairpin | holding_smoking_pipe | looking_at_viewer | sitting | solo | thighs | bare_legs | bare_shoulders | claw_ring | smile | white_dress | detached_sleeves | high_heels | fur_collar | closed_mouth | black_footwear | kiseru | leg_tattoo | yellow_dress | fur_trim | simple_background | white_background | medium_breasts | vision_(genshin_impact) | blue_dress | official_alternate_costume | tassel_hair_ornament | cleavage | blue_collar | bracelet | navel | collarbone | blush | stomach | armpits | arms_up | black_bra | parted_lips | black_panties | curtains | underwear_only | white_shirt | alternate_costume | black_skirt | high-waist_skirt | teacup | hand_up | holding_cup | long_sleeves | necklace |
|----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------|:---------------|:--------------|:---------------|:---------------|:-------------|:----------|:-----------------------|:--------------------|:----------|:-------|:---------|:------------|:-----------------|:------------|:--------|:--------------|:-------------------|:-------------|:-------------|:---------------|:-----------------|:---------|:-------------|:---------------|:-----------|:--------------------|:-------------------|:-----------------|:--------------------------|:-------------|:-----------------------------|:-----------------------|:-----------|:--------------|:-----------|:--------|:-------------|:--------|:----------|:----------|:----------|:------------|:--------------|:----------------|:-----------|:-----------------|:--------------|:--------------------|:--------------|:-------------------|:---------|:----------|:--------------|:---------------|:-----------|
| 0 | 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 | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 1 | 17 |  |  |  |  |  | X | X | X | | X | X | X | X | X | | X | | | X | X | X | X | X | | X | | | | | | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 2 | 12 |  |  |  |  |  | X | X | X | | X | X | X | | X | | X | | | X | X | X | X | X | | X | X | | | | | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 3 | 6 |  |  |  |  |  | X | X | X | X | X | X | X | | X | X | X | X | | X | X | X | X | X | | X | X | | | | | | | | | X | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 4 | 25 |  |  |  |  |  | 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 | X | X | X | | | | | | | | | |
| 6 | 7 |  |  |  |  |  | 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: ningguang/凝光/凝光 (Genshin Impact)

数据集内容: 包含500张图像及其标签。

核心标签: long_hair, white_hair, hair_ornament, red_eyes, parted_bangs, breasts, tassel, very_long_hair, sidelocks, large_breasts

数据来源: 图像从多个网站（如danbooru, pixiv, zerochan等）爬取，爬虫系统由DeepGHS Team开发。

数据集包列表

名称	图像数量	大小	下载链接	类型	描述
raw	500	1.05 GiB	下载	Waifuc-Raw	原始数据，包含元信息（最小边对齐至1400像素）
1200	500	878.04 MiB	下载	IMG+TXT	数据集，短边不超过1200像素
stage3-p480-1200	1315	1.63 GiB	下载	IMG+TXT	三阶段裁剪数据集，区域不小于480x480像素

数据集加载示例

使用waifuc加载原始数据集的示例代码：

python import os import zipfile

from huggingface_hub import hf_hub_download from waifuc.source import LocalSource

下载原始归档文件

zip_file = hf_hub_download( repo_id=CyberHarem/ningguang_genshin, 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])

标签集群列表

#	样本数	图像示例	标签
0	8	至	1girl, black_gloves, china_dress, crossed_legs, elbow_gloves, hair_stick, hairpin, holding_smoking_pipe, looking_at_viewer, sitting, solo, thighs, bare_legs, bare_shoulders, claw_ring, smile, white_dress, detached_sleeves, high_heels, fur_collar, closed_mouth, black_footwear, kiseru, leg_tattoo, yellow_dress
1	17	至	1girl, bare_shoulders, black_gloves, china_dress, fur_collar, hairpin, solo, claw_ring, detached_sleeves, elbow_gloves, holding_smoking_pipe, looking_at_viewer, white_dress, fur_trim, smile, hair_stick, simple_background, white_background
2	12	至	1girl, black_gloves, china_dress, detached_sleeves, elbow_gloves, fur_collar, hairpin, looking_at_viewer, smile, solo, white_dress, bare_shoulders, hair_stick, white_background, claw_ring, simple_background, closed_mouth, fur_trim, medium_breasts
3	6	至	1girl, bare_shoulders, black_gloves, china_dress, elbow_gloves, fur_collar, hair_stick, hairpin, looking_at_viewer, sitting, solo, white_dress, claw_ring, crossed_legs, thighs, vision_(genshin_impact), detached_sleeves, smile, closed_mouth
4	25	至	1girl, bare_shoulders, blue_dress, official_alternate_costume, solo, tassel_hair_ornament, looking_at_viewer, cleavage, smile, blue_collar, bracelet, vision_(genshin_impact), fur_trim, detached_sleeves, closed_mouth, thighs, white_background
5	7	至	1girl, bare_shoulders, hairpin, navel, solo, cleavage, collarbone, hair_stick, looking_at_viewer, smile, tassel_hair_ornament, blush, stomach, thighs, armpits, arms_up, black_bra, parted_lips, black_panties, closed_mouth, curtains, underwear_only
6	7	至	1girl, black_gloves, looking_at_viewer, smile, solo, white_shirt, alternate_costume, black_skirt, high-waist_skirt, elbow_gloves, hair_stick, hairpin, closed_mouth, teacup, hand_up, holding_cup, long_sleeves, necklace, sitting, tassel_hair_ornament

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《原神》中的角色凝光，通过自动化爬取系统从Danbooru、Pixiv、Zerochan等多个知名图站收集了500张图像及其标签。原始数据经过精心处理，提供了三种不同规格的打包版本：raw版本保留元信息并将最小边对齐至1400像素；1200版本限制短边不超过1200像素；stage3-p480-1200版本则采用三级裁剪策略，确保图像区域不小于480×480像素。数据集的核心标签如长白发、红瞳等已被精简保留，便于下游任务使用。

特点

该数据集展现出多层次的精细结构特征。在标签层面，除了保留角色核心描述标签外，还提供了基于标签聚类的分组结果，揭示了不同服装变体（如旗袍、蓝色礼服、白色衬衫搭配黑裙等）和场景风格（如坐姿、持烟斗、内衣装等）的分布模式。每个聚类簇均附有样本图像和完整标签列表，便于研究者挖掘角色在不同装扮下的视觉特征。数据集规模适中，包含500张原始图像及经过裁剪增强后的1315张子图，兼顾了数据多样性与存储效率。

使用方法

数据集的使用方式灵活多样。研究者可直接下载预打包的ZIP文件，根据需求选择raw、1200或stage3-p480-1200版本。对于需要原始元数据的场景，推荐使用Waifuc框架加载raw数据集，通过简单的Python代码即可遍历图像、文件名及标签信息。数据集兼容HuggingFace Hub的下载机制，支持通过hf_hub_download函数自动获取文件。这种设计使得该数据集能够无缝集成到现有的文本到图像模型训练流程中，特别适用于角色定制化生成任务。

背景与挑战

背景概述

在生成式人工智能与数字娱乐交汇的浪潮中，面向特定角色的文本到图像数据集成为推动细粒度风格迁移与可控生成的关键资源。CyberHarem/ningguang_genshin数据集由DeepGHS团队于近期构建，聚焦于《原神》中高人气角色凝光，收录了500张经过标注的二次元图像。该数据集的核心研究问题在于如何通过自动爬取与标签清洗技术，为动漫角色构建高质量、多姿态的训练样本库，从而支撑下游的扩散模型微调与个性化生成任务。作为CyberHarem系列的重要组成部分，该数据集不仅为角色一致性生成提供了基准，也推动了社区驱动的二次元数据标准化进程，在爱好者与研究者中产生了广泛影响。

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严苛性。领域层面，文本到图像生成需应对角色多风格呈现与背景干扰的平衡，凝光复杂的服饰细节（如旗袍纹理、发饰与烟斗）要求模型具备精细的语义理解能力，而仅500张样本量难以覆盖所有姿态与光照条件，易导致生成结果中的特征混淆。构建过程中，自动爬取系统虽依托Danbooru、Pixiv等多源站点，但不同平台的标注格式差异与噪声标签（如冗余或缺失标签）增加了清洗难度；此外，图像版权归属与审核问题（如“not-for-all-audiences”标签）要求数据集在合规性与可用性之间权衡。多分辨率裁剪与聚类分析虽提升了数据多样性，却引入了样本分布不均与过拟合风险，进一步挑战了数据集的可扩展性。

常用场景

经典使用场景

在文本到图像生成领域，CyberHarem/ningguang_genshin 数据集为基于扩散模型和生成对抗网络的动漫角色图像生成提供了高质量的标注训练样本。研究者利用其包含的500张凝光角色图像及精细的标签信息，构建和微调能够精准描绘角色外貌特征（如白发、红瞳、长鬓角）与服饰细节（如旗袍、云肩、挂饰）的条件生成模型。该数据集的多分辨率版本（如1200像素和三级裁剪版本）适应了不同训练需求，成为风格迁移、角色一致性生成和图像超分辨率等任务中不可或缺的基准资源。

实际应用

在实际应用层面，CyberHarem/ningguang_genshin 数据集支撑了游戏内容创作和虚拟偶像生态的自动化生产流程。开发者可借助该数据集训练定制化的图像生成模型，快速产出符合角色设定的同人插画、宣传物料或交互式角色立绘。结合 waifuc 框架的原始数据加载接口，该数据集还赋能了二次元社区的个性化内容生成工具，例如自动合成角色不同表情和动作的组合图像，显著降低了人工绘制成本，提升了数字娱乐产业的创作效率。

衍生相关工作

基于此数据集，学术界和工业界衍生出了一系列经典工作。在方法层面，研究者开发了针对动漫角色的标签解耦网络，实现了对发型、服饰等属性的独立编辑；在应用层面，出现了融合该数据集的角色迁移学习框架，能够将凝光的视觉风格适配至其他虚拟角色。此外，数据集的聚类分析结果启发了基于图神经网络的服饰搭配推荐系统，以及面向角色多姿态生成的分层条件扩散模型，这些工作共同推动了动漫图像生成领域的进步。