CyberHarem/sonya_fireemblem

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

# Dataset of sonya (Fire Emblem)

This is the dataset of sonya (Fire Emblem), containing 250 images and their tags.

The core tags of this character are `purple_hair, long_hair, breasts, large_breasts, earrings, 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 | 250 | 341.81 MiB | [Download](https://huggingface.co/datasets/CyberHarem/sonya_fireemblem/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). |
| 800 | 250 | 182.66 MiB | [Download](https://huggingface.co/datasets/CyberHarem/sonya_fireemblem/resolve/main/dataset-800.zip) | IMG+TXT | dataset with the shorter side not exceeding 800 pixels. |
| stage3-p480-800 | 590 | 366.11 MiB | [Download](https://huggingface.co/datasets/CyberHarem/sonya_fireemblem/resolve/main/dataset-stage3-p480-800.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. |
| 1200 | 250 | 296.80 MiB | [Download](https://huggingface.co/datasets/CyberHarem/sonya_fireemblem/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. |
| stage3-p480-1200 | 590 | 535.36 MiB | [Download](https://huggingface.co/datasets/CyberHarem/sonya_fireemblem/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/sonya_fireemblem',
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 | 14 |  |  |  |  |  | 1boy, 1girl, hetero, solo_focus, jewelry, nipples, penis, blush, cum_in_pussy, open_mouth, sex, vaginal, navel, thighhighs, ahegao, heart, spread_legs, sweat, cowgirl_position, tongue_out, brown_eyes, completely_nude, girl_on_top, saliva, uncensored |
| 1 | 12 |  |  |  |  |  | 1girl, 1boy, blush, hetero, mosaic_censoring, penis, solo_focus, fellatio, cum_in_mouth, heart, jewelry, dark-skinned_male, interracial, nipples, nude, circlet, gloves, thighhighs |
| 2 | 5 |  |  |  |  |  | 1girl, fellatio, hetero, multiple_penises, solo_focus, 2boys, double_penetration, jewelry, mmf_threesome, nipples, uncensored, blush, completely_nude, navel, spitroast, spread_legs, testicles, vaginal, black_gloves, dark-skinned_male, gangbang, gloved_handjob, interracial, pregnant, pussy_juice, thighhighs |
| 3 | 38 |  |  |  |  |  | jewelry, 1girl, cleavage, solo, cape, circlet, looking_at_viewer, smile, simple_background, black_gloves, dress, thighhighs, white_background |
| 4 | 6 |  |  |  |  |  | 1girl, fake_animal_ears, looking_at_viewer, pantyhose, rabbit_ears, solo, cleavage, gloves, jewelry, playboy_bunny, smile, leotard, official_alternate_costume, cape, circlet, easter_egg, open_mouth, thighs |
| 5 | 5 |  |  |  |  |  | 1girl, looking_at_viewer, solo, cleavage, navel, smile, bare_shoulders, blue_sky, choker, cloud, collarbone, day, ocean, outdoors, purple_bikini, thighs, water, alternate_costume, bikini_pull, closed_mouth, jewelry, thigh_strap, tongue_out, twitter_username, wading |
| 6 | 5 |  |  |  |  |  | 2girls, yuri, closed_eyes, french_kiss, nail_polish, nude, blush, short_hair, black_nails, jewelry, nipples, saliva, sweat, tongue_out |

### Table Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1boy | 1girl | hetero | solo_focus | jewelry | nipples | penis | blush | cum_in_pussy | open_mouth | sex | vaginal | navel | thighhighs | ahegao | heart | spread_legs | sweat | cowgirl_position | tongue_out | brown_eyes | completely_nude | girl_on_top | saliva | uncensored | mosaic_censoring | fellatio | cum_in_mouth | dark-skinned_male | interracial | nude | circlet | gloves | multiple_penises | 2boys | double_penetration | mmf_threesome | spitroast | testicles | black_gloves | gangbang | gloved_handjob | pregnant | pussy_juice | cleavage | solo | cape | looking_at_viewer | smile | simple_background | dress | white_background | fake_animal_ears | pantyhose | rabbit_ears | playboy_bunny | leotard | official_alternate_costume | easter_egg | thighs | bare_shoulders | blue_sky | choker | cloud | collarbone | day | ocean | outdoors | purple_bikini | water | alternate_costume | bikini_pull | closed_mouth | thigh_strap | twitter_username | wading | 2girls | yuri | closed_eyes | french_kiss | nail_polish | short_hair | black_nails |
|----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:-------|:--------|:---------|:-------------|:----------|:----------|:--------|:--------|:---------------|:-------------|:------|:----------|:--------|:-------------|:---------|:--------|:--------------|:--------|:-------------------|:-------------|:-------------|:------------------|:--------------|:---------|:-------------|:-------------------|:-----------|:---------------|:--------------------|:--------------|:-------|:----------|:---------|:-------------------|:--------|:---------------------|:----------------|:------------|:------------|:---------------|:-----------|:-----------------|:-----------|:--------------|:-----------|:-------|:-------|:--------------------|:--------|:--------------------|:--------|:-------------------|:-------------------|:------------|:--------------|:----------------|:----------|:-----------------------------|:-------------|:---------|:-----------------|:-----------|:---------|:--------|:-------------|:------|:--------|:-----------|:----------------|:--------|:--------------------|:--------------|:---------------|:--------------|:-------------------|:---------|:---------|:-------|:--------------|:--------------|:--------------|:-------------|:--------------|
| 0 | 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 | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 1 | 12 |  |  |  |  |  | X | X | X | X | X | X | X | X | | | | | | X | | X | | | | | | | | | | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 2 | 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 | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 3 | 38 |  |  |  |  |  | | X | | | X | | | | | | | | | X | | | | | | | | | | | | | | | | | | X | | | | | | | | X | | | | | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 4 | 6 |  |  |  |  |  | | X | | | X | | | | | X | | | | | | | | | | | | | | | | | | | | | | X | X | | | | | | | | | | | | X | X | X | X | X | | | | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | |
| 5 | 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 | | | | | | | |
| 6 | 5 |  |  |  |  |  | | | | | 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

数据集基本信息

• 名称: Dataset of sonya (Fire Emblem)
• 许可证: MIT
• 任务类别: text-to-image
• 标签: art, not-for-all-audiences
• 大小类别: n<1K

数据集内容

• 图像数量: 250张
• 核心标签: purple_hair, long_hair, breasts, large_breasts, earrings, purple_eyes
• 图像来源: 从多个网站爬取，包括danbooru, pixiv, zerochan等

数据集包列表

名称	图像数量	大小	类型	描述
raw	250	341.81 MiB	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）
800	250	182.66 MiB	IMG+TXT	短边不超过800像素的数据集
stage3-p480-800	590	366.11 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素
1200	250	296.80 MiB	IMG+TXT	短边不超过1200像素的数据集
stage3-p480-1200	590	535.36 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素

标签聚类结果

原始文本版本

#	样本数量	图像示例	标签
0	14		1boy, 1girl, hetero, solo_focus, jewelry, nipples, penis, blush, cum_in_pussy, open_mouth, sex, vaginal, navel, thighhighs, ahegao, heart, spread_legs, sweat, cowgirl_position, tongue_out, brown_eyes, completely_nude, girl_on_top, saliva, uncensored
1	12		1girl, 1boy, blush, hetero, mosaic_censoring, penis, solo_focus, fellatio, cum_in_mouth, heart, jewelry, dark-skinned_male, interracial, nipples, nude, circlet, gloves, thighhighs
2	5		1girl, fellatio, hetero, multiple_penises, solo_focus, 2boys, double_penetration, jewelry, mmf_threesome, nipples, uncensored, blush, completely_nude, navel, spitroast, spread_legs, testicles, vaginal, black_gloves, dark-skinned_male, gangbang, gloved_handjob, interracial, pregnant, pussy_juice, thighhighs
3	38		jewelry, 1girl, cleavage, solo, cape, circlet, looking_at_viewer, smile, simple_background, black_gloves, dress, thighhighs, white_background
4	6		1girl, fake_animal_ears, looking_at_viewer, pantyhose, rabbit_ears, solo, cleavage, gloves, jewelry, playboy_bunny, smile, leotard, official_alternate_costume, cape, circlet, easter_egg, open_mouth, thighs
5	5		1girl, looking_at_viewer, solo, cleavage, navel, smile, bare_shoulders, blue_sky, choker, cloud, collarbone, day, ocean, outdoors, purple_bikini, thighs, water, alternate_costume, bikini_pull, closed_mouth, jewelry, thigh_strap, tongue_out, twitter_username, wading
6	5		2girls, yuri, closed_eyes, french_kiss, nail_polish, nude, blush, short_hair, black_nails, jewelry, nipples, saliva, sweat, tongue_out

搜集汇总

数据集介绍

构建方式

在动漫与游戏角色图像数据集构建领域，针对《火焰之纹章》系列角色索尼娅，本数据集通过自动化爬取系统从Danbooru、Pixiv、Zerochan等多个知名图像站点收集了250幅图像及其对应的标签信息。该爬取系统由DeepGHS团队开发，确保了数据来源的广泛性与多样性。数据集的核心标签涵盖紫色长发、耳饰、紫色眼眸等角色特征，并在预处理阶段对这些标签进行了精简处理。为满足不同应用场景，数据集提供了多种分辨率版本，包括短边不超过800像素与1200像素的标准版，以及经过三阶段裁剪、面积不低于480×480像素的增强版，以适配不同模型训练需求。

特点

该数据集呈现出鲜明的多层次特点。在标签结构上，通过聚类分析揭示了角色在不同情境下的视觉表现，从单人肖像到多人互动场景均有覆盖，展现了角色形象的丰富性。标签体系不仅包含角色外貌特征，还涵盖了服装变体、场景元素及互动关系等维度，形成了结构化的语义标注。值得注意的是，数据集包含了从正常着装到成人内容的多样化图像，并提供了马赛克审查与未审查两种版本，体现了对内容敏感性的考量。这种多维度的标签聚类为角色风格迁移、服装变体生成等任务提供了细粒度的数据支持。

使用方法

数据集的使用方式灵活多样。对于原始数据，推荐通过Waifuc库进行加载，用户可从HuggingFace Hub下载压缩包后，使用LocalSource接口快速读取图像及其元数据，包括文件名与标签信息。针对不同训练需求，用户可直接下载预处理好的图像-标签配对版本，如800像素或1200像素的缩略版，或采用三阶段裁剪的高质量版本。这些版本无需额外处理即可直接输入到文本到图像生成模型中进行训练或微调。此外，数据集提供的标签聚类结果可用于挖掘角色在不同装扮下的视觉模式，为风格迁移或角色定制化生成提供参考。

背景与挑战

背景概述

在生成式人工智能与数字艺术交汇的浪潮中，面向特定虚拟角色的文本到图像数据集成为推动模型精细化创作的关键资源。由DeepGHS团队于近期构建的CyberHarem/sonya_fireemblem数据集，聚焦于《火焰之纹章》系列中索尼娅这一角色，系统收录了250张经过多源爬取的高质量图像及其标签信息。该数据集的核心研究问题在于如何通过精细化的视觉特征标签（如紫色长发、紫色眼眸、耳饰等）与多尺度图像包（raw、800、1200像素及三级裁剪版本），为动漫角色定制化生成模型提供标准化的训练素材。其影响力体现在为角色一致性生成、风格迁移及跨模态检索等领域奠定了数据基础，同时推动了Waifuc框架在自动标注与数据预处理方面的应用。

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：动漫角色生成需在保持角色身份特征（如服饰、发型、配饰）的同时，适应多样化的场景与姿态变化，这对标签的细粒度与完整性提出了极高要求。在构建过程中，团队遭遇了数据源异构性的难题——从Danbooru、Pixiv等平台爬取的图像在分辨率、画风及版权标注上差异显著，需通过自动裁剪与标签清洗实现标准化。此外，数据集中包含的敏感内容（如成人向聚类）增加了伦理审查与合规使用的难度，而不足千张的样本量也限制了模型在复杂交互场景下的泛化能力，亟需通过数据增强或迁移学习策略加以弥补。

常用场景

经典使用场景

在文本到图像生成领域，CyberHarem/sonya_fireemblem 数据集常被用于角色一致性生成与概念微调。研究者利用其包含的250张高标注度图像，结合诸如Stable Diffusion等扩散模型，通过低秩适应（LoRA）或DreamBooth等技术，对特定动漫角色进行个性化定制。该数据集提供了多种分辨率与裁剪版本的图像-标签对，支持从原始元数据到标准化输入的灵活加载，使得模型能够精准学习角色外观与风格，从而生成保持角色身份特征的新颖图像。

解决学术问题

该数据集有效解决了动漫角色生成中的身份保持难题，即如何从有限样本中提取并泛化角色核心视觉概念。学术研究中，它常用于探索少样本学习下的概念解耦与风格迁移，评估模型在维持角色关键属性（如发色、瞳色、服饰）不变的前提下，生成多样化姿态与场景的鲁棒性。其贡献在于为动漫图像生成领域提供了一个标准化、标签丰富的基准，推动了可控图像合成与个性化生成技术的进步。

衍生相关工作

该数据集衍生了一系列经典工作，包括基于标签聚类的角色概念挖掘研究，如通过分析图像聚类结果（如不同服饰与姿势组）揭示角色视觉变体分布。此外，它被用于评估图像-标签对齐方法的有效性，推动了如Waifuc等自动化数据爬取与处理工具的发展。相关工作还涉及跨数据集的角色迁移学习，以及利用多分辨率版本进行图像质量与生成效率的权衡分析，为大规模动漫数据集构建提供了方法论参考。