PRISM-36K: A Benchmark Dataset for AI-Generated Image Attribution

PRISM-36K: A Benchmark Dataset for AI-Generated Image Attribution PRISM-36K is a benchmark dataset of 36,000 AI-generated images for model-attribution research — the task of identifying which generative modelproduced a given image.It accompanies the paper "PRISM: Phase-enhanced Radial-based Image Signature Mapping for AI-Generated Image Attribution" (Ricco, Onofri, Cima, Cresci, Di Pietro; arXiv:2509.15270). What is in the dataset The dataset contains 36,000 PNG images at 512 × 512 pixels, balanced across six text-to-image generators with 6,000 images per model: DALL-E 2 (Ramesh et al., 2022) — closed, accessed via OpenAI API FuseDream (Liu et al., 2021) — GAN + CLIP guidance PixArt-α (Chen et al., 2024) — diffusion transformer SANA (Xie et al., 2024) — diffusion transformer Stable Diffusion 1.4 (Rombach et al., 2022) — latent diffusion VQGAN-CLIP (Esser et al., 2021) — GAN + CLIP guidance Each generator produces 150 images per prompt over a fixed set of 40 author-written English prompts (20 short + 20 long, paired by topic).All images are stored in lossless PNG format to preserve frequency-domain artefacts that are critical to spectral attribution methods. What makes this dataset useful Prompt-matched generations. The same 40 prompts are issued to every generator, so cross-model differences reflect generator-specific signatures rather than prompt drift. Architectural diversity. The six generators span GAN-based, CLIP-guided, and transformer-based diffusion families, with both open-weight and closed-API systems represented. Reproducible splits. 100 random prompt-level train/test splits used in the paper are shipped as splits/splits_100.csv; one canonical "average split" (splits/average_split.json) is provided for direct reproduction of all figures and tables. Lossless integrity. Every image ships with a SHA-256 hash in checksums/SHA256SUMS (BSD-style, compatible with sha256sum -c) so users can verify their downloads. Rich metadata. Per-image manifest (metadata/images.csv) and prompt manifest (metadata/prompts.csv) support filtering by model, prompt length, prompt pair, or specific generation iteration. Repository layout PRISM-36K/├── README.md├── LICENSE.txt├── CITATION.cff├── CHANGELOG.md├── metadata/│ ├── prompts.csv│ └── images.csv├── splits/│ ├── average_split.json│ └── splits_100.csv├── images/│ ├── DALLE-2/│ ├── FuseDream/│ ├── PixArt-alpha/│ ├── SANA/│ ├── StableDiffusion-1.4/│ └── VQGAN-CLIP/└── checksums/ └── SHA256SUMS Image filename convention: <ModelName>_<promptid>_<iter>.png, with promptid ∈ 1..40 and iter ∈ 1..150. Intended uses Training and evaluating model-attribution classifiers for AI-generated images. Benchmarking real vs. fake detectors in a controlled multi-source setting. Studying frequency-domain and spectral fingerprints of generative models. Research on content provenance, generative-AI accountability, and related forensic problems. Companion resources Paper: arXiv:2509.15270 Image-generation scripts (the code used to produce these images): github.com/emarich18-res/PRISM-36K PRISM classifier and evaluation code: released upon full paper acceptance. Licensing Dataset (images and metadata): Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0). Note on DALL-E 2 images. The 6,000 images in images/dalle2/ were generated via OpenAI's paid API and are subject to OpenAI's usage policies in addition to CC BY 4.0: users intending to use these images beyond academic research should consult OpenAI's current terms of service. Note on NVIDIA-SANA images. The 6,000 images in images/sana/ are licensed under the Apache License 2.0 usage policies in addition to CC BY 4.0. Citing PRISM-36K If you use this dataset, please cite both the paper and this Zenodo record. BibTeX entries and a CFF citation file are provided in the repository (README.md, CITATION.cff). Limitations Closed-set scope. The dataset covers six specific generators; it is not designed to support open-set attribution to unseen models. English-only prompts authored by the dataset creators; no multilingual or in-the-wild prompts are included. Synthetic only. No real photographs are included; for real vs. fake benchmarks, real images must be sourced from a complementary dataset. No identifiable individuals. Prompts were authored to elicit generic scenes (objects, animals, landscapes); the dataset contains no images of identifiable real persons by design.# PRISM-36K：用于AI生成图像归因的基准数据集 PRISM-36K是一款专为**模型归因（model-attribution）**研究打造的基准数据集，包含36000张AI生成图像——该任务的目标为识别给定图像由哪一款生成式模型创作。本数据集配套论文《PRISM：面向AI生成图像归因的相位增强径向图像特征映射》（Ricco、Onofri、Cima、Cresci、Di Pietro；arXiv:2509.15270）。 ## 数据集内容 本数据集包含36000张分辨率为512×512像素的PNG格式无损图像，在6款文本到图像生成器（text-to-image generator）间均衡分布，每个模型对应6000张图像： 1. DALL-E 2（Rames等人，2022）——闭源模型，通过OpenAI API访问 2. FuseDream（Liu等人，2021）——基于生成对抗网络（GAN）+ CLIP引导 3. PixArt-α（Chen等人，2024）——扩散Transformer 4. SANA（Xie等人，2024）——扩散Transformer 5. Stable Diffusion 1.4（Rombach等人，2022）——潜在扩散模型（latent diffusion） 6. VQGAN-CLIP（Esser等人，2021）——基于生成对抗网络（GAN）+ CLIP引导 每款生成器针对固定的40条人工撰写的英文提示词（分为20条短提示与20条长提示，按主题配对），每条提示词生成150张图像。所有图像均以无损PNG格式存储，以保留对光谱归因方法至关重要的频域伪影。 ## 数据集核心优势 1. **提示词匹配生成**：所有生成器均使用同一套40条提示词，因此跨模型差异反映的是生成器专属的特征印记，而非提示词漂移带来的偏差。 2. **架构多样性**：6款生成器涵盖了基于生成对抗网络、CLIP引导以及基于Transformer的扩散模型家族，同时包含开源权重与闭源API系统两类范式。 3. **可复现划分方案**：论文中使用的100组随机提示级训练/测试划分已存储于`splits/splits_100.csv`；同时提供标准的“平均划分”版本（`splits/average_split.json`），可直接复现所有图表与表格结果。 4. **无损完整性校验**：每张图像均附带SHA-256哈希值，存储于`checksums/SHA256SUMS`（BSD格式，兼容`sha256sum -c`命令），使用者可验证下载文件的完整性。 5. **丰富元数据**：图像级清单（`metadata/images.csv`）与提示词级清单（`metadata/prompts.csv`）支持按模型、提示词长度、提示词配对或具体生成迭代次数进行筛选。 ## 仓库布局 PRISM-36K/ ├── README.md ├── LICENSE.txt ├── CITATION.cff ├── CHANGELOG.md ├── metadata/ │ ├── prompts.csv │ └── images.csv ├── splits/ │ ├── average_split.json │ └── splits_100.csv ├── images/ │ ├── DALLE-2/ │ ├── FuseDream/ │ ├── PixArt-alpha/ │ ├── SANA/ │ ├── StableDiffusion-1.4/ │ └── VQGAN-CLIP/ └── checksums/ └── SHA256SUMS 图像文件名命名规则：`<ModelName>_<promptid>_<iter>.png`，其中`promptid`取值范围为1~40，`iter`取值范围为1~150。 ## 预期用途 1. 训练与评估针对AI生成图像的模型归因分类器 2. 在可控多源场景下对真伪图像检测器进行基准测试 3. 研究生成式模型的频域与光谱指纹特征 4. 开展内容溯源、生成式AI问责制及相关取证问题的研究 ## 配套资源 1. 配套论文：arXiv:2509.15270 2. 图像生成脚本（用于生成本数据集图像的代码）：github.com/emarich18-res/PRISM-36K 3. PRISM分类器与评估代码：将在论文正式录用后公开 ## 授权协议 数据集（含图像与元数据）采用知识共享署名-非商业性使用-相同方式共享4.0国际许可协议（CC BY-NC-SA 4.0）。 > 关于DALL-E 2图像的说明：`images/dalle2/`目录下的6000张图像通过OpenAI付费API生成，除遵循CC BY 4.0协议外，同时受OpenAI使用条款约束；若使用者计划将这些图像用于学术研究以外的场景，请查阅OpenAI当前的服务条款。 > 关于NVIDIA-SANA图像的说明：`images/sana/`目录下的6000张图像除遵循CC BY 4.0协议外，同时受Apache License 2.0使用条款约束。 ## 引用本数据集 若您使用本数据集，请同时引用该论文与本Zenodo记录。BibTeX引用条目与CFF引用文件已在仓库中提供（详见`README.md`与`CITATION.cff`）。 ## 局限性说明 1. **闭集范围**：本数据集仅覆盖6款特定生成器，未设计用于对未见过的模型开展开集归因任务。 2. **仅英文提示词**：提示词均由数据集创作者撰写的英文内容，未包含多语言或真实场景下的提示词。 3. **仅合成图像**：未包含真实摄影图像；若需开展真伪图像基准测试，需从其他互补数据集获取真实图像。 4. **无可识别个人信息**：提示词均设计为生成通用场景（物体、动物、风景），本数据集未包含任何可识别的真实人物图像。