DiffusionDB: A Large-scale Prompt Gallery Dataset for Text-to-Image Generative Models
This dataset enables research into prompt-model interactions, deepfake detection, and human-AI tools for text-to-image models, but it is incremental as it provides new data rather than a novel method.
The authors introduced DiffusionDB, a 6.5TB dataset with 14 million images and 1.8 million unique prompts from Stable Diffusion, to address the challenge of understanding how text prompts affect image generation in diffusion models. They analyzed prompt characteristics, identified hyperparameters and styles causing errors, and highlighted potential misuse like misinformation generation.
With recent advancements in diffusion models, users can generate high-quality images by writing text prompts in natural language. However, generating images with desired details requires proper prompts, and it is often unclear how a model reacts to different prompts or what the best prompts are. To help researchers tackle these critical challenges, we introduce DiffusionDB, the first large-scale text-to-image prompt dataset totaling 6.5TB, containing 14 million images generated by Stable Diffusion, 1.8 million unique prompts, and hyperparameters specified by real users. We analyze the syntactic and semantic characteristics of prompts. We pinpoint specific hyperparameter values and prompt styles that can lead to model errors and present evidence of potentially harmful model usage, such as the generation of misinformation. The unprecedented scale and diversity of this human-actuated dataset provide exciting research opportunities in understanding the interplay between prompts and generative models, detecting deepfakes, and designing human-AI interaction tools to help users more easily use these models. DiffusionDB is publicly available at: https://poloclub.github.io/diffusiondb.