🏠 Project Page | 💻 GitHub | 📖 Paper | 🤗 MultiRef-Benchmark
MultiRef-Dataset is a large-scale dataset of over 38,000 samples for training controllable image generation models to use multiple visual references.
The MultiRef-Dataset was created to facilitate research and development of generative models that can synthesize new images by combining elements from various visual sources. All 38,076 samples are synthetically generated using… See the full description on the dataset page:
https://huggingface.co/datasets/ONE-Lab/MultiRef-dataset.
