SyntheticFur

SyntheticFur数据集是由谷歌创建的，专门用于机器学习训练的高质量合成毛皮数据集。该数据集包含约140,000张通过光线追踪生成的合成毛皮渲染图像，以及相应的栅格化输入缓冲区和模拟数据文件。数据集中的图像展示了不同皮肤基元和预定义光照环境下的各种运动。创建过程利用Houdini软件进行程序化生成，旨在通过神经渲染技术显著提升毛皮图形的质量，解决实时应用中毛皮渲染的复杂性和高成本问题。

The SyntheticFur dataset, developed by Google, is a high-quality synthetic fur dataset specifically designed for machine learning training. It contains approximately 140,000 ray-traced synthetic fur rendered images, along with corresponding rasterized input buffers and simulation data files. The images in the dataset showcase various motions under different skin primitives and predefined lighting environments. The dataset was created using Houdini software for procedural generation, aiming to significantly improve the quality of fur graphics through neural rendering techniques and address the complexity and high costs of fur rendering in real-time applications.