ReNe (Relighting NeRF)
收藏资源简介:
ReNe数据集是由University of Bologna创建的,专注于真实世界物体的新视角合成和重光照研究。该数据集包含20个场景,总计40000张图像,每个场景有2000张图像,从50个不同视点和40种不同的光照条件下拍摄。数据集的创建过程利用了两个机器人臂,一个持相机,另一个持光源,以确保精确的相机和光源位置。ReNe数据集适用于研究光照条件下的新视角合成,旨在解决逆渲染中的复杂问题,如物体的几何形状、材质特性和光照条件的估计,这些问题的有效解决对于游戏、机器人和增强现实等领域具有重要价值。
The ReNe Dataset was developed by the University of Bologna, focusing on novel view synthesis and relighting research for real-world objects. It contains a total of 40,000 images across 20 scenes, with 2,000 images per scene captured from 50 distinct viewpoints and under 40 different lighting conditions. The dataset was constructed using two robotic arms: one mounted with a camera and the other with a light source, to ensure precise positioning of both the camera and the light source. The ReNe Dataset is suitable for research on novel view synthesis under varying lighting conditions, aiming to address complex problems in inverse rendering, such as the estimation of object geometry, material properties, and lighting conditions. Effective solutions to these issues hold significant value for fields including gaming, robotics, and augmented reality.




