Gaussian Primitive
Gaussian primitives are 3D representations, often Gaussian functions, used in novel view synthesis and scene reconstruction to model objects and scenes efficiently. Current research focuses on improving the accuracy and efficiency of these models, particularly within the Gaussian splatting framework, by addressing issues like memory footprint, primitive pruning strategies, and the incorporation of additional data sources (e.g., LiDAR, depth priors) to enhance reconstruction quality from sparse viewpoints. This work is significant because it enables real-time rendering of high-fidelity 3D scenes, with applications ranging from virtual and augmented reality to robotics and 3D modeling software.
Papers
RetinaGS: Scalable Training for Dense Scene Rendering with Billion-Scale 3D Gaussians
Bingling Li, Shengyi Chen, Luchao Wang, Kaimin Liao, Sijie Yan, Yuanjun Xiong
Effective Rank Analysis and Regularization for Enhanced 3D Gaussian Splatting
Junha Hyung, Susung Hong, Sungwon Hwang, Jaeseong Lee, Jaegul Choo, Jin-Hwa Kim
LetsGo: Large-Scale Garage Modeling and Rendering via LiDAR-Assisted Gaussian Primitives
Jiadi Cui, Junming Cao, Fuqiang Zhao, Zhipeng He, Yifan Chen, Yuhui Zhong, Lan Xu, Yujiao Shi, Yingliang Zhang, Jingyi Yu
CompGS: Efficient 3D Scene Representation via Compressed Gaussian Splatting
Xiangrui Liu, Xinju Wu, Pingping Zhang, Shiqi Wang, Zhu Li, Sam Kwong