Learning View-Dependent Splatting Kernels 事件

Learning View-Dependent Splatting Kernels arXiv:2605.25426v1 Announce Type: cross Abstract: We present a differentiable framework to automatically learn view-dependent 2D kernels in a splatting-based pipeline to improve reconstruction quality and representation efficiency for novel 3D view synthesis. Our volumetric primitive is defined as a bounding ellipsoid and a 3D-kernel latent vector. We first learn a projection network to output a 2D-kernel latent, taking the attributes of the ellipsoid a