InFusion: Inpainting 3D Gaussians via Learning Depth Completion from Diffusion Prior

arXiv:2404.11613v1 Announce Type: new
Abstract: 3D Gaussians have recently emerged as an efficient representation for novel view synthesis. This work studies its editability with a particular focus on the inpainting task, which aims to supplement an incomplete set of 3D Gaussians with additional points for visually harmonious rendering. Compared to 2D inpainting, the crux of inpainting 3D Gaussians is to figure out the rendering-relevant properties of the introduced points, whose optimization largely benefits from their initial 3D positions. To this …

abstract arxiv cs.cv diffusion focus inpainting novel prior rendering representation set studies synthesis type via view work