NeRF-Editing: Geometry Editing of Neural Radiance Fields
NeRF‑Editing introduces an interactive framework that lets users freely deform the geometry of neural radiance fields by coupling an explicit mesh with implicit NeRF representations, propagating mesh vertex changes through tetrahedral ARAP optimization to bend rays during rendering, enabling realistic edits and animations on synthetic and real‑world scenes, a first reported at CVPR 2022.
This paper presents a method for freely editing the geometry of neural radiance fields (NeRF), allowing users to interactively deform 3D scenes reconstructed from images. The approach combines explicit mesh representation with implicit NeRF modeling, enabling intuitive geometric editing through control point manipulation.
The core innovation lies in propagating discrete mesh vertex deformations to continuous 3D space using tetrahedral meshes and ARAP (as-rigid-as-possible) energy minimization. This allows the system to bend rays during volume rendering, producing deformed scene renderings that satisfy user editing intentions.
Experimental results demonstrate successful editing on both synthetic and real-world scenes, including complex objects like Lego bulldozers, chairs, giraffes, dinosaurs, and laptops. The method enables creating animation sequences through interpolated deformations and represents the first approach supporting user-controlled geometric deformation of NeRF networks.
The research was conducted by teams from the Institute of Computing Technology, Chinese Academy of Sciences, and Alibaba's Taobao Technology Department, and published at IEEE CVPR 2022.
DaTaobao Tech
Official account of DaTaobao Technology
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.