AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

Urban Radiance Field Representation with Deformable Neural Mesh Primitives

Fan Lu, Yan Xu, Guang Chen, Hongsheng Li, Kwan-Yee Lin, Changjun Jiang

ICCV 2023 · 3D vision › NeRF and radiance fields · approach: Dynamic and editable NeRF · 55 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
optimize DNMP shapes via a pretrained latent space instead of direct vertex offsets (ablation)DNMP · KITTI-360PSNR: 21.41 -> 23.41Improved
add shape optimization of primitives (ablation)DNMP · KITTI-360PSNR: 20.36 -> 23.41Improved
add hierarchical DNMPs (0.5m and 1m) (ablation)DNMP · KITTI-360LPIPS: 0.322 -> 0.305Improved
increase intersection points J from 2 to 8 (ablation)DNMP · KITTI-360PSNR: 23.23 -> 23.43No clear effect
vary radiance feature dimension (15, 21, 27) (ablation)DNMP · KITTI-360PSNR: 23.21 / 23.41 / 23.28No clear effect
use larger DNMP radius of 2 m instead of 0.5 m (ablation)DNMP · KITTI-360PSNR: 23.21 -> 22.61Hurt
use lightweight MLP version (ablation)DNMP · KITTI-360PSNR: 23.41 -> 23.27No clear effect
DNMP versus Mip-NeRF 360 for urban view synthesisMip-NeRF 360 · KITTI-360LPIPS: 0.355 -> 0.305Improved
DNMP versus Mip-NeRF 360 on WaymoMip-NeRF 360 · Waymo Open DatasetSSIM: 0.893 -> 0.892Mixed
explicit surface modeling with DNMP for 45 degree rotated viewsMip-NeRF · KITTI-360 rotated viewsPSNR: 14.33 -> 15.25 (left)Improved
rasterization-based rendering instead of volumetric sampling (Mip-NeRF 360)Mip-NeRF 360 · Efficiency (A100)rendering time (ms/1K pixel): 10.39 -> 2.07Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy