AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

Disentangled Generation and Aggregation for Robust Radiance Fields

Shihe Shen, Huachen Gao, Wangze Xu, Rui Peng, Luyang Tang, Kaiqiang Xiong, Jianbo Jiao, Ronggang Wang

ECCV 2024 · 3D vision › NeRF and radiance fields · approach: Single-view and streaming radiance fields · 2 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
Use triplane generator (deep network prior) instead of directly optimizing triplane grids (ablation)pose-triplane joint optimization · LLFF (unknown poses)PSNR: 21.47 -> 24.93Improved
Add Disentangled Plane Aggregation to baseline (ablation)pose-triplane baseline · LLFF (unknown poses)PSNR: 21.47 -> 23.41Improved
Add DPA on top of triplane generator (ablation)generator model · LLFF (unknown poses)PSNR: 24.93 -> 25.21Improved
Add two-stage warm-start training (ablation)generator + DPA · LLFF (unknown poses)PSNR: 25.21 -> 25.90Improved
Use DPA instead of product or sum aggregation (ablation)triplane aggregation · LLFF HornsPSNR: 21.64 -> 23.66 (vs product)Improved
Use sum instead of product aggregation (ablation)triplane aggregation · LLFF Hornspose error: rotation 1.002 -> 0.794; PSNR 21.64 -> 21.29Mixed
Use only first training stage (generator throughout) (ablation)full method · NeRF-Synthetic DrumsPSNR / training time: 26.10 -> 24.98; time 01:21 -> 04:00Hurt
Skip generator stage and optimize triplane directly (ablation)full method · NeRF-Synthetic DrumsPSNR: 26.10 -> 19.93Hurt
Disentangled generation and aggregation versus HASH-BARFjoint pose-NeRF optimization · LLFF (mean)PSNR: 24.79 -> 25.90Improved
Evaluation on NeRF-Synthetic Materials sceneOurs · NeRF-Synthetic MaterialsPSNR: 27.91 (GARF) vs 25.78Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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