AI Paper Atlas

Research map / 3D vision / Face and shape

ActorsNeRF: Animatable Few-shot Human Rendering with Generalizable NeRFs

Jiteng Mu, Shen Sang, Nuno Vasconcelos, Xiaolong Wang

ICCV 2023 · 3D vision › Face and shape · approach: Full-body avatar reconstruction · 31 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
add category-level prior with 2-level canonical space (ActorsNeRF) over HumanNeRFfew-shot human NeRF · ZJU-MoCap Person 387, 5-shotLPIPS (x10^3): 65.64 -> 46.06Improved
add category-level prior (ActorsNeRF) over HumanNeRFfew-shot human NeRF · AIST++ Person 16, 5-shotPSNR: 24.37 -> 25.22Improved
increase number of support frames from 5 to 300 (ablation)ActorsNeRF · ZJU-MoCap Person 387LPIPS (x10^3): 46.06 -> 36.18Improved
ActorsNeRF vs baselines at 300-shot on AIST++ActorsNeRF · AIST++ Person 20, 300-shotPSNR: 26.00 (HumanNeRF) vs 25.78No clear effect
ActorsNeRF vs NeuralBody on PSNR at few shotsActorsNeRF · ZJU-MoCap Person 393, 5-shotPSNR: 27.54 (NeuralBody) vs 27.20Mixed
use ActorsNeRF in short-video setting with partial body coverageActorsNeRF · ZJU-MoCap Person 387, short-videoLPIPS (x10^3): 81.82 -> 50.88Improved
add deformation network (instance-level canonical space) (ablation)ActorsNeRF · ZJU-MoCap 10-shot, Person 393LPIPS (x10^3): 44.13 -> 42.56Improved
add deformation network (ablation)ActorsNeRF · ZJU-MoCap 10-shot, Person 387PSNR: 27.33 -> 27.15No clear effect
add pixel-aligned local features (ablation)ActorsNeRF · ZJU-MoCap 10-shot, Person 393LPIPS (x10^3): 42.77 -> 42.56No clear effect
add SMPL local features (ablation)ActorsNeRF · ZJU-MoCap 10-shot, Person 394LPIPS (x10^3): 37.13 -> 35.93Improved

Later papers that compared with this method

1 later accepted paper reports ActorsNeRF as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

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