Research map / 3D vision / Face and shape
Animatable Implicit Neural Representations for Creating Realistic Avatars From Videos
Xiaowei Zhou, Sida Peng, Zhen Xu, Junting Dong, Qianqian Wang, Shangzhan Zhang, Qing Shuai, Hujun Bao
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add pose-dependent displacement field to SMPL blend weights (ablation) | NeRF with SMPL blend weight field · Human3.6M novel poses | PSNR: 23.61 -> 23.78 | Improved |
| use neural blend weight field instead of SMPL blend weights (ablation) | NeRF with SMPL blend weight field · Human3.6M novel poses | PSNR: 23.61 -> 23.64 | No clear effect |
| use pose-dependent displacement field instead of neural blend weight field (ablation) | NeRF-NBW · Human3.6M novel poses | PSNR: 23.65 -> 23.78 | Improved |
| use signed distance field instead of density field (ablation) | NeRF-PDF · SyntheticHuman reconstruction | P2S: 1.78 -> 0.70 | Improved |
| use signed distance field instead of density field (ablation) | NeRF-PDF · Human3.6M image synthesis (novel poses) | PSNR: 23.78 -> 23.67 | No clear effect |
| use VolSDF rendering instead of NeuS rendering (ablation) | SDF-PDF · SyntheticHuman reconstruction | P2S: 0.79 -> 0.70 | Improved |
| use bigger color head network instead of original NeRF network (ablation) | NeRF-PDF · Human3.6M novel poses | PSNR: 22.55 -> 23.65 | Improved |
| train on 100 frames instead of 1 frame (ablation) | SDF-PDF · Human3.6M S9 | PSNR: 22.46 -> 24.14 | Improved |
| train on 800 frames instead of 200 frames (ablation) | SDF-PDF · Human3.6M S9 | PSNR: 24.68 -> 24.48 | Hurt |
| train on 3 views instead of 1 view (ablation) | SDF-PDF · Human3.6M S9 | PSNR: 24.21 -> 24.45 | No clear effect |
| use pose-dependent deformation instead of NB when trained on a single view | NB · ZJU-MoCap single view novel poses | PSNR: 20.32 -> 22.98 | Improved |
| use SDF-PDF instead of NB with 4 views | NB · ZJU-MoCap 4 views training poses | PSNR: 28.15 -> 27.45 | Hurt |
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.
- D-NeRF — D-NeRF: Neural Radiance Fields for Dynamic Scenes (CVPR 2021) · better on 2 of 2 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
- Animatable Neural Radiance Fields for Modeling Dynamic Human Bodies (ICCV 2021 · similarity 0.97)
- 3DGS-Avatar: Animatable Avatars via Deformable 3D Gaussian Splatting (CVPR 2024 · similarity 0.90)
- Pose Modulated Avatars from Video (ICLR 2024 · similarity 0.90)
- REACTO: Reconstructing Articulated Objects from a Single Video (CVPR 2024 · similarity 0.89)
- Relightable and Animatable Neural Avatar from Sparse-View Video (CVPR 2024 · similarity 0.88)
- AniGS: Animatable Gaussian Avatar from a Single Image with Inconsistent Gaussian Reconstruction (CVPR 2025 · similarity 0.88)
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.
