Research map / 3D vision / Gaussian splatting
Total-Recon: Deformable Scene Reconstruction for Embodied View Synthesis
Chonghyuk Song, Gengshan Yang, Kangle Deng, Jun-Yan Zhu, Deva Ramanan
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 depth supervision loss (ablation) | Total-Recon · Ablation, 6 sequences | Acc@0.1m: .154 -> .898 | Improved |
| add depth supervision loss (ablation) | Total-Recon · Ablation, 6 sequences | LPIPS: .372 -> .268 | Improved |
| add deformation field on top of root-body motion (ablation) | Total-Recon · Ablation, 6 sequences | LPIPS: .296 -> .268 | Improved |
| add PoseNet initialization of root-body poses (ablation) | Total-Recon · Ablation, 6 sequences | LPIPS: .293 -> .268 | Improved |
| add explicit root-body pose modeling | Total-Recon · Ablation, 6 sequences | convergence | Improved |
| add depth supervision to HyperNeRF baseline | HyperNeRF · Stereo-view synthesis mean | Acc@0.1m: .198 -> .439 | Improved |
| add depth supervision to D2NeRF baseline | D2NeRF · Stereo-view synthesis mean | LPIPS: .611 -> .599 | No clear effect |
| hierarchical object-centric motion decomposition versus depth-supervised monocular deformable NeRFs | Total-Recon · Stereo-view synthesis mean | LPIPS: .428 -> .278 | Improved |
Later papers that compared with this method
2 later accepted papers report Total-Recon as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- HoliGS: Holistic Gaussian Splatting for Embodied View Synthesis (NeurIPS 2025 · better on 2 of 2 shared results)
- OSN: Infinite Representations of Dynamic 3D Scenes from Monocular Videos (ICML 2024 · better on 3 of 3 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
- Restage4D: Reanimating Deformable 3D Reconstruction from a Single Video (NeurIPS 2025 · similarity 0.89)
- HSR: Holistic 3D Human-Scene Reconstruction from Monocular Videos (ECCV 2024 · similarity 0.89)
- BANMo: Building Animatable 3D Neural Models From Many Casual Videos (CVPR 2022 · similarity 0.89)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.88)
- Reconstruct, Inpaint, Test-Time Finetune: Dynamic Novel-view Synthesis from Monocular Videos (NeurIPS 2025 · similarity 0.87)
- PKU-DyMVHumans: A Multi-View Video Benchmark for High-Fidelity Dynamic Human Modeling (CVPR 2024 · similarity 0.87)
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.
