Research map / 3D vision / Gaussian splatting
Surfel-Based Gaussian Inverse Rendering for Fast and Relightable Dynamic Human Reconstruction From Monocular Videos
Yiqun Zhao, Chenming Wu, Binbin Huang, Yihao Zhi, Chen Zhao, Jingdong Wang, Shenghua Gao
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 |
|---|---|---|---|
| update Gaussian shape during PBR stage instead of freezing it after stage I (ablation) | SGIA · RANA ablation | albedo PSNR: 18.78 -> 23.17 | Improved |
| add progressive normal-consistency optimization (ablation) | SGIA · RANA ablation | normal error (degrees): 12.89 -> 11.31 | Improved |
| add progressive normal-consistency optimization (ablation) | SGIA · RANA ablation | albedo PSNR: 20.21 -> 23.17 | Improved |
| use 2DGS surfels instead of 3DGS with learnable normals (ablation) | SGIA · RANA ablation | normal error (degrees): 12.33 -> 11.71 | Improved |
| use cosine-similarity normal consistency instead of l1 loss (ablation) | SGIA · RANA ablation | normal error (degrees): 20.33 -> 11.71 | Improved |
| use cosine-similarity normal consistency instead of l1 loss (ablation) | SGIA · RANA ablation | relighting PSNR: 15.87 -> 19.77 | Improved |
| add template-mesh ambient occlusion (ablation) | SGIA · RANA ablation | albedo PSNR: 21.08 -> 21.31 | Improved |
| add template-mesh ambient occlusion (ablation) | SGIA · RANA ablation | normal error (degrees): 11.87 -> 10.69 | Improved |
| bake occlusion from SMPL template mesh instead of full clothed mesh (ablation) | SGIA · RANA ablation | rendering time: 2s -> 0.2s | Improved |
| bake occlusion from SMPL template mesh instead of full clothed mesh (ablation) | SGIA · RANA ablation | albedo PSNR: 22.01 -> 21.93 | No clear effect |
| add white-light regularization (ablation) | SGIA · RANA ablation | albedo PSNR: 22.05 -> 22.44 | Improved |
| add smoothness regularization on materials (ablation) | SGIA · RANA ablation | albedo PSNR: 21.83 -> 22.44 | Improved |
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.
- IntrinsicAvatar — IntrinsicAvatar: Physically Based Inverse Rendering of Dynamic Humans from Monocular Videos via Explicit Ray Tracing (CVPR 2024) · better on 5 of 8 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
- IntrinsicAvatar: Physically Based Inverse Rendering of Dynamic Humans from Monocular Videos via Explicit Ray Tracing (CVPR 2024 · similarity 0.89)
- Human Gaussian Splatting: Real-time Rendering of Animatable Avatars (CVPR 2024 · similarity 0.87)
- Real-time High-fidelity Gaussian Human Avatars with Position-based Interpolation of Spatially Distributed MLPs (CVPR 2025 · similarity 0.87)
- 3DGS-Avatar: Animatable Avatars via Deformable 3D Gaussian Splatting (CVPR 2024 · similarity 0.87)
- Expressive Gaussian Human Avatars from Monocular RGB Video (NeurIPS 2024 · similarity 0.87)
- GAF: Gaussian Avatar Reconstruction from Monocular Videos via Multi-view Diffusion (CVPR 2025 · similarity 0.86)
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.
