Research map / 3D vision / Face and shape
ARCH++: Animation-Ready Clothed Human Reconstruction Revisited
Tong He, Yuanlu Xu, Shunsuke Saito, Stefano Soatto, Tony Tung
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 |
|---|---|---|---|
| supervise in both posed and canonical space instead of posed only (ablation) | ARCH++ · Joint-space ablation | Mean Chamfer: 1.192 -> 0.825 | Improved |
| supervise in both spaces instead of canonical only (ablation) | ARCH++ · Joint-space ablation | Mean Chamfer: 0.917 -> 0.825 | Improved |
| joint supervision versus canonical only in posed space (ablation) | ARCH++ · Posed space | P2S: 0.716 -> 0.662 | Improved |
| add point-based geometry encoder to joint supervision (ablation) | ARCH++ · Joint-space ablation | Mean Chamfer: 0.825 -> 0.717 | Improved |
| add point-based geometry encoder to joint supervision (ablation) | ARCH++ · Joint-space ablation | Mean P2S: 0.641 -> 0.483 | Improved |
| add mesh refinement (moulding-inpainting) (ablation) | ARCH++ · Joint-space ablation | Mean Normal error: 0.036 -> 0.035 | Improved |
| add mesh refinement (moulding-inpainting) (ablation) | ARCH++ · Joint-space ablation | Mean Chamfer: 0.717 -> 0.717 | No clear effect |
| point-based encoder instead of RBF features (ablation) | ARCH++ · Geometry encoder ablation | Chamfer: 0.85 -> 0.61 | Improved |
| point-based encoder instead of voxel encoder (ablation) | ARCH++ · Geometry encoder ablation | Chamfer: 0.63 -> 0.61 | Improved |
| image-space input of normals instead of image-space regression (ablation) | ARCH++ · Normal refinement ablation | Mean normal error: 0.035 -> 0.035 | No clear effect |
| object-space normal regression instead of baseline (ablation) | ARCH++ · Normal refinement ablation | Mean normal error: 0.037 -> 0.037 | No clear effect |
| add adversarial loss to normal map estimation (ablation) | Image translation network · Normal map estimation | performance | Hurt |
Later papers that compared with this method
4 later accepted papers report ARCH++ as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- FRESA: Feedforward Reconstruction of Personalized Skinned Avatars from Few Images (CVPR 2025 · better on 6 of 6 shared results)
- DINAR: Diffusion Inpainting of Neural Textures for One-Shot Human Avatars (ICCV 2023 · better on 1 of 1 shared result)
- S-PIFu: Integrating Parametric Human Models with PIFu for Single-view Clothed Human Reconstruction (NeurIPS 2022 · better on 5 of 5 shared results)
- Photorealistic Monocular 3D Reconstruction of Humans Wearing Clothing (CVPR 2022 · better on 4 of 4 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
- ARCH: Animatable Reconstruction of Clothed Humans (CVPR 2020 · similarity 0.96)
- DLCA-Recon: Dynamic Loose Clothing Avatar Reconstruction from Monocular Videos (AAAI 2024 · similarity 0.88)
- FRESA: Feedforward Reconstruction of Personalized Skinned Avatars from Few Images (CVPR 2025 · similarity 0.88)
- D^3-Human: Dynamic Disentangled Digital Human from Monocular Video (CVPR 2025 · similarity 0.88)
- VS: Reconstructing Clothed 3D Human from Single Image via Vertex Shift (CVPR 2024 · similarity 0.88)
- GALA: Generating Animatable Layered Assets from a Single Scan (CVPR 2024 · 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.
