Research map / 3D vision / Face and shape
Large-scale Codec Avatars: The Unreasonable Effectiveness of Large-scale Avatar Pretraining
Junxuan Li, Rawal Khirodkar, Egor Zakharov, Jihyun Lee, Zhaoen Su, Yuan Dong, Julieta Martinez, Kai Li, Qingyang Tan, Takaaki Shiratori, Matthew Hu, Peihong Guo and 28 more
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 |
|---|---|---|---|
| pretrain on in-the-wild data then post-train on studio data, instead of mixing both (ablation) | LCA · Capture-Studio test | PSNR: 30.025 -> 30.514 | Improved |
| pretrain on in-the-wild data then post-train on studio data, instead of mixing both (ablation) | LCA · In-the-Wild test | PSNR: 27.998 -> 28.175 | Improved |
| add in-the-wild data (mixed or pretraining) to studio-only training (ablation) | LCA · In-the-Wild test | PSNR: 26.632 -> 28.175 | Improved |
| add in-the-wild pretraining to studio-only training, evaluated on studio domain (ablation) | LCA · Capture-Studio test | PSNR: 29.741 -> 30.514 | Improved |
| scale pretraining data from 10K to 1M videos (ablation) | LCA · In-the-Wild test | PSNR: 27.665 -> 28.175 | Improved |
| scale pretraining data from 10K to 1M videos (ablation) | LCA · Capture-Studio test | PSNR: 30.393 -> 30.514 | No clear effect |
| use LCA instead of ExAvatar (multi-view) | 3D avatar reconstruction · In-the-Wild multi-view | PSNR: 18.010 -> 27.802 | Improved |
| use LCA instead of LHM (single-view) | 3D avatar reconstruction · In-the-Wild single-view | PSNR: 18.322 -> 27.685 | Improved |
| add loose garment support (two-level learnable deformation) in post-training (ablation) | LCA · loose garment avatars (qualitative) | visual animation quality | Improved |
Closest papers by meaning
- SplatCtrlA: Generalizable Single Image to Fully Controllable 3D Avatar (ECCV 2026 · similarity 0.90)
- Vid2Avatar-Pro: Authentic Avatar from Videos in the Wild via Universal Prior (CVPR 2025 · similarity 0.90)
- Codec Avatar Studio: Paired Human Captures for Complete, Driveable, and Generalizable Avatars (NeurIPS 2024 · similarity 0.89)
- Expressive Gaussian Human Avatars from Monocular RGB Video (NeurIPS 2024 · similarity 0.88)
- FRESA: Feedforward Reconstruction of Personalized Skinned Avatars from Few Images (CVPR 2025 · similarity 0.88)
- WildAvatar: Learning In-the-wild 3D Avatars from the Web (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.
