Large-scale Codec Avatars: The Unreasonable Effectiveness of Large-scale Avatar Pretraining
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866908943754723328 |
|---|---|
| author | Li, Junxuan Khirodkar, Rawal He, Chengan Jiang, Zhongshi Nam, Giljoo Yang, Lingchen Lee, Jihyun Zakharov, Egor Su, Zhaoen Abdrashitov, Rinat Dong, Yuan Martinez, Julieta Li, Kai Tan, Qingyang Shiratori, Takaaki Hu, Matthew Guo, Peihong Huang, Xuhua Zarei, Ariyan Pesavento, Marco Xu, Yichen Wen, He Deng, Teng Borsos, Wyatt Thakrar, Anjali Bazin, Jean-Charles Stoll, Carsten Hidalgo, Ginés Booth, James Wang, Lucy Ma, Xiaowen Rong, Yu Thalanki, Sairanjith Cao, Chen Häne, Christian Kar, Abhishek Bouaziz, Sofien Saragih, Jason Sheikh, Yaser Saito, Shunsuke |
| author_facet | Li, Junxuan Khirodkar, Rawal He, Chengan Jiang, Zhongshi Nam, Giljoo Yang, Lingchen Lee, Jihyun Zakharov, Egor Su, Zhaoen Abdrashitov, Rinat Dong, Yuan Martinez, Julieta Li, Kai Tan, Qingyang Shiratori, Takaaki Hu, Matthew Guo, Peihong Huang, Xuhua Zarei, Ariyan Pesavento, Marco Xu, Yichen Wen, He Deng, Teng Borsos, Wyatt Thakrar, Anjali Bazin, Jean-Charles Stoll, Carsten Hidalgo, Ginés Booth, James Wang, Lucy Ma, Xiaowen Rong, Yu Thalanki, Sairanjith Cao, Chen Häne, Christian Kar, Abhishek Bouaziz, Sofien Saragih, Jason Sheikh, Yaser Saito, Shunsuke |
| contents | High-quality 3D avatar modeling faces a critical trade-off between fidelity and generalization. On the one hand, multi-view studio data enables high-fidelity modeling of humans with precise control over expressions and poses, but it struggles to generalize to real-world data due to limited scale and the domain gap between the studio environment and the real world. On the other hand, recent large-scale avatar models trained on millions of in-the-wild samples show promise for generalization across a wide range of identities, yet the resulting avatars are often of low-quality due to inherent 3D ambiguities. To address this, we present Large-Scale Codec Avatars (LCA), a high-fidelity, full-body 3D avatar model that generalizes to world-scale populations in a feedforward manner, enabling efficient inference. Inspired by the success of large language models and vision foundation models, we present, for the first time, a pre/post-training paradigm for 3D avatar modeling at scale: we pretrain on 1M in-the-wild videos to learn broad priors over appearance and geometry, then post-train on high-quality curated data to enhance expressivity and fidelity. LCA generalizes across hair styles, clothing, and demographics while providing precise, fine-grained facial expressions and finger-level articulation control, with strong identity preservation. Notably, we observe emergent generalization to relightability and loose garment support to unconstrained inputs, and zero-shot robustness to stylized imagery, despite the absence of direct supervision. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_02320 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Large-scale Codec Avatars: The Unreasonable Effectiveness of Large-scale Avatar Pretraining Li, Junxuan Khirodkar, Rawal He, Chengan Jiang, Zhongshi Nam, Giljoo Yang, Lingchen Lee, Jihyun Zakharov, Egor Su, Zhaoen Abdrashitov, Rinat Dong, Yuan Martinez, Julieta Li, Kai Tan, Qingyang Shiratori, Takaaki Hu, Matthew Guo, Peihong Huang, Xuhua Zarei, Ariyan Pesavento, Marco Xu, Yichen Wen, He Deng, Teng Borsos, Wyatt Thakrar, Anjali Bazin, Jean-Charles Stoll, Carsten Hidalgo, Ginés Booth, James Wang, Lucy Ma, Xiaowen Rong, Yu Thalanki, Sairanjith Cao, Chen Häne, Christian Kar, Abhishek Bouaziz, Sofien Saragih, Jason Sheikh, Yaser Saito, Shunsuke Computer Vision and Pattern Recognition Graphics High-quality 3D avatar modeling faces a critical trade-off between fidelity and generalization. On the one hand, multi-view studio data enables high-fidelity modeling of humans with precise control over expressions and poses, but it struggles to generalize to real-world data due to limited scale and the domain gap between the studio environment and the real world. On the other hand, recent large-scale avatar models trained on millions of in-the-wild samples show promise for generalization across a wide range of identities, yet the resulting avatars are often of low-quality due to inherent 3D ambiguities. To address this, we present Large-Scale Codec Avatars (LCA), a high-fidelity, full-body 3D avatar model that generalizes to world-scale populations in a feedforward manner, enabling efficient inference. Inspired by the success of large language models and vision foundation models, we present, for the first time, a pre/post-training paradigm for 3D avatar modeling at scale: we pretrain on 1M in-the-wild videos to learn broad priors over appearance and geometry, then post-train on high-quality curated data to enhance expressivity and fidelity. LCA generalizes across hair styles, clothing, and demographics while providing precise, fine-grained facial expressions and finger-level articulation control, with strong identity preservation. Notably, we observe emergent generalization to relightability and loose garment support to unconstrained inputs, and zero-shot robustness to stylized imagery, despite the absence of direct supervision. |
| title | Large-scale Codec Avatars: The Unreasonable Effectiveness of Large-scale Avatar Pretraining |
| topic | Computer Vision and Pattern Recognition Graphics |
| url | https://arxiv.org/abs/2604.02320 |