Animatable 3D Gaussians for modeling dynamic humans

Abstract

We present an animatable 3D Gaussian representation for synthesizing high-fidelity human videos under novel views and poses in real time. Given multi-view videos of a human subject, we learn a collection of 3D Gaussians in the canonical space of the rest pose. Each Gaussian is associated with a few basic properties (i.e., position, opacity, scale, rotation, spherical harmonics coefficients) representing the average human appearance across all video frames, as well as a latent code and a set of blend weights for dynamic appearance correction and pose transformation. The latent code is fed to an Multi-layer Perceptron (MLP) with a target pose to correct Gaussians in the canonical space to capture appearance changes under the target pose. The corrected Gaussians are then transformed to the target pose using linear blend skinning (LBS) with their blend weights. High-fidelity human images under novel views and poses can be rendered in real time through Gaussian splatting. Compared to state-of-the-art NeRF-based methods, our animatable Gaussian representation produces more compelling results with well captured details, and achieves superior rendering performance.

Graphical abstract

Keywords

free-view videos / image-based rendering / Gaussian splatting

Yukun Xu is a postgraduate at the State Key Lab of CAD&CG, Zhejiang University, China. His research interests include computer graphics and digital human

Keyang Ye received the bachelor’s degree from National University of Defense Technology, China in 2022. He is currently working toward the PhD degree in the Graphics and Parallel Systems Lab of Zhejiang University, China. His research interests include animation and rendering

Tianjia Shao is a professor in the State Key Laboratory of CAD&CG, Zhejiang University, China. Previously, he was a Lecturer in the School of Computing, University of Leeds, UK. He received his PhD in computer science from Institute for Advanced Study, Tsinghua University, and his BS from the Department of Automation, Tsinghua University, China. His research focuses on 3D modeling, digital human, and computer animation

Yanlin Weng received the bachelor’s and master’s degrees in control science and engineering from Zhejiang University, China and the PhD degree in computer science from the University of Wisconsin - Milwaukee, USA. She is currently an associate professor with the School of Computer Science and Technology, Zhejiang University. Her research interests include computer graphics and multimedia

References

[1] Loper M, Mahmood N, Romero J, Pons-Moll G, Black M J. SMPL: a skinned multi-person linear model. In: Whitton M C, ed. Seminal Graphics Papers: Pushing the Boundaries, Volume 2. New York: ACM, 2023, 851−866

[2] Allen B, Curless B, Popović Z . The space of human body shapes: reconstruction and parameterization from range scans. ACM Transactions on Graphics, 2003, 22( 3): 587–594

[3] Mildenhall B, Srinivasan P P, Tancik M, Barron J T, Ramamoorthi R, Ng R. NeRF: representing scenes as neural radiance fields for view synthesis. In: Proceedings of the 16th European Conference on Computer Vision. 2020, 405−421

[4] Peng S, Dong J, Wang Q, Zhang S, Shuai Q, Zhou X, Bao H. Animatable neural radiance fields for modeling dynamic human bodies. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021, 14294−14303

[5] Zhao F, Yang W, Zhang J, Lin P, Zhang Y, Yu J, Xu L. HumanNeRF: efficiently generated human radiance field from sparse inputs. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022, 7733−7743

[6] Chen Y, Wang X, Chen X, Zhang Q, Li X, Guo Y, Wang J, Wang F. UV volumes for real-time rendering of editable free-view human performance. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023, 16621−16631

[7] Peng S, Zhang Y, Xu Y, Wang Q, Shuai Q, Bao H, Zhou X. Neural body: implicit neural representations with structured latent codes for novel view synthesis of dynamic humans. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, 9050−9059

[8] Lin H, Peng S, Xu Z, Yan Y, Shuai Q, Bao H, Zhou X. Efficient neural radiance fields for interactive free-viewpoint video. In: Proceedings of the SIGGRAPH Asia 2022 Conference Papers. 2022, 39

[9] Kerbl B, Kopanas G, Leimkuehler T, Drettakis G . 3D Gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics, 2023, 42( 4): 139

[10] Yang Z, Gao X, Zhou W, Jiao S, Zhang Y, Jin X. Deformable 3D Gaussians for high-fidelity monocular dynamic scene reconstruction. 2023, arXiv preprint arXiv: 2309.13101

[11] Jacobson A, Deng Z, Kavan L, Lewis J P. Skinning: real-time shape deformation (full text not available). In: Proceedings of the ACM SIGGRAPH 2014 Courses. 2014, 24

[12] Joo H, Simon T, Sheikh Y. Total capture: a 3D deformation model for tracking faces, hands, and bodies. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018, 8320−8329

[13] Osman A A A, Bolkart T, Black M J. STAR: sparse trained articulated human body regressor. In: Proceedings of the 16th European Conference on Computer Vision. 2020, 598−613

[14] Zhang C, Pujades S, Black M J, Pons-Moll G. Detailed, accurate, human shape estimation from clothed 3D scan sequences. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017, 5484−5493

[15] Guan P, Reiss L, Hirshberg D A, Weiss A, Black M J . DRAPE: dressing any PErson. ACM Transactions on Graphics, 2012, 31( 4): 35

[16] Xu F, Liu Y, Stoll C, Tompkin J, Bharaj G, Dai Q, Seidel H P, Kautz J, Theobalt C. Video-based characters: creating new human performances from a multi-view video database. In: Proceedings of the ACM SIGGRAPH 2011 Papers. 2011, 32

[17] Habermann M, Liu L, Xu W, Zollhoefer M, Pons-Moll G, Theobalt C . Real-time deep dynamic characters. ACM Transactions on Graphics, 2021, 40( 4): 94

[18] Lombardi S, Simon T, Saragih J, Schwartz G, Lehrmann A, Sheikh Y . Neural volumes: learning dynamic renderable volumes from images. ACM Transactions on Graphics, 2019, 38( 4): 65

[19] Wu M, Wang Y, Hu Q, Yu J. Multi-view neural human rendering. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020, 1679−1688

[20] Bagautdinov T, Wu C, Simon T, Prada F, Shiratori T, Wei S E, Xu W, Sheikh Y, Saragih J . Driving-signal aware full-body avatars. ACM Transactions on Graphics, 2021, 40( 4): 143

[21] Ma S, Simon T, Saragih J, Wang D, Li Y, De La Torre F, Sheikh Y. Pixel codec avatars. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, 64−73

[22] Yang G, Vo M, Neverova N, Ramanan D, Vedaldi A, Joo H. BANMo: building animatable 3D neural models from many casual videos. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022, 2853−2863

[23] Xu Z, Peng S, Lin H, He G, Sun J, Shen Y, Bao H, Zhou X. 4K4D: real-time 4D view synthesis at 4K resolution. 2023, arXiv preprint arXiv: 2310.11448

[24] Xu Z, Peng S, Geng C, Mou L, Yan Z, Sun J, Bao H, Zhou X. Relightable and animatable neural avatar from sparse-view video. 2023, arXiv preprint arXiv: 2308.07903

[25] Peng B, Hu J, Zhou J, Gao X, Zhang J . IntrinsicNGP: intrinsic coordinate based hash encoding for human NeRF. IEEE Transactions on Visualization and Computer Graphics, 2024, 30( 8): 5679–5692

[26] Zheng Z, Huang H, Yu T, Zhang H, Guo Y, Liu Y. Structured local radiance fields for human avatar modeling. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022, 15872−15882

[27] Wang L, Zhang J, Liu X, Zhao F, Zhang Y, Zhang Y, Wu M, Yu J, Xu L. Fourier PlenOctrees for dynamic radiance field rendering in real-time. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022, 13514−13524

[28] Jiang T, Chen X, Song J, Hilliges O. InstantAvatar: learning avatars from monocular video in 60 seconds. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023, 16922−16932

[29] Müller T, Evans A, Schied C, Keller A . Instant neural graphics primitives with a multiresolution hash encoding. ACM Transactions on Graphics, 2022, 41( 4): 102

[30] Buehler C, Bosse M, McMillan L, Gortler S, Cohen M. Unstructured lumigraph rendering. In: Whitton M C, ed. Seminal Graphics Papers: Pushing the Boundaries, Volume 2. New York: ACM, 2023, 52

[31] Davis A, Levoy M, Durand F . Unstructured light fields. Computer Graphics Forum, 2012, 31( 2pt1): 305–314

[32] Eisemann M, De Decker B, Magnor M, Bekaert P, De Aguiar E, Ahmed N, Theobalt C, Sellent A . Floating textures. Computer Graphics Forum, 2008, 27( 2): 409–418

[33] Yu A, Li R, Tancik M, Li H, Ng R, Kanazawa A. PlenOctrees for real-time rendering of neural radiance fields. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021, 5732−5741

[34] Garbin S J, Kowalski M, Johnson M, Shotton J, Valentin J. FastNeRF: high-fidelity neural rendering at 200FPS. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021, 14326−14335

[35] Zielonka W, Bagautdinov T, Saito S, Zollhöfer M, Thies J, Romero J. Drivable 3D Gaussian avatars. 2023, arXiv preprint arXiv: 2311.08581

[36] Li Z, Zheng Z, Wang L, Liu Y. Animatable Gaussians: learning pose-dependent Gaussian maps for high-fidelity human avatar modeling. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2024, 19711−19722

[37] Wang L, Zhao X, Sun J, Zhang Y, Zhang H, Yu T, Liu Y. StyleAvatar: real-time photo-realistic portrait avatar from a single video. In: Proceedings of the ACM SIGGRAPH 2023 Conference Proceedings. 2023, 67

[38] Jena R, Iyer G S, Choudhary S, Smith B, Chaudhari P, Gee J. SplatArmor: articulated Gaussian splatting for animatable humans from monocular RGB videos. 2023, arXiv preprint arXiv: 2311.10812

[39] Moreau A, Song J, Dhamo H, Shaw R, Zhou Y, Pérez-Pellitero E. Human Gaussian splatting: real-time rendering of animatable avatars. 2023, arXiv preprint arXiv: 2311.17113

[40] Kocabas M, Chang J H R, Gabriel J, Tuzel O, Ranjan A. HUGS: human Gaussian splats. 2023, arXiv preprint arXiv: 2311.17910

[41] Hu S, Liu Z. GauHuman: articulated Gaussian splatting from monocular human videos. 2023, arXiv preprint arXiv: 2312.02973

[42] Lei J, Wang Y, Pavlakos G, Liu L, Daniilidis K. GART: Gaussian articulated template models. 2023, arXiv preprint arXiv: 2311.16099

[43] Hu L, Zhang H, Zhang Y, Zhou B, Liu B, Zhang S, Nie L. GaussianAvatar: towards realistic human avatar modeling from a single video via animatable 3D Gaussians. 2023, arXiv preprint arXiv: 2312.02134

[44] Xiang J, Gao X, Guo Y, Zhang J. FlashAvatar: high-fidelity digital avatar rendering at 300FPS. 2023, arXiv preprint arXiv: 2312.02214

[45] Lin S, Ryabtsev A, Sengupta S, Curless B, Seitz S, Kemelmacher-Shlizerman I. Real-time high-resolution background matting. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, 8758−8767

[46] Shoemake K, Duff T. Matrix animation and polar decomposition. In: Proceedings of the Conference on Graphics Interface. 1992, 258−264

[47] Zhang R, Isola P, Efros A A, Shechtman E, Wang O. The unreasonable effectiveness of deep features as a perceptual metric. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018, 586−595

[48] Ionescu C, Papava D, Olaru V, Sminchisescu C . Human3. 6M: large scale datasets and predictive methods for 3D human sensing in natural environments. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36( 7): 1325–1339

[49] Joo H, Simon T, Li X, Liu H, Tan L, Gui L, Banerjee S, Godisart T, Nabbe B, Matthews I, Kanade T, Nobuhara S, Sheikh Y . Panoptic studio: a massively multiview system for social interaction capture. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019, 41( 1): 190–204

[50] Liu Y, Li Z, Liu Y, Wang H. TexVocab: texture vocabulary-conditioned human avatars. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024, 1715−1725

Acknowledgements

This work was supported by the National Key R&D Program of China (No. 2022YFF0902302) and the National Natural Science Foundation of China (Grant Nos. 62172357 & 62322209).

Competing interests

The authors declare that they have no competing interests or financial conflicts to disclose.

RIGHTS & PERMISSIONS

2025 Higher Education Press