Pith. sign in

REVIEW 3 major objections 5 minor 63 references

Real-time High-fidelity Gaussian Human Avatars with Position-based Interpolation of Spatially Distributed MLPs

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a Gaussian human avatar representation based on pose-only spatially distributed MLPs and per-Gaussian offset bases reconstructs high-fidelity pose-dependent appearance and renders at 166 fps, about 16 times faster…

desk verdict Genuinely new basis-interpolation idea and an honest speed win, but the quality comparison is mostly in-sample; the single held-out result is much closer. read the letter →

arxiv 2504.12909 v2 pith:HB5EKJTA submitted 2025-04-17 cs.GR cs.CV

Youyi Zhan , Tianjia Shao , Yin Yang , Kun Zhou This is my paper · ORCID
classification cs.GRcs.CV
keywords 3DGaussianSplattinghumanavatarreconstructionpose-dependentappearancespatiallydistributedMLPsoffsetbasisinverse-distanceinterpolationcontrolpointsreal-timerendering
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper aims to establish that a 3D Gaussian human avatar can capture pose-dependent appearance details that currently require a heavy convolutional network, while still rendering in real time. The proposed representation sets about 300 small MLPs at fixed anchor points on a canonical template; each MLP takes only the pose as input and emits coefficient vectors, and each Gaussian obtains its property offsets by inverse-distance interpolation of the nearest three anchor outputs combined with its own learned offset basis. The paper argues that because the basis vectors are learned freely while coefficients vary smoothly, the representation can produce sharply different Gaussian properties in neighboring regions without paying a per-Gaussian network cost. It reports better quantitative quality than the compared methods on the tested multi-view sequences, with rendering at 166 fps versus 10 fps for AnimatableGaussians.

What carries the argument

The load-bearing object is the pair formed by the anchor coefficients $w^j_a = \mathcal{E}^j(\theta)$ output by $F=300$ spatially distributed MLPs located at anchor points, and the per-Gaussian offset bases $\{\delta\Lambda_k\}_{k=1}^{B}$, with $B=15$, covering rotation, scale, opacity, and SH coefficients. These are joined by inverse-distance interpolation: each pose passes through the MLPs once, and every Gaussian's coefficient vector is the distance-weighted combination $w_g = \frac{\sum_j \gamma(x_0,x^j_a) w^j_a}{\sum_j \gamma(x_0,x^j_a)}$ of its three nearest anchors, with $\gamma(x,y)=1/\|x-y\|^2$; the property offset is then $\delta\Lambda=\sum_k w_g[k]\,\delta\Lambda_k$ added to the neutral properties. Smoothly interpolated coefficients are deliberately combined with freely learned basis vectors, which is what lets the representation output sharply different Gaussian offsets in neighboring regions, and the control-point mechanism of Sec. 3.2 applies the same interpolation idea to position offsets to keep Gaussians on a surface layer.

What would settle it

Render a sequence with the canonical template deliberately perturbed (for example, shift or rotate the mesh by a few centimeters before anchor points, control points, and neutral Gaussian positions are sampled), then compare PSNR and LPIPS with the unperturbed run; a sharp drop would show the fixed-template assumption is load-bearing.

Watch

Extended reading notes

Core claim

The central claim is that pose-dependent Gaussian appearance decomposes into a neutral mean, per-Gaussian learned offset bases $\{\delta\Lambda_k\}$, and pose-driven coefficients that are spatially interpolated from a small set of anchor MLPs, and that this decomposition recovers detail a single MLP misses while avoiding the per-frame cost of a large CNN. Each Gaussian's coefficient vector is $w_g = \frac{\sum_j \gamma(x_0,x^j_a)\,w^j_a}{\sum_j \gamma(x_0,x^j_a)}$ over the three nearest anchors, with $\gamma(x,y)=1/\|x-y\|^2$, and its property offset is $\delta\Lambda = \sum_{k=1}^{B} w_g[k]\,\delta\Lambda_k$ added to the neutral properties. The same coefficient mechanism, routed through control points, also produces Gaussian position offsets and keeps Gaussians on a surface layer during novel-pose animation. On the sequences tested, the paper reports better appearance metrics than the compared methods and a rendering speed of 166 fps versus 10 fps for AnimatableGaussians.

Load-bearing premise

The load-bearing premise is that the canonical template mesh and skinning weights taken from AnimatableGaussians are accurate enough that the anchor points, control points, and neutral Gaussian positions sampled once on that mesh stay valid across all training and novel poses.

Editorial extensions

If this is right

  • High-fidelity pose-dependent avatars no longer need a heavy per-frame CNN: a single pose-vector pass through 300 small MLPs produces all Gaussian properties, which is why the reported speed reaches 166 fps.
  • The number of MLP evaluations per frame becomes independent of the number of Gaussians, so raising the Gaussian count from 200K upward carries a much smaller speed penalty than in network-per-Gaussian designs.
  • The proposed basis combination permits high-frequency detail without per-Gaussian network inference, a key difference from single-MLP designs that the ablation study supports.
  • Tying position offsets to control points makes novel-pose animation more robust, since Gaussians move with the surface rather than drifting freely inside the body.
  • The use of PCA projection at test time anchors the method to the training pose distribution, so novel poses are interpreted within that space, which the paper reports improves generalization.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An implicit consequence is that each Gaussian's offset basis defines a small linear subspace of appearance change; one could share basis sets across identities and learn only coefficients, which the paper does not explore.
  • The same anchor-and-control-point interpolation machinery should transfer to any articulated object with a template mesh and skinning weights, since nothing in the representation is human-specific beyond the pose vector.
  • Because each MLP is evaluated once per pose and the coefficients are then interpolated, the per-frame cost scales with the number of anchors, not the number of Gaussians; a compiled or quantized version of the MLP ensemble could push rendering further toward interactive VR budgets, though the paper does not test that.
  • A direct extension would vary the interpolation kernel (for example, learned weights or Gaussian falloff) and measure whether the reported detail gain persists; the paper fixes the inverse-distance kernel.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a real-time Gaussian human avatar representation in which a set of spatially distributed MLPs, located at anchor points on a canonical template mesh, take only the pose vector as input and output coefficients. Each Gaussian’s property offsets are obtained by combining these interpolated coefficients with per-Gaussian offset basis vectors, and the Gaussian position offsets are generated by interpolation from control points so that Gaussians remain on a surface layer. The method is evaluated on AvatarReX, THuman4.0, and ActorsHQ sequences against 3DGS-Avatar, MeshAvatar, AnimatableGaussians, and DEGAS, reporting better appearance quality and a large rendering speedup (166 fps versus 10 fps).

Significance. If the quantitative claims hold, the paper makes a meaningful contribution to real-time high-fidelity human avatar rendering: the speed-quality trade-off relative to AnimatableGaussians is notable, the idea of interpolating coefficients rather than high-frequency properties is well motivated and supported by ablations, and the control-point surface constraint is a practical solution to a real artifact. The authors ship code, provide a supplementary viewer, and include ablations on PCA components and Gaussian counts, which are strengths. However, the central quality claim is currently supported mainly by metrics computed on training frames, and the single held-out evaluation is narrow, so the headline result is not yet established to the standard expected for a high-impact venue.

major comments (3)
  1. [Sec. 4.1, Table 1; Supplementary Sec. 2] The quantitative comparison in Table 1 and the ablations in Tables 4-6 are evaluated on the first 500 frames of the first 2000 training frames, as stated in Supplementary Section 2. This is train-set leakage: PSNR/SSIM/LPIPS/FID computed on training images can be inflated by overfitting, and FID on training frames is not a valid distribution distance. The only held-out evidence is Table 2, where the margin over AnimatableGaussians shrinks to +0.216 dB PSNR and -0.0015 LPIPS. Please rerun the main comparison and the ablations on held-out frames (e.g., the last 500 frames) and report both fitting quality and held-out quality separately; the central claim of better appearance quality requires held-out results.
  2. [Sec. 4.1, Table 2] The novel-pose evaluation uses a single sequence (subject00) and a single camera view ("cam18") with no error bars or multiple random seeds. Given that the margin over AnimatableGaussians on this held-out test is small, the current evidence is too narrow to support the general claim of "better appearance quality with finer details" under novel poses. Please add multiple sequences and multiple evaluation views, and report per-sequence results with variance.
  3. [Sec. 3.3, Testing] The PCA projection of novel poses into the space of training poses is a test-time operation that moves test inputs toward the training distribution. The paper should state explicitly that the reported novel-pose results are obtained after this projection and discuss the implications for the generalization claim. It should also confirm that all comparison baselines are given exactly the same PCA treatment; if they are not, the comparison is not apples-to-apples, and if they are, the term "novel pose" should be qualified.
minor comments (5)
  1. [Sec. 1] The first paragraph of the introduction contains a duplicated word: "learning the the human appearance" should read "learning the human appearance."
  2. [Sec. 3.2] The phrase "position-ware MLP" should be "position-aware MLP."
  3. [Sec. 4; Supplementary Sec. 2] The main paper states that each sequence contains 1000 to 2000 frames, but Supplementary Section 2 says the avatarrex_zzr sequence uses the first 2000 frames for training. If the actual frame counts are shorter, this is inconsistent; please clarify the exact training and evaluation splits for every sequence.
  4. [Sec. 3.2] The relationship between the appearance coefficients and the position-offset coefficients is not defined in the main text; the supplementary clarifies that the MLP outputs 2B coefficients, half used for Gaussians and half for control points. This detail should be moved to the main text for clarity.
  5. [Sec. 3.3; Supplementary Code 1] The pose vector dimensionality is described as excluding finger joints, but Supplementary Code 1 uses a 63-dimensional input. Please state the exact dimensionality of the pose vector used in the main text.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the representation is an end-to-end fitted model, and the self-citations are not load-bearing.

full rationale

The paper's derivation chain is not circular. Section 3.1 defines a parametric Gaussian avatar: neutral properties, per-Gaussian offset basis, MLPs mapping pose to anchor coefficients, Eq. (2) inverse-distance interpolation to Gaussian coefficients, and Eq. (3) linear combination to produce property offsets. These equations define the model rather than deriving a prediction from its own output; the basis and MLP weights are optimized against multi-view images, so there is no self-definitional reduction. Section 3.2's control-point interpolation is likewise a constraint or parameterization, not a disguised fit of the evaluation target. The self-citations [32] and [54] appear only in related-work discussion and an ablation baseline, respectively, and neither is used to justify the central claim. The canonical template mesh and skinning weights are inherited from the external AnimatableGaussians [28] as inputs, not conclusions. The test-time PCA projection (Sec. 3.3) does move novel poses toward the training pose space, which is a legitimate generalization caveat, but it does not make the rendered image equal to a training image by construction. One experimental-validity concern should be noted: Supplementary Sec. 2 states metrics in Tables 1, 4, 5, and 6 are computed on the first 500 of the first 2000 training frames, so those numbers measure training fit rather than prediction; however, Table 2 is a genuinely held-out novel-pose evaluation, and the speed comparison in Table 3 is independent of fitting. These caveats affect how much confidence the headline quality claim deserves, not whether the derivation is circular.

Assumptions & free parameters 9 free parameters · 6 assumptions · 0 invented entities

No new physical entities are introduced; the Gaussian offset basis and control points are learned representations and geometric sampling tools. The method rests on an inherited template/registration pipeline and a set of hand-set hyperparameters, most of which are validated only through ablations on the same benchmark sequences.

free parameters (9)
  • Number of Gaussians N = 200,000
    Chosen by ablation in Supplementary Sec. 3; more Gaussians improve PSNR slightly but reduce FPS.
  • Number of anchor MLPs F = 300
    Chosen by ablation in Table 5 to balance quality and speed (50, 300, 800 tested).
  • Basis number B = 15
    Chosen by ablation in Table 6 (5, 15, 40 tested); 15 and 40 give comparable results.
  • Number of control points C = 10,000
    Set in Sec. 3.3 without an ablation, hand-selected.
  • PCA components = 20
    Chosen in Supplementary Sec. 4; fewer components lose details, no PCA produces artifacts.
  • Loss weights for LPIPS and control-point smoothness = 0.1 and 0.1 in L = L1 + 0.1*LPIPS + 0.1*Lctrl + Lscale
    Hand-set weighting in Eq. (8), not ablated.
  • Learning rates = 5e-4 for neutral properties and MLPs, 1.6e-4 for position offsets, basis five times smaller
    Hand-set in Sec. 3.3 Training; not ablated.
  • MLP architecture = 4 hidden layers: 512, 256, 256, 256
    Hand-set in Supplementary Sec. 1; no architecture search reported.
  • Training iterations = 800,000
    Fixed in Sec. 3.3, not ablated.
assumptions (6)
  • domain assumption The SMPL-X registrations and canonical template mesh provided by AnimatableGaussians [28] are accurate enough for all datasets and poses.
    Sec. 3 states the method follows [15,28,59] for masks, SMPL-X registration, and template mesh; the neutral Gaussian positions, anchors, and control points are fixed on this mesh (Sec. 3.3).
  • domain assumption Distance-weighted interpolation of coefficients from the three nearest anchor MLPs yields correct spatial variation of appearance.
    Eq. (2) defines the interpolation; no theoretical guarantee is given, only ablations.
  • domain assumption A linear combination of B=15 per-Gaussian offset basis vectors can represent the pose-dependent property changes, including high-frequency details.
    Eq. (3) and Sec. 3.1; ablation shows 15 basis is comparable to 40, but the sufficiency claim is empirical.
  • domain assumption Projecting novel poses into the training pose space with PCA does not remove appearance-relevant pose information.
    Sec. 3.3 Testing; without PCA, artifacts appear (Supp. Fig. 2), so the claim depends on this projection being valid.
  • domain assumption Constraining neighboring control points to have similar position offsets keeps Gaussians on a surface layer and prevents interior artifacts.
    Sec. 3.2 and Eq. (6); validated only through ablation (Fig. 7).
  • domain assumption The pose vector without finger joints is sufficient for body appearance modeling.
    Sec. 3.3 states this belief; no experiment with finger joints is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Real-time High-fidelity Gaussian Human Avatars with Position-based Interpolation of Spatially Distributed MLPs." pith.science (2026). https://pith.science/paper/HB5EKJTA

@misc{pith2026250412909,
  author       = {Pith},
  title        = {Pith review of: Real-time High-fidelity Gaussian Human Avatars with Position-based Interpolation of Spatially Distributed MLPs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HB5EKJTA}},
  note         = {Machine review of arXiv:2504.12909}
}
read the original abstract

Many works have succeeded in reconstructing Gaussian human avatars from multi-view videos. However, they either struggle to capture pose-dependent appearance details with a single MLP, or rely on a computationally intensive neural network to reconstruct high-fidelity appearance but with rendering performance degraded to non-real-time. We propose a novel Gaussian human avatar representation that can reconstruct high-fidelity pose-dependence appearance with details and meanwhile can be rendered in real time. Our Gaussian avatar is empowered by spatially distributed MLPs which are explicitly located on different positions on human body. The parameters stored in each Gaussian are obtained by interpolating from the outputs of its nearby MLPs based on their distances. To avoid undesired smooth Gaussian property changing during interpolation, for each Gaussian we define a set of Gaussian offset basis, and a linear combination of basis represents the Gaussian property offsets relative to the neutral properties. Then we propose to let the MLPs output a set of coefficients corresponding to the basis. In this way, although Gaussian coefficients are derived from interpolation and change smoothly, the Gaussian offset basis is learned freely without constraints. The smoothly varying coefficients combined with freely learned basis can still produce distinctly different Gaussian property offsets, allowing the ability to learn high-frequency spatial signals. We further use control points to constrain the Gaussians distributed on a surface layer rather than allowing them to be irregularly distributed inside the body, to help the human avatar generalize better when animated under novel poses. Compared to the state-of-the-art method, our method achieves better appearance quality with finer details while the rendering speed is significantly faster under novel views and novel poses.

Figures

Figures reproduced from arXiv: 2504.12909 by the authors.

Figure 1
Figure 1. Our method can model high-fidelity human avatars that can be animated under novel poses and rendered in real-time. Compared to [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Pipeline overview. (a) We define the spatially distributed MLPs on anchor points, which are uniformly sampled on the template [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the control point. The Gaussian position [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison with the state-of-the-art methods [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparison of several design choices. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Ablation study on Gaussian offset basis. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 8
Figure 8. Figure 8: Results of different view number. dering speed, as shown in Tab. 4 “w/o basis (2500 MLPs)”. We also shows that other possible designs are not as good as our basis combination. For example, following Ye et al. [54], we assign a learnable vector to each Gaussian. Each Ga…
Figure 9
Figure 9. Figure 9: Our method achieves high-quality human avatar reconstruction and animation under novel poses. [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 1
Figure 1. Figure 1: Qualitative comparison of different number of Guas [PITH_FULL_IMAGE:figures/full_fig_p012_1.png]
Figure 2
Figure 2. Figure 2: Ablation study on PCA components. 2 [PITH_FULL_IMAGE:figures/full_fig_p013_2.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 29 canonical work pages

  1. [1]

    Driving-signal aware full-body avatars

    Timur Bagautdinov, Chenglei Wu, Tomas Simon, Fabi ´an Prada, Takaaki Shiratori, Shih-En Wei, Weipeng Xu, Yaser Sheikh, and Jason Saragih. Driving-signal aware full-body avatars. ACM Transactions on Graphics (TOG), 40(4):1–17,

    work page

  2. [2]

    Detailed full-body reconstructions of moving peo- ple from monocular rgb-d sequences

    Federica Bogo, Michael J Black, Matthew Loper, and Javier Romero. Detailed full-body reconstructions of moving peo- ple from monocular rgb-d sequences. In Proceedings of the IEEE international conference on computer vision , pages 2300–2308, 2015. 2

    work page 2015

  3. [3]

    Uv volumes for real-time rendering of editable free-view human performance

    Yue Chen, Xuan Wang, Xingyu Chen, Qi Zhang, Xiaoyu Li, Yu Guo, Jue Wang, and Fei Wang. Uv volumes for real-time rendering of editable free-view human performance. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16621–16631, 2023. 2

    work page 2023

  4. [4]

    Meshavatar: Learning high-quality triangu- lar human avatars from multi-view videos

    Yushuo Chen, Zerong Zheng, Zhe Li, Chao Xu, and Yebin Liu. Meshavatar: Learning high-quality triangu- lar human avatars from multi-view videos. arXiv preprint arXiv:2407.08414, 2024. 3, 6

    arXiv 2024

  5. [5]

    High-quality streamable free-viewpoint video

    Alvaro Collet, Ming Chuang, Pat Sweeney, Don Gillett, Den- nis Evseev, David Calabrese, Hugues Hoppe, Adam Kirk, and Steve Sullivan. High-quality streamable free-viewpoint video. ACM Transactions on Graphics (ToG) , 34(4):1–13,

  6. [6]

    Ram-avatar: Real-time photo-realistic avatar from monoc- ular videos with full-body control

    Xiang Deng, Zerong Zheng, Yuxiang Zhang, Jingxiang Sun, Chao Xu, Xiaodong Yang, Lizhen Wang, and Yebin Liu. Ram-avatar: Real-time photo-realistic avatar from monoc- ular videos with full-body control. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1996–2007, 2024. 2

    work page 1996

  7. [7]

    Reconstructing personalized se- mantic facial nerf models from monocular video

    Xuan Gao, Chenglai Zhong, Jun Xiang, Yang Hong, Yudong Guo, and Juyong Zhang. Reconstructing personalized se- mantic facial nerf models from monocular video. ACM Transactions on Graphics (TOG), 41(6):1–12, 2022. 3

    2022

  8. [8]

    Learning neural volumetric representations of dy- namic humans in minutes

    Chen Geng, Sida Peng, Zhen Xu, Hujun Bao, and Xiaowei Zhou. Learning neural volumetric representations of dy- namic humans in minutes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 8759–8770, 2023. 2

    2023

Show all 63 references
  1. [9]

    Vid2avatar: 3d avatar reconstruction from videos in the wild via self-supervised scene decomposition

    Chen Guo, Tianjian Jiang, Xu Chen, Jie Song, and Otmar Hilliges. Vid2avatar: 3d avatar reconstruction from videos in the wild via self-supervised scene decomposition. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12858–12868, 2023. 2

    2023

  2. [10]

    Livecap: Real-time human performance capture from monocular video

    Marc Habermann, Weipeng Xu, Michael Zollhoefer, Ger- ard Pons-Moll, and Christian Theobalt. Livecap: Real-time human performance capture from monocular video. ACM Transactions On Graphics (TOG), 38(2):1–17, 2019. 2

    2019

  3. [11]

    Real-time deep dynamic characters

    Marc Habermann, Lingjie Liu, Weipeng Xu, Michael Zoll- hoefer, Gerard Pons-Moll, and Christian Theobalt. Real-time deep dynamic characters. ACM Transactions on Graphics (ToG), 40(4):1–16, 2021. 2

    2021

  4. [12]

    Gans trained by a two time-scale update rule converge to a local nash equilib- rium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilib- rium. Advances in neural information processing systems , 30, 2017. 5

    2017

  5. [13]

    Expressive gaussian human avatars from monocular rgb video

    Hezhen Hu, Zhiwen Fan, Tianhao Wu, Yihan Xi, Seoyoung Lee, Georgios Pavlakos, and Zhangyang Wang. Expressive gaussian human avatars from monocular rgb video. arXiv preprint arXiv:2407.03204, 2024. 3

    2024 arXiv

  6. [14]

    Gauhuman: Articu- lated gaussian splatting from monocular human videos

    Shoukang Hu, Tao Hu, and Ziwei Liu. Gauhuman: Articu- lated gaussian splatting from monocular human videos. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 20418–20431, 2024. 3

    2024

  7. [15]

    Humanrf: High-fidelity neural radiance fields for humans in motion

    Mustafa Is ¸ık, Martin R ¨unz, Markos Georgopoulos, Taras Khakhulin, Jonathan Starck, Lourdes Agapito, and Matthias Nießner. Humanrf: High-fidelity neural radiance fields for humans in motion. ACM Transactions on Graphics (TOG), 42(4):1–12, 2023. 3, 5

    2023

  8. [16]

    Splatarmor: Articulated gaussian splatting for animat- able humans from monocular rgb videos

    Rohit Jena, Ganesh Subramanian Iyer, Siddharth Choud- hary, Brandon Smith, Pratik Chaudhari, and James Gee. Splatarmor: Articulated gaussian splatting for animat- able humans from monocular rgb videos. arXiv preprint arXiv:2311.10812, 2023. 3

    2023 arXiv

  9. [17]

    In- stantavatar: Learning avatars from monocular video in 60 seconds

    Tianjian Jiang, Xu Chen, Jie Song, and Otmar Hilliges. In- stantavatar: Learning avatars from monocular video in 60 seconds. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 16922– 16932, 2023. 2

    2023

  10. [18]

    Neuman: Neural human radiance field from a single video

    Wei Jiang, Kwang Moo Yi, Golnoosh Samei, Oncel Tuzel, and Anurag Ranjan. Neuman: Neural human radiance field from a single video. In European Conference on Computer Vision, pages 402–418. Springer, 2022. 2

    2022

  11. [19]

    Uv gaussians: Joint learning of mesh deformation and gaus- sian textures for human avatar modeling

    Yujiao Jiang, Qingmin Liao, Xiaoyu Li, Li Ma, Qi Zhang, Chaopeng Zhang, Zongqing Lu, and Ying Shan. Uv gaussians: Joint learning of mesh deformation and gaus- sian textures for human avatar modeling. arXiv preprint arXiv:2403.11589, 2024. 3

    2024 arXiv

  12. [20]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42(4):139–1,

  13. [21]

    Hugs: Human gaussian splats

    Muhammed Kocabas, Jen-Hao Rick Chang, James Gabriel, Oncel Tuzel, and Anurag Ranjan. Hugs: Human gaussian splats. In Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition , pages 505–515, 2024. 3

    2024

  14. [22]

    Deliffas: Deformable light fields for fast avatar synthesis

    Youngjoong Kwon, Lingjie Liu, Henry Fuchs, Marc Haber- mann, and Christian Theobalt. Deliffas: Deformable light fields for fast avatar synthesis. Advances in Neural Informa- tion Processing Systems, 36, 2024. 2 9

    2024

  15. [23]

    Gen- eralizable human gaussians for sparse view synthesis

    Youngjoong Kwon, Baole Fang, Yixing Lu, Haoye Dong, Cheng Zhang, Francisco Vicente Carrasco, Albert Mosella- Montoro, Jianjin Xu, Shingo Takagi, Daeil Kim, et al. Gen- eralizable human gaussians for sparse view synthesis. In European Conference on Computer Vision, pages 451–46...

    2025

  16. [24]

    Gart: Gaussian articulated template mod- els

    Jiahui Lei, Yufu Wang, Georgios Pavlakos, Lingjie Liu, and Kostas Daniilidis. Gart: Gaussian articulated template mod- els. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition , pages 19876–19887,

  17. [25]

    Tava: Template-free animatable volumetric actors

    Ruilong Li, Julian Tanke, Minh V o, Michael Zollh ¨ofer, J¨urgen Gall, Angjoo Kanazawa, and Christoph Lassner. Tava: Template-free animatable volumetric actors. In Eu- ropean Conference on Computer Vision , pages 419–436. Springer, 2022. 2

    2022

  18. [26]

    Learning a model of facial shape and expression from 4d scans

    Tianye Li, Timo Bolkart, Michael J Black, Hao Li, and Javier Romero. Learning a model of facial shape and expression from 4d scans. ACM Trans. Graph., 36(6):194–1, 2017. 3

    2017

  19. [27]

    Posevocab: Learning joint-structured pose embeddings for human avatar modeling

    Zhe Li, Zerong Zheng, Yuxiao Liu, Boyao Zhou, and Yebin Liu. Posevocab: Learning joint-structured pose embeddings for human avatar modeling. In ACM SIGGRAPH 2023 Con- ference Proceedings, pages 1–11, 2023. 2, 6

    2023

  20. [28]

    Ani- matable gaussians: Learning pose-dependent gaussian maps for high-fidelity human avatar modeling

    Zhe Li, Zerong Zheng, Lizhen Wang, and Yebin Liu. Ani- matable 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, pages 19711–19722, 2024. 1, 2, 3, 4, 5, 6

    2024

  21. [29]

    Neural actor: Neural free-view synthesis of human actors with pose con- trol

    Lingjie Liu, Marc Habermann, Viktor Rudnev, Kripasindhu Sarkar, Jiatao Gu, and Christian Theobalt. Neural actor: Neural free-view synthesis of human actors with pose con- trol. ACM transactions on graphics (TOG) , 40(6):1–16,

  22. [30]

    Texvo- cab: Texture vocabulary-conditioned human avatars

    Yuxiao Liu, Zhe Li, Yebin Liu, and Haoqian Wang. Texvo- cab: Texture vocabulary-conditioned human avatars. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1715–1725, 2024. 2

    2024

  23. [31]

    Pixel codec avatars

    Shugao Ma, Tomas Simon, Jason Saragih, Dawei Wang, Yuecheng Li, Fernando De La Torre, and Yaser Sheikh. Pixel codec avatars. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 64–73,

  24. [32]

    3d gaussian blendshapes for head avatar animation

    Shengjie Ma, Yanlin Weng, Tianjia Shao, and Kun Zhou. 3d gaussian blendshapes for head avatar animation. In ACM SIGGRAPH 2024 Conference Papers, pages 1–10, 2024. 3

    2024

  25. [33]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. In European Conference on Computer Vision, pages 405–421. Springer, 2020. 1, 2

    2020

  26. [34]

    Expressive whole-body 3d gaussian avatar

    Gyeongsik Moon, Takaaki Shiratori, and Shunsuke Saito. Expressive whole-body 3d gaussian avatar. arXiv preprint arXiv:2407.21686, 2024. 3

    2024 arXiv

  27. [35]

    Human gaussian splatting: Real-time rendering of animatable avatars

    Arthur Moreau, Jifei Song, Helisa Dhamo, Richard Shaw, Yiren Zhou, and Eduardo P ´erez-Pellitero. Human gaussian splatting: Real-time rendering of animatable avatars. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 788–798, 2024. 1, 3

    2024

  28. [36]

    Instant neural graphics primitives with a multires- olution hash encoding

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a multires- olution hash encoding. ACM Trans. Graph. , 41(4):102:1– 102:15, 2022. 2

    2022

  29. [37]

    Ash: Animatable gaussian splats for efficient and photoreal human rendering

    Haokai Pang, Heming Zhu, Adam Kortylewski, Christian Theobalt, and Marc Habermann. Ash: Animatable gaussian splats for efficient and photoreal human rendering. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1165–1175, 2024. 3

    2024

  30. [38]

    ihuman: Instant animatable digital humans from monocular videos

    Pramish Paudel, Anubhav Khanal, Ajad Chhatkuli, Danda Pani Paudel, and Jyoti Tandukar. ihuman: Instant animatable digital humans from monocular videos. arXiv preprint arXiv:2407.11174, 2024. 3

    2024 arXiv

  31. [39]

    Expressive body capture: 3d hands, face, and body from a single image

    Georgios Pavlakos, Vasileios Choutas, Nima Ghorbani, Timo Bolkart, Ahmed AA Osman, Dimitrios Tzionas, and Michael J Black. Expressive body capture: 3d hands, face, and body from a single image. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognitio...

    2019

  32. [40]

    Ani- matable neural radiance fields for modeling dynamic human bodies

    Sida Peng, Junting Dong, Qianqian Wang, Shangzhan Zhang, Qing Shuai, Xiaowei Zhou, and Hujun Bao. Ani- matable neural radiance fields for modeling dynamic human bodies. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 14314–14323, 2021. 2

    2021

  33. [41]

    Neural body: Implicit neural representations with structured latent codes for novel view synthesis of dynamic humans

    Sida Peng, Yuanqing Zhang, Yinghao Xu, Qianqian Wang, Qing Shuai, Hujun Bao, and Xiaowei Zhou. Neural body: Implicit neural representations with structured latent codes for novel view synthesis of dynamic humans. In Proceed- ings of the IEEE/CVF Conference on Computer Vision a...

    2021

  34. [42]

    3dgs-avatar: Animatable avatars via deformable 3d gaussian splatting

    Zhiyin Qian, Shaofei Wang, Marko Mihajlovic, Andreas Geiger, and Siyu Tang. 3dgs-avatar: Animatable avatars via deformable 3d gaussian splatting. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5020–5030, 2024. 1, 3, 6

    2024

  35. [43]

    Degas: Detailed expressions on full- body gaussian avatars

    Zhijing Shao, Duotun Wang, Qing-Yao Tian, Yao-Dong Yang, Hengyu Meng, Zeyu Cai, Bo Dong, Yu Zhang, Kang Zhang, and Zeyu Wang. Degas: Detailed expressions on full- body gaussian avatars. arXiv preprint arXiv:2408.10588 ,

    arXiv

  36. [44]

    Splattingavatar: Realistic real-time human avatars with mesh-embedded gaussian splatting

    Zhijing Shao, Zhaolong Wang, Zhuang Li, Duotun Wang, Xiangru Lin, Yu Zhang, Mingming Fan, and Zeyu Wang. Splattingavatar: Realistic real-time human avatars with mesh-embedded gaussian splatting. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogniti...

    2024

  37. [45]

    Haha: Highly articulated gaussian human avatars with textured mesh prior

    David Svitov, Pietro Morerio, Lourdes Agapito, and Alessio Del Bue. Haha: Highly articulated gaussian human avatars with textured mesh prior. arXiv preprint arXiv:2404.01053,

    arXiv

  38. [46]

    Scanning 3d full human bodies using kinects

    Jing Tong, Jin Zhou, Ligang Liu, Zhigeng Pan, and Hao Yan. Scanning 3d full human bodies using kinects. IEEE trans- actions on visualization and computer graphics, 18(4):643– 650, 2012. 2

    2012

  39. [47]

    Stylea- vatar: Real-time photo-realistic portrait avatar from a single 10 video

    Lizhen Wang, Xiaochen Zhao, Jingxiang Sun, Yuxiang Zhang, Hongwen Zhang, Tao Yu, and Yebin Liu. Stylea- vatar: Real-time photo-realistic portrait avatar from a single 10 video. In ACM SIGGRAPH 2023 Conference Proceedings , pages 1–10, 2023. 2

    2023

  40. [48]

    Arah: Animatable volume rendering of articulated hu- man sdfs

    Shaofei Wang, Katja Schwarz, Andreas Geiger, and Siyu Tang. Arah: Animatable volume rendering of articulated hu- man sdfs. In European conference on computer vision, pages 1–19. Springer, 2022. 2

    2022

  41. [49]

    Gomavatar: Efficient an- imatable human modeling from monocular video using gaussians-on-mesh

    Jing Wen, Xiaoming Zhao, Zhongzheng Ren, Alexander G Schwing, and Shenlong Wang. Gomavatar: Efficient an- imatable human modeling from monocular video using gaussians-on-mesh. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 2059–20...

    2024

  42. [50]

    Hu- mannerf: Free-viewpoint rendering of moving people from monocular video

    Chung-Yi Weng, Brian Curless, Pratul P Srinivasan, Jonathan T Barron, and Ira Kemelmacher-Shlizerman. Hu- mannerf: Free-viewpoint rendering of moving people from monocular video. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern Recognition , pages 162...

    2022

  43. [51]

    Modeling clothing as a separate layer for an animatable hu- man avatar

    Donglai Xiang, Fabian Prada, Timur Bagautdinov, Weipeng Xu, Yuan Dong, He Wen, Jessica Hodgins, and Chenglei Wu. Modeling clothing as a separate layer for an animatable hu- man avatar. ACM Transactions on Graphics (TOG) , 40(6): 1–15, 2021. 2

    2021

  44. [52]

    Dressing avatars: Deep photorealistic appearance for physically sim- ulated clothing

    Donglai Xiang, Timur Bagautdinov, Tuur Stuyck, Fabian Prada, Javier Romero, Weipeng Xu, Shunsuke Saito, Jing- fan Guo, Breannan Smith, Takaaki Shiratori, et al. Dressing avatars: Deep photorealistic appearance for physically sim- ulated clothing. ACM Transactions on Graphics (...

    2022

  45. [53]

    Driv- able avatar clothing: Faithful full-body telepresence with dy- namic clothing driven by sparse rgb-d input

    Donglai Xiang, Fabian Prada, Zhe Cao, Kaiwen Guo, Chen- glei Wu, Jessica Hodgins, and Timur Bagautdinov. Driv- able avatar clothing: Faithful full-body telepresence with dy- namic clothing driven by sparse rgb-d input. In SIGGRAPH Asia 2023 Conference Papers, pages 1–11, 2023. 2

    2023

  46. [54]

    Animatable 3d gaussians for high-fidelity synthesis of human motions

    Keyang Ye, Tianjia Shao, and Kun Zhou. Animatable 3d gaussians for high-fidelity synthesis of human motions. arXiv preprint arXiv:2311.13404, 2023. 1, 3, 7

    2023 arXiv

  47. [55]

    Monohuman: Animatable human neu- ral field from monocular video

    Zhengming Yu, Wei Cheng, Xian Liu, Wayne Wu, and Kwan-Yee Lin. Monohuman: Animatable human neu- ral field from monocular video. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16943–16953, 2023. 2

    2023

  48. [56]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 5

    2018

  49. [57]

    Gps- gaussian: Generalizable pixel-wise 3d gaussian splatting for real-time human novel view synthesis

    Shunyuan Zheng, Boyao Zhou, Ruizhi Shao, Boning Liu, Shengping Zhang, Liqiang Nie, and Yebin Liu. Gps- gaussian: Generalizable pixel-wise 3d gaussian splatting for real-time human novel view synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Re...

    2024

  50. [58]

    Structured local radiance fields for human avatar modeling

    Zerong Zheng, Han Huang, Tao Yu, Hongwen Zhang, Yan- dong Guo, and Yebin Liu. Structured local radiance fields for human avatar modeling. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 15893–15903, 2022. 2, 3, 5, 1

    2022

  51. [59]

    Avatarrex: Real-time expressive full- body avatars

    Zerong Zheng, Xiaochen Zhao, Hongwen Zhang, Boning Liu, and Yebin Liu. Avatarrex: Real-time expressive full- body avatars. ACM Transactions on Graphics (TOG), 42(4): 1–19, 2023. 2, 3, 5, 1 11 Real-time High-fidelity Gaussian Human Avatars with Position-based Interpolation of S...

    2023

  52. [60]

    More Implementation Details Template Mesh. For the canonical template mesh used in our method, we use SMPL-X [39] mesh as the template mesh for avatars wearing tight clothing, and follow Ani- matableGaussians [28] to obtain template mesh for avatars wearing loose clothing. Spa...

  53. [61]

    22010710

    Experiment Details In the main paper, the quantitative results in Tab. 1, Tab. 4, Tab. 5, and Tab. 6 are evaluated onavatarrex zzr sequence from AvatarRex [59] dataset. We use the first 2000 frames for training and calculate metrics on the first 500 frames from “22010710” came...

    2000

  54. [62]

    Quantitative experiments are shown in Tab

    Ablation Study on Gaussian Number We conduct experiments on the number of Gaussians used in our method. Quantitative experiments are shown in Tab. 1. Although reducing the number of Gaussians can greatly improve rendering speed, we find that fewer Gaus- sians make it more diff...

  55. [63]

    Ablation Study on PCA Components We use 20 PCA components during testing. Fig. 2 shows the results of using different numbers of components and no PCA. The left results show that using fewer PCA com- ponents yields fewer details. The right results show that artifacts can appea...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.