Pith. sign in

REVIEW 4 major objections 7 minor 1 cited by

RMAvatar: Photorealistic Human Avatar Reconstruction from Monocular Video Based on Rectified Mesh-embedded Gaussians

T0 review · 4 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read RMAvatar reconstructs photorealistic animatable avatars from monocular video by binding Gaussian splats to a SMPL mesh and rectifying them with a pose-conditioned network, reaching state-of-the-art rendering and animation quality on…

desk verdict A sensible hybrid avatar method whose internal ablations support the rectification module, but whose state-of-the-art claim rests on uncontrolled pose inputs and a too-thin baseline set. read the letter →

arxiv 2501.07104 v1 pith:QVOMKUPI submitted 2025-01-13 cs.CV

classification cs.CV
keywords humanavatarreconstruction3DGaussiansplattingmesh-embeddedGaussiansmonocularvideonon-rigiddeformationSMPLnovelviewsynthesisanimatable
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

RMAvatar tries to reconstruct a photorealistic, animatable clothed human avatar from a single monocular video. The paper argues that a hybrid representation—3D Gaussian splats bound to the faces of a posed SMPL mesh—captures gross motion and shape, while a pose-conditioned rectification network adds the fine non-rigid detail (cloth wrinkles, hair, hand motion) that linear blend skinning misses. If the claims hold, the method would give higher novel-view synthesis and animation quality than prior NeRF- and Gaussian-based avatar methods on the PeopleSnapshot and ZJU-MoCap benchmarks, and do so with the speed of Gaussian rasterization. The paper reports state-of-the-art PSNR, SSIM, and LPIPS scores on both datasets, with ablation studies showing the rectification module is the main contributor.

What carries the argument

The central representation is the mesh-embedded Gaussian: each 3D Gaussian is attached to a parent triangle of the posed SMPL mesh, so its local position, rotation, and scale are transformed to world space by the triangle's frame, then refined by a Gaussian rectification module—a five-layer MLP mapping the encoded Gaussian position and the current pose to small offsets $(\delta\mu, \delta r, \delta s)$. The mesh guides low-frequency motion; the rectifier adds pose-dependent non-rigid detail; three regularization terms keep Gaussians near their parent triangles, prevent overly large scaling, and keep the rectifier's offsets small.

What would settle it

Render the same ZJU-MoCap subjects from held-out viewpoints using ground-truth SMPL fits from the multi-view capture instead of Anim-NeRF-refined poses; if RMAvatar's margin over SplattingAvatar shrinks, its gain is mostly correcting pose error rather than modeling authentic non-rigid deformation.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that binding Gaussian splats to mesh triangles and correcting them with a pose-dependent MLP yields the best reported novel-view and animation quality for monocular human avatar reconstruction. The mesh supplies a compact, topology-complete scaffold that places Gaussians correctly under LBS-driven motion, and the rectification module predicts per-Gaussian offsets in position, rotation, and scale as a function of pose and Gaussian location, compensating for the non-rigid deformations LBS cannot express. The paper states that this decomposition of motion into rigid mesh guidance plus learned non-rigid rectification surpasses both pure-Gaussian methods like GaussianAvatar and the mesh-embedded SplattingAvatar baseline on PSNR, SSIM, and LPIPS, and produces cleaner geometry for loose clothing and hair.

Load-bearing premise

The load-bearing premise is that the SMPL poses used for training and animation are accurate enough for the mesh to provide reliable guidance, and that the pose-conditioned rectifier can learn non-rigid deformation without overfitting to pose-specific errors or failing on out-of-distribution poses.

Editorial extensions

If this is right

  • A direct corollary is that the same hybrid scaffold can be dropped into real-time rendering pipelines, since rasterization uses 3D Gaussian splatting rather than volumetric ray marching.
  • If the pose-conditioned rectifier generalizes, avatars animated to out-of-distribution poses should stay coherent on loose clothing and hair, where pure LBS mesh warping fails.
  • The method's sensitivity to pose accuracy is reduced relative to mesh-only baselines because the rectifier can absorb small pose errors, which matters for monocular capture where fitted SMPL is noisy.
  • The ablation results imply that the rectification module, not densification, is the main source of the quality gain, so future avatar methods need not grow Gaussian counts to improve fidelity.

Reading between the lines

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

  • An unstated corollary is that the rectifier may be compensating for pose-estimation error as much as for true non-rigid deformation; if so, its benefit should shrink when trained with ground-truth poses, which is directly testable on multi-view datasets.
  • The same binding-plus-rectification pattern could transfer to other parametric templates such as FLAME for heads, where expression-dependent wrinkles require non-rigid correction beyond mesh motion.
  • The offset regularization caps how far Gaussians can leave their parent triangles, so very loose garments or large cloth folds that require big displacements may be under-modelled; testing on a video with a long skirt or wide sleeves would probe this boundary.
  • Because the rectifier is a single MLP over pose and position, its capacity is fixed; training on longer sequences with more pose variety should push generalization further, but the paper does not quantify this scaling behavior.
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

4 major / 7 minor

Summary. The paper proposes RMAvatar, a hybrid representation for animatable human avatars from monocular video. 3D Gaussian splats are embedded on the triangular faces of an SMPL mesh; the mesh drives their rigid motion via LBS, and a pose-conditioned 5-layer MLP predicts offsets to position, rotation, and scale to capture non-rigid deformations such as cloth and hair. Training uses RGB/SSIM/LPIPS losses plus position, scaling, and offset regularizers. Experiments on PeopleSnapshot and ZJU-MoCap report PSNR/SSIM/LPIPS improvements over Anim-NeRF, InstantAvatar, GaussianAvatar, and SplattingAvatar, and a qualitative animation comparison.

Significance. If the results are taken at face value, the representation is a sensible marriage of explicit mesh guidance (robust under novel poses) and Gaussian splatting appearance, with the rectification MLP addressing the known LBS limitation for non-rigid deformations. The method is described in sufficient detail to be re-implemented, and the ablation isolates the contribution of the rectification module. The paper does not ship code or machine-checked proofs; its main deliverable is the empirical claim. That claim is currently weakened by (i) an uncontrolled comparison protocol with respect to SMPL pose refinement, (ii) a ZJU-MoCap evaluation that tests novel poses rather than held-out views, and (iii) missing comparisons against closely related Gaussian-on-mesh methods. With a controlled evaluation, the method could be a solid contribution.

major comments (4)
  1. [§4.1, Tables 1–2] The comparison is not controlled for SMPL pose inputs. Section 4.1 states 'For fair comparison, we use the provided poses optimized by Anim-NeRF [4] in our training and inference' (PeopleSnapshot) and 'We use Anim-NeRF [40] to get the refined poses of ZJU-MoCap in our training and inference,' but it never states whether Anim-NeRF, InstantAvatar, GaussianAvatar, or SplattingAvatar were retrained with these same refined poses or used the same ZJU frame selection. Since the paper itself notes in §4.3 that SplattingAvatar's quality 'is highly dependent on the motion accuracy of the underlying mesh,' the 0.3–3.0 dB advantages reported in Tables 1–2 may reflect better pose inputs rather than the rectification module. Please re-run all baselines under identical pose refinement and data splits, or provide evidence that the published baseline numbers were produced under the same protocol, and include a pose-sensitivity study (e.g., RMAvatar with raw vs. refined SMPL).
  2. [Table 3, §4.1–4.3] The ZJU-MoCap evaluation does not demonstrate novel-view synthesis. Table 3 shows training on frames 1–456 and testing on frames 457–end for each subject, which is temporal extrapolation at the same camera viewpoint, not a held-out view. The claim in §4.3 of 'novel view synthesis' is therefore overstated. Either add a held-out camera evaluation (train on one camera, test on another) or explicitly re-label the experiment as novel-pose synthesis and adjust the conclusion accordingly.
  3. [§2.2–2.3, Tables 1–2] Three closely related Gaussian-avatar methods discussed in the related work—3DGS-Avatar [45], GoMAvatar [55], and GART [26]—are absent from the quantitative comparison. GoMAvatar in particular is a Gaussians-on-mesh method with the same design philosophy. Without these baselines (or a clear justification for their exclusion, such as incompatible pose estimation requirements), the claim of far exceeding the state of the art is not established. Please add these methods under the same protocol or narrow the claim to the compared set.
  4. [Table 4, §4.4] The ablation support for the OffsetLoss and ScalingLoss components is weak. On female-3-casual, the full model achieves PSNR 32.51 versus 32.48 without OffsetLoss and 32.45 without ScalingLoss, with identical SSIM and LPIPS in both cases; no variance or multiple seeds are reported. This does not support the statement in §4.4 that the offset and scaling losses 'significantly improve' reconstruction quality. Please report mean and standard deviation over multiple runs or over additional subjects, and use a test that accounts for run-to-run variance.
minor comments (7)
  1. [Eq. (12), §3.4] The sentence 'Note that L_rgb, L_ssim and L_lpips are L1-norm losses' is incorrect: SSIM and LPIPS losses are not L1 norms. Please replace it with proper definitions (e.g., 1−SSIM and the LPIPS distance).
  2. [§4.3] The claim 'Our method far exceeds the SoTA methods' is too strong for gains of 0.3–0.6 dB on several ZJU subjects (e.g., subject 386: 30.61 vs 30.31 PSNR). Substantiate with statistical analysis or temper the language.
  3. [Figure 4, §4.3] Please specify how out-of-distribution poses were generated (e.g., novel SMPL poses from other sequences or retargeting) and provide quantitative metrics for animation frames if possible.
  4. [§4.1] For ZJU-MoCap, specify which camera viewpoint was used per subject and confirm that the same camera and frame split were used for all compared methods.
  5. [§4.3] The sentence attributing SplattingAvatar's artifacts to 'Walking on mesh strategy' is speculative; either provide a citation or evidence, or soften it.
  6. [Eq. (13), §3.4] The position regularization L_pos = ||max(μ−ε_pos,0)||2 constrains only positive deviations from the parent triangle; if the intent is to constrain both directions, the absolute value should be used inside the max. Please clarify.
  7. [Throughout] There are numerous typos and spacing artifacts (e.g., 'threshhold' near Eq. (14), 'recitification' in §3.3, 'farexceeds' in §4.3). A careful proofread is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the method is a per-scene supervised reconstruction evaluated on held-out frames, with no prediction derived from its own inputs.

full rationale

RMAvatar's central derivation is a per-scene supervised learning pipeline, not a derivation that returns its own inputs. The Gaussian initialization (Eqs. 5-7) places splats on a posed SMPL mesh; the rectification MLP (Eq. 8) maps encoded Gaussian positions and pose to offsets; Eqs. 9-11 apply those offsets; and Eq. 16 combines RGB, SSIM, LPIPS, and regularization losses. Every learned component is optimized against training views and evaluated on held-out frames, as shown by the disjoint train/test frame indices in Table 3 for ZJU-MoCap and the InstantAvatar split for PeopleSnapshot. No predicted quantity is used to define the model or its loss, and no benchmark number is fitted as a parameter. The reported state-of-the-art claim is an empirical comparison in Tables 1-2 against external baselines, not an analytic consequence of the formulation. The only self-citations are to Anim-NeRF pose refinement and Instant-NVR as related work; pose refinement is an input preprocessing step, not evidence for the method's novelty, and the comparison numbers come from external baselines. The concern that baselines may not have used identical refined poses is an experimental-control issue, not circular reasoning. Therefore the derivation is self-contained with no circular step.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The method's empirical success depends mainly on hand-tuned hyperparameters and on assumptions about pose accuracy and monocular observability. No free physical constants or invented entities are introduced.

free parameters (5)
  • Loss weights in Eq. (16) = lambda_rgb=1.5, lambda_ssim=0.2, lambda_lpips=0.05, lambda_pos=0.01, lambda_scaling=1, lambda_offset=1
    Hand-set to balance photometric and regularization terms; they directly affect rendering quality and are not derived from data.
  • Position and scaling thresholds = epsilon_pos=1, epsilon_scaling=0.6
    Chosen by hand to keep Gaussians near their parent triangles and prevent elongation; they control the regularization losses in Eqs. (13) and (14).
  • Optimizer hyperparameters = position LR 0.008 decaying to 1e-5, scale 0.017, rotation 0.001, opacity 0.05, MLP 1e-4
    Standard 3DGS and SplattingAvatar settings re-used here; no sensitivity analysis is provided.
  • Densification and pruning schedule = density control every 500 iterations, opacity reset every 5000 from iteration 10,000, off at iteration 35,000
    Copied from SplattingAvatar; determines the number and distribution of Gaussians per triangle.
  • Training setup choices = 50,000 iterations; ZJU-MoCap sampled 1 of every 4 frames; clip selected with a complete turning action
    These choices affect benchmark comparisons and are not justified by sensitivity analysis.
assumptions (5)
  • domain assumption SMPL template plus LBS is an adequate body prior; clothing and hair deviations are encoded by Gaussians and the rectification MLP.
    Gaussians are bound to SMPL triangles; the method does not model cloth or hair meshes, and the Conclusion states limited ability to model these details.
  • domain assumption LBS cannot capture non-rigid deformation, and the residual is a deterministic function of pose and Gaussian position learnable by a 5-layer MLP.
    This is the central motivation of Section 3.3; if the residual is too complex or multi-modal, the MLP cannot learn it.
  • domain assumption Monocular video provides sufficient constraints to optimize per-scene Gaussians and the MLP without significant overfitting.
    The Introduction calls monocular reconstruction under-constrained; no explicit regularization beyond offset and scaling losses is provided.
  • standard math 3D Gaussian Splatting's differentiable rasterization and covariance projection (Eqs. 1 to 4) are correct building blocks.
    The method relies on 3DGS as a black-box renderer and does not re-derive it.
  • domain assumption Anim-NeRF pose refinement gives poses accurate enough for mesh-guided Gaussian placement.
    Section 4.1 states refined poses from Anim-NeRF are used; Section 4.3 notes there are pose errors in ZJU-MoCap.
invented entities (1)
  • None
    purpose: No new physical or conceptual entities are introduced.
    The Gaussian rectification module is a learned MLP, not a new particle, force, dimension, or conserved quantity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RMAvatar: Photorealistic Human Avatar Reconstruction from Monocular Video Based on Rectified Mesh-embedded Gaussians." pith.science (2026). https://pith.science/paper/QVOMKUPI

@misc{pith2026250107104,
  author       = {Pith},
  title        = {Pith review of: RMAvatar: Photorealistic Human Avatar Reconstruction from Monocular Video Based on Rectified Mesh-embedded Gaussians},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QVOMKUPI}},
  note         = {Machine review of arXiv:2501.07104}
}
read the original abstract

We introduce RMAvatar, a novel human avatar representation with Gaussian splatting embedded on mesh to learn clothed avatar from a monocular video. We utilize the explicit mesh geometry to represent motion and shape of a virtual human and implicit appearance rendering with Gaussian Splatting. Our method consists of two main modules: Gaussian initialization module and Gaussian rectification module. We embed Gaussians into triangular faces and control their motion through the mesh, which ensures low-frequency motion and surface deformation of the avatar. Due to the limitations of LBS formula, the human skeleton is hard to control complex non-rigid transformations. We then design a pose-related Gaussian rectification module to learn fine-detailed non-rigid deformations, further improving the realism and expressiveness of the avatar. We conduct extensive experiments on public datasets, RMAvatar shows state-of-the-art performance on both rendering quality and quantitative evaluations. Please see our project page at https://rm-avatar.github.io.

Figures

Figures reproduced from arXiv: 2501.07104 by the authors.

Figure 1
Figure 1. Overview of RMAvatar. Given a sequence of monocular images and a neutral SMPL template, we first obtain the deformed mesh under current pose of the person via SMPL tracking. Our initialization strategy is to embed Gaussian splats on mesh in the local coordinates of each triangle and then transform them to world space based on triangle’s shape. To improve the representation ability of Gaussian splats for non-rigid cl… view at source ↗
Figure 2
Figure 2. Comparison of novel view synthesis on PeopleSnapshot [2]. Our method is able to reconstruct intricate texture details. follow the same data split. We compare our approach with Anim-NeRF [4], InstantAvatar [17], GaussianAvatar [13] and SplattingAvatar [48] on this dataset. For fair compari￾son, we use the provided poses optimized by Anim-NeRF [4] in our training and inference. Sen Peng et al.: Preprint submitted to E… view at source ↗
Figure 3
Figure 3. Comparison of novel view synthesis on ZJU-MoCap [41]. Our method reconstructs complicated cloth textures. ZJU-MoCap [41] Dataset. We use six subjects (377, 386, 387, 392, 393, and 394) [56] of the ZJU-MoCap dataset and compare with GaussianAvatar [13] and SplattingA￾vatar [48] on monocular videos of the dataset. We select a video from a camera viewpoint and use a segment of the video during training. Specifically, w… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Comparison of animation on out-of-distribution poses on PeopleSnapshot [2]. Our method generates consistent representations for avatars on novel poses. 𝛿𝑟. In order to limit the offset value to be small, we design offset loss for these three attributes. 4.3. Comparison…
Figure 5
Figure 5. Figure 5: Ablation on female-3-casual of PeopleSnapshot. Our full model mitigates rendering artifacts with Gaussian rectification module and regularization losses [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. DevilSight: Augmenting Monocular Human Avatar Reconstruction through a Virtual Perspective

    cs.CV 2025-08 reject novelty 5.0 of 10

    A monocular human avatar reconstruction method generates pseudo back-view videos with a fine-tuned diffusion model and uses them as extra training data for a 3D Gaussian avatar.

Reference graph

Works this paper leans on

78 extracted references · 69 canonical work pages · cited by 1 Pith paper

  1. [4]

    Animatable neural radiance fields from monocular rgb videos

    Chen,J.,Zhang,Y.,Kang,D.,Zhe,X.,Bao,L.,Jia,X.,Lu,H.,2021b. Animatable neural radiance fields from monocular rgb videos. arXiv preprint arXiv:2106.13629

  2. [40]

    Animatable neural radiance fields for modeling dynamic human bodies, in: Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pp

    Peng, S., Dong, J., Wang, Q., Zhang, S., Shuai, Q., Zhou, X., Bao, H., 2021a. Animatable neural radiance fields for modeling dynamic human bodies, in: Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pp. 14314–14323

  3. [45]

    3dgs- avatar: Animatable avatars via deformable 3d gaussian splatting, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Qian,Z.,Wang,S.,Mihajlovic,M.,Geiger,A.,Tang,S.,2024b. 3dgs- avatar: Animatable avatars via deformable 3d gaussian splatting, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 5020–5030

  4. [55]

    Wen, J., Zhao, X., Ren, Z., Schwing, A.G., Wang, S., 2024. Goma- vatar: Efficient animatable human modeling from monocular video using gaussians-on-mesh, in: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, pp. 2059–2069

  5. [26]

    Gart: Gaussian articulated template models, in: Proceedings of the IEEE/CVFConferenceonComputerVisionandPatternRecognition, pp

    Lei, J., Wang, Y., Pavlakos, G., Liu, L., Daniilidis, K., 2024. Gart: Gaussian articulated template models, in: Proceedings of the IEEE/CVFConferenceonComputerVisionandPatternRecognition, pp. 19876–19887

  6. [1]

    Learningtoreconstructpeopleinclothingfromasinglergb camera, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Alldieck, T., Magnor, M., Bhatnagar, B.L., Theobalt, C., Pons-Moll, G.,2019. Learningtoreconstructpeopleinclothingfromasinglergb camera, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 1175–1186

  7. [2]

    Video based reconstruction of 3d people models, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Alldieck,T.,Magnor,M.,Xu,W.,Theobalt,C.,Pons-Moll,G.,2018. Video based reconstruction of 3d people models, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 8387–8397

  8. [3]

    14124–14133

    Chen, A., Xu, Z., Zhao, F., Zhang, X., Xiang, F., Yu, J., Su, H., 2021a.Mvsnerf:Fastgeneralizableradiancefieldreconstructionfrom multi-view stereo, in: Proceedings of the IEEE/CVF international conference on computer vision, pp. 14124–14133

Show all 78 references
  1. [5]

    Fast-snarf: A fast deformer for articulated neural fields

    Chen, X., Jiang, T., Song, J., Rietmann, M., Geiger, A., Black, M.J., Hilliges, O., 2023. Fast-snarf: A fast deformer for articulated neural fields. IEEE Transactions on Pattern Analysis and Machine Intelligence 45, 11796–11809

  2. [6]

    Monogaussianavatar: Monocular gaussian point-based head avatar, in: ACM SIGGRAPH 2024 Conference Papers, pp

    Chen, Y., Wang, L., Li, Q., Xiao, H., Zhang, S., Yao, H., Liu, Y., 2024a. Monogaussianavatar: Monocular gaussian point-based head avatar, in: ACM SIGGRAPH 2024 Conference Papers, pp. 1–9

  3. [7]

    Monogaussianavatar: Monocular gaussian point-based head avatar, in: ACM SIGGRAPH Conference Proceedings, pp

    Chen, Y., Wang, L., Li, Q., Xiao, H., Zhang, S., Yao, H., Liu, Y., 2024b. Monogaussianavatar: Monocular gaussian point-based head avatar, in: ACM SIGGRAPH Conference Proceedings, pp. 1–9

  4. [8]

    arXiv preprint arXiv:2407.08414

    Chen,Y.,Zheng,Z.,Li,Z.,Xu,C.,Liu,Y.,2024c.Meshavatar:Learn- ing high-quality triangular human avatars from multi-view videos. arXiv preprint arXiv:2407.08414

  5. [9]

    Stereo radiancefields(srf):Learningviewsynthesisforsparseviewsofnovel scenes, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Chibane, J., Bansal, A., Lazova, V., Pons-Moll, G., 2021. Stereo radiancefields(srf):Learningviewsynthesisforsparseviewsofnovel scenes, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 7911–7920

  6. [10]

    Feng, Y., Liu, W., Bolkart, T., Yang, J., Pollefeys, M., Black, M.J.,

  7. [11]

    Learning neural volumetric representations of dynamic humans in minutes, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Geng, C., Peng, S., Xu, Z., Bao, H., Zhou, X., 2023. Learning neural volumetric representations of dynamic humans in minutes, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 8759–8770

  8. [12]

    Hdhumans: A hybrid approach for high-fidelity digital humans

    Habermann, M., Liu, L., Xu, W., Pons-Moll, G., Zollhoefer, M., Theobalt, C., 2023. Hdhumans: A hybrid approach for high-fidelity digital humans. Proceedings of the ACM on Computer Graphics and Interactive Techniques 6, 1–23

  9. [13]

    Hu, L., Zhang, H., Zhang, Y., Zhou, B., Liu, B., Zhang, S., Nie, L., 2024a. Gaussianavatar: Towards realistic human avatar modeling from a single video via animatable 3d gaussians, in: Proceedings of theIEEE/CVFConferenceonComputerVisionandPatternRecogni- tion, pp. 634–644

  10. [14]

    Gauhuman:Articulatedgaussiansplat- tingfrommonocularhumanvideos,in:ProceedingsoftheIEEE/CVF ConferenceonComputerVisionandPatternRecognition,pp.20418– 20431

    Hu,S.,Hu,T.,Liu,Z.,2024b. Gauhuman:Articulatedgaussiansplat- tingfrommonocularhumanvideos,in:ProceedingsoftheIEEE/CVF ConferenceonComputerVisionandPatternRecognition,pp.20418– 20431

  11. [15]

    Humanrf: High-fidelity neural radiancefieldsforhumansinmotion

    Işık, M., Rünz, M., Georgopoulos, M., Khakhulin, T., Starck, J., Agapito, L., Nießner, M., 2023. Humanrf: High-fidelity neural radiancefieldsforhumansinmotion. ACMTransactionsonGraphics (TOG) 42, 1–12

  12. [16]

    Selfrecon: Self reconstructionyourdigitalavatarfrommonocularvideo,in:Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Jiang, B., Hong, Y., Bao, H., Zhang, J., 2022a. Selfrecon: Self reconstructionyourdigitalavatarfrommonocularvideo,in:Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 5605–5615

  13. [17]

    Instantavatar: Learning avatars from monocular video in 60 seconds, in: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Jiang, T., Chen, X., Song, J., Hilliges, O., 2023. Instantavatar: Learning avatars from monocular video in 60 seconds, in: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 16922–16932

  14. [18]

    Neu- man: Neural human radiance field from a single video, in: European Conference on Computer Vision, Springer

    Jiang, W., Yi, K.M., Samei, G., Tuzel, O., Ranjan, A., 2022b. Neu- man: Neural human radiance field from a single video, in: European Conference on Computer Vision, Springer. pp. 402–418

  15. [19]

    Panoptic studio: A massively multiview system for social motion capture, in: Proceedings of the IEEE international conference on computer vision, pp

    Joo, H., Liu, H., Tan, L., Gui, L., Nabbe, B., Matthews, I., Kanade, T., Nobuhara, S., Sheikh, Y., 2015. Panoptic studio: A massively multiview system for social motion capture, in: Proceedings of the IEEE international conference on computer vision, pp. 3334–3342

  16. [20]

    Virtual tour: An immersive low cost telepresence system, in: 2020 IEEE conference on virtual reality and 3D user interfaces abstracts and workshops (VRW), IEEE

    Kachach,R.,Perez,P.,Villegas,A.,Gonzalez-Sosa,E.,2020. Virtual tour: An immersive low cost telepresence system, in: 2020 IEEE conference on virtual reality and 3D user interfaces abstracts and workshops (VRW), IEEE. pp. 504–506

  17. [21]

    3d gaussiansplattingforreal-timeradiancefieldrendering

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G., 2023. 3d gaussiansplattingforreal-timeradiancefieldrendering. ACMTrans. Graph. 42, 139–1

  18. [22]

    Amethodforstochasticoptimiza- tion,in:Internationalconferenceonlearningrepresentations(ICLR), San Diego, California

    Kinga,D.,Adam,J.B.,etal.,2015. Amethodforstochasticoptimiza- tion,in:Internationalconferenceonlearningrepresentations(ICLR), San Diego, California;. p. 6

  19. [23]

    Hugs: Human gaussian splats, in: Proceedings of the IEEE/CVF conferenceoncomputervisionandpatternrecognition,pp.505–515

    Kocabas, M., Chang, J.H.R., Gabriel, J., Tuzel, O., Ranjan, A., 2024. Hugs: Human gaussian splats, in: Proceedings of the IEEE/CVF conferenceoncomputervisionandpatternrecognition,pp.505–515

  20. [24]

    Deliffas: Deformable light fields for fast avatar synthesis

    Kwon, Y., Liu, L., Fuchs, H., Habermann, M., Theobalt, C., 2024. Deliffas: Deformable light fields for fast avatar synthesis. Advances in Neural Information Processing Systems 36

  21. [25]

    1440–1449

    Lassner,C.,Zollhofer,M.,2021.Pulsar:Efficientsphere-basedneural rendering,in:ProceedingsoftheIEEE/CVFConferenceonComputer Vision and Pattern Recognition, pp. 1440–1449

  22. [27]

    Pose space deformation: a unifiedapproachtoshapeinterpolationandskeleton-drivendeforma- tion, in: Seminal Graphics Papers: Pushing the Boundaries, Volume 2, pp

    Lewis, J.P., Cordner, M., Fong, N., 2023. Pose space deformation: a unifiedapproachtoshapeinterpolationandskeleton-drivendeforma- tion, in: Seminal Graphics Papers: Pushing the Boundaries, Volume 2, pp. 811–818

  23. [28]

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

    Li, T., Bolkart, T., Black, M.J., Li, H., Romero, J., 2017. Learning a model of facial shape and expression from 4D scans. ACM Transac- tionsonGraphics,(Proc.SIGGRAPHAsia)36,194:1–194:17. URL: https://doi.org/10.1145/3130800.3130813

  24. [29]

    Li, Z., Zheng, Z., Wang, L., Liu, Y., 2024. 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, pp. 19711–19722

  25. [30]

    Avatarcap: Animatableavatarconditionedmonocularhumanvolumetriccapture, in:EuropeanConferenceonComputerVision,Springer.pp.322–341

    Li, Z., Zheng, Z., Zhang, H., Ji, C., Liu, Y., 2022. Avatarcap: Animatableavatarconditionedmonocularhumanvolumetriccapture, in:EuropeanConferenceonComputerVision,Springer.pp.322–341

  26. [31]

    Lin,S.,Li,Z.,Su,Z.,Zheng,Z.,Zhang,H.,Liu,Y.,2024.Layga:Lay- eredgaussianavatarsforanimatableclothingtransfer,in:SIGGRAPH Conference Papers, pp. 1–11

  27. [32]

    Loper, M., Mahmood, N., Romero, J., Pons-Moll, G., Black, M.J.,

  28. [33]

    Leap: Learning articulated occupancy of people, in: Proceedings of the IEEE/CVFConferenceonComputerVisionandPatternRecognition, pp

    Mihajlovic, M., Zhang, Y., Black, M.J., Tang, S., 2021. Leap: Learning articulated occupancy of people, in: Proceedings of the IEEE/CVFConferenceonComputerVisionandPatternRecognition, pp. 10461–10471

  29. [34]

    Nerf: Representing scenes as neural radiance fields for view synthesis

    Mildenhall,B.,Srinivasan,P.P.,Tancik,M.,Barron,J.T.,Ramamoor- thi, R., Ng, R., 2021. Nerf: Representing scenes as neural radiance fields for view synthesis. Communications of the ACM 65, 99–106

  30. [35]

    Human gaussian splatting: Real-time rendering of ani- matable avatars, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Moreau,A.,Song,J.,Dhamo,H.,Shaw,R.,Zhou,Y.,Pérez-Pellitero, E., 2024. Human gaussian splatting: Real-time rendering of ani- matable avatars, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 788–798

  31. [36]

    Instant neural graphics primitives with a multiresolution hash encoding

    Müller, T., Evans, A., Schied, C., Keller, A., 2022. Instant neural graphics primitives with a multiresolution hash encoding. ACM transactions on graphics (TOG) 41, 1–15

  32. [37]

    Niemeyer, M., Mescheder, L., Oechsle, M., Geiger, A., 2020. Differ- entiable volumetric rendering: Learning implicit 3d representations without3dsupervision,in:ProceedingsoftheIEEE/CVFconference on computer vision and pattern recognition, pp. 3504–3515

  33. [38]

    Unisurf: Unifying neural implicitsurfacesandradiancefieldsformulti-viewreconstruction,in: ProceedingsoftheIEEE/CVFInternationalConferenceonComputer Vision, pp

    Oechsle, M., Peng, S., Geiger, A., 2021. Unisurf: Unifying neural implicitsurfacesandradiancefieldsformulti-viewreconstruction,in: ProceedingsoftheIEEE/CVFInternationalConferenceonComputer Vision, pp. 5589–5599

  34. [39]

    Expressive body capture: 3D hands, face, and body from a single image, in: Proceedings IEEE Conf

    Pavlakos,G.,Choutas,V.,Ghorbani,N.,Bolkart,T.,Osman,A.A.A., Tzionas, D., Black, M.J., 2019. Expressive body capture: 3D hands, face, and body from a single image, in: Proceedings IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), pp. 10975– 10985

  35. [41]

    Peng, S., Zhang, Y., Xu, Y., Wang, Q., Shuai, Q., Bao, H., Zhou, X., 2021b. Neural body: Implicit neural representations with structured latentcodesfornovelviewsynthesisofdynamichumans,in:Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. ...

  36. [42]

    Dy- namic point fields, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Prokudin, S., Ma, Q., Raafat, M., Valentin, J., Tang, S., 2023. Dy- namic point fields, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 7964–7976

  37. [43]

    D-nerf:Neuralradiancefieldsfordynamicscenes,in:Proceedingsof theIEEE/CVFConferenceonComputerVisionandPatternRecogni- tion, pp

    Pumarola, A., Corona, E., Pons-Moll, G., Moreno-Noguer, F., 2021. D-nerf:Neuralradiancefieldsfordynamicscenes,in:Proceedingsof theIEEE/CVFConferenceonComputerVisionandPatternRecogni- tion, pp. 10318–10327

  38. [44]

    20299–20309

    Qian, S., Kirschstein, T., Schoneveld, L., Davoli, D., Giebenhain, S., Nießner,M.,2024a.Gaussianavatars:Photorealisticheadavatarswith rigged3dgaussians,in:ProceedingsoftheIEEE/CVFConferenceon Computer Vision and Pattern Recognition, pp. 20299–20309

  39. [46]

    Adop: Approximate differentiable one-pixel point rendering

    Rückert, D., Franke, L., Stamminger, M., 2022. Adop: Approximate differentiable one-pixel point rendering. ACM Transactions on Graphics (ToG) 41, 1–14. Sen Peng et al.:Preprint submitted to Elsevier Page 10 of 11

  40. [47]

    Control4d: Efficient 4d portrait editing with text, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Shao, R., Sun, J., Peng, C., Zheng, Z., Zhou, B., Zhang, H., Liu, Y., 2024a. Control4d: Efficient 4d portrait editing with text, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4556–4567

  41. [48]

    Splattingavatar: Realistic real-time human avatars with mesh-embedded gaussian splatting, in: Proceedings of theIEEE/CVFConferenceonComputerVisionandPatternRecogni- tion, pp

    Shao, Z., Wang, Z., Li, Z., Wang, D., Lin, X., Zhang, Y., Fan, M., Wang, Z., 2024b. Splattingavatar: Realistic real-time human avatars with mesh-embedded gaussian splatting, in: Proceedings of theIEEE/CVFConferenceonComputerVisionandPatternRecogni- tion, pp. 1606–1616

  42. [49]

    Npc: Neural point charactersfromvideo,in:ProceedingsoftheIEEE/CVFInternational Conference on Computer Vision, pp

    Su, S.Y., Bagautdinov, T., Rhodin, H., 2023. Npc: Neural point charactersfromvideo,in:ProceedingsoftheIEEE/CVFInternational Conference on Computer Vision, pp. 14795–14805

  43. [50]

    Gaussianeditor: Editing3dgaussiansdelicatelywithtextinstructions,in:Proceedings oftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition, pp

    Wang,J.,Fang,J.,Zhang,X.,Xie,L.,Tian,Q.,2024. Gaussianeditor: Editing3dgaussiansdelicatelywithtextinstructions,in:Proceedings oftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition, pp. 20902–20911

  44. [51]

    Fourier plenoctrees for dynamic radiance field rendering in real-time, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Wang,L.,Zhang,J.,Liu,X.,Zhao,F.,Zhang,Y.,Zhang,Y.,Wu,M., Yu, J., Xu, L., 2022a. Fourier plenoctrees for dynamic radiance field rendering in real-time, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13524–13534

  45. [52]

    Wang,S.,Schwarz,K.,Geiger,A.,Tang,S.,2022b.Arah:Animatable volumerenderingofarticulatedhumansdfs,in:Europeanconference on computer vision, Springer. pp. 1–19

  46. [53]

    Image qualityassessment:fromerrorvisibilitytostructuralsimilarity

    Wang, Z., Bovik, A.C., Sheikh, H.R., Simoncelli, E.P., 2004. Image qualityassessment:fromerrorvisibilitytostructuralsimilarity. IEEE transactions on image processing 13, 600–612

  47. [54]

    A systematic review on the visualization of avatars and agents in ar & vr displayed using head-mounted displays

    Weidner, F., Boettcher, G., Arboleda, S.A., Diao, C., Sinani, L., Kunert, C., Gerhardt, C., Broll, W., Raake, A., 2023. A systematic review on the visualization of avatars and agents in ar & vr displayed using head-mounted displays. IEEE Transactions on Visualization and Compu...

  48. [56]

    Humannerf: Free-viewpoint rendering of mov- ing people from monocular video, in: Proceedings of the IEEE/CVF conference on computer vision and pattern Recognition, pp

    Weng,C.Y.,Curless,B.,Srinivasan,P.P.,Barron,J.T.,Kemelmacher- Shlizerman, I., 2022. Humannerf: Free-viewpoint rendering of mov- ing people from monocular video, in: Proceedings of the IEEE/CVF conference on computer vision and pattern Recognition, pp. 16210– 16220

  49. [57]

    Synsin: End- to-end view synthesis from a single image, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Wiles, O., Gkioxari, G., Szeliski, R., Johnson, J., 2020. Synsin: End- to-end view synthesis from a single image, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 7467–7477

  50. [58]

    Multiface: Adatasetforneuralfacerendering

    Wuu, C.h., Zheng, N., Ardisson, S., Bali, R., Belko, D., Brockmeyer, E.,Evans,L.,Godisart,T.,Ha,H.,Huang,X.,etal.,2022. Multiface: Adatasetforneuralfacerendering. arXivpreprintarXiv:2207.11243

  51. [59]

    1802–1812

    Xiang,J.,Gao,X.,Guo,Y.,Zhang,J.,2024.Flashavatar:High-fidelity headavatarwithefficientgaussianembedding,in:Proceedingsofthe IEEE/CVFConferenceonComputerVisionandPatternRecognition, pp. 1802–1812

  52. [60]

    Xu,Q.,Xu,Z.,Philip,J.,Bi,S.,Shu,Z.,Sunkavalli,K.,Neumann,U.,

  53. [61]

    Gaussian head avatar: Ultra high-fidelity head avatar via dynamic gaussians, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Xu, Y., Chen, B., Li, Z., Zhang, H., Wang, L., Zheng, Z., Liu, Y., 2024a. Gaussian head avatar: Ultra high-fidelity head avatar via dynamic gaussians, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 1931–1941

  54. [62]

    Gphm: Gaussian parametric headmodelformonocularheadavatarreconstruction

    Xu, Y., Su, Z., Wu, Q., Liu, Y., 2024b. Gphm: Gaussian parametric headmodelformonocularheadavatarreconstruction. arXivpreprint arXiv:2407.15070

  55. [63]

    Relightableandanimatableneuralavatarfromsparse-view video, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Xu, Z., Peng, S., Geng, C., Mou, L., Yan, Z., Sun, J., Bao, H., Zhou, X.,2024c. Relightableandanimatableneuralavatarfromsparse-view video, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 990–1000

  56. [64]

    Differentiable surface splatting for point-based geometry processing

    Yifan,W.,Serena,F.,Wu,S.,Öztireli,C.,Sorkine-Hornung,O.,2019. Differentiable surface splatting for point-based geometry processing. ACM Transactions on Graphics (TOG) 38, 1–14

  57. [65]

    Monohuman: Animatable human neural field from monocular video, in: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Yu, Z., Cheng, W., Liu, X., Wu, W., Lin, K.Y., 2023. Monohuman: Animatable human neural field from monocular video, in: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 16943–16953

  58. [66]

    Routledge

    Zackariasson,P.,2012.TheVideoGameIndustry:Formation,Present State, and Future. Routledge

  59. [67]

    Differ- entiable point-based radiance fields for efficient view synthesis, in: SIGGRAPH Asia 2022 Conference Papers, pp

    Zhang, Q., Baek, S.H., Rusinkiewicz, S., Heide, F., 2022. Differ- entiable point-based radiance fields for efficient view synthesis, in: SIGGRAPH Asia 2022 Conference Papers, pp. 1–12

  60. [68]

    The unreasonableeffectivenessofdeepfeaturesasaperceptualmetric,in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp

    Zhang,R.,Isola,P.,Efros,A.A.,Shechtman,E.,Wang,O.,2018. The unreasonableeffectivenessofdeepfeaturesasaperceptualmetric,in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 586–595

  61. [69]

    ACM Transactions on Graphics (TOG) 41, 1–17

    Zhao, F., Jiang, Y., Yao, K., Zhang, J., Wang, L., Dai, H., Zhong, Y., Zhang,Y.,Wu,M.,Xu,L.,etal.,2022.Humanperformancemodeling and rendering via neural animated mesh. ACM Transactions on Graphics (TOG) 41, 1–17

  62. [70]

    Zheng, S., Zhou, B., Shao, R., Liu, B., Zhang, S., Nie, L., Liu, Y., 2024a. Gps-gaussian: Generalizable pixel-wise 3d gaussian splatting for real-time human novel view synthesis, in: Proceedings of the IEEE/CVFConferenceonComputerVisionandPatternRecognition, pp. 19680–19690

  63. [71]

    Im avatar: Implicit morphable head avatars from videos, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Zheng, Y., Abrevaya, V.F., Bühler, M.C., Chen, X., Black, M.J., Hilliges, O., 2022a. Im avatar: Implicit morphable head avatars from videos, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13545–13555

  64. [72]

    Pointavatar: Deformable point-based head avatars from videos, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Zheng, Y., Yifan, W., Wetzstein, G., Black, M.J., Hilliges, O., 2023. Pointavatar: Deformable point-based head avatars from videos, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 21057–21067

  65. [73]

    Physavatar: Learning the physics of dressed 3d avatars from visual observations,in:EuropeanConferenceonComputerVision(ECCV), pp

    Zheng, Y., Zhao, Q., Yang, G., Yifan, W., Xiang, D., Dubost, F., Lagun, D., Beeler, T., Tombari, F., Guibas, L., Wetzstein, G., 2024b. Physavatar: Learning the physics of dressed 3d avatars from visual observations,in:EuropeanConferenceonComputerVision(ECCV), pp. 262–284

  66. [74]

    Structured local radiance fields for human avatar modeling, in: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Zheng, Z., Huang, H., Yu, T., Zhang, H., Guo, Y., Liu, Y., 2022b. Structured local radiance fields for human avatar modeling, in: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 15893–15903

  67. [75]

    Ewa volume splatting,in:ProceedingsVisualization,2001.VIS’01.,IEEE.pp.29– 538

    Zwicker, M., Pfister, H., Van Baar, J., Gross, M., 2001. Ewa volume splatting,in:ProceedingsVisualization,2001.VIS’01.,IEEE.pp.29– 538. Sen Peng et al.:Preprint submitted to Elsevier Page 11 of 11

  68. [2015]

    ACM Trans

    SMPL: A skinned multi-person linear model. ACM Trans. Graphics (Proc. SIGGRAPH Asia) 34, 248:1–248:16

  69. [2022]

    5438–5448

    Point-nerf: Point-based neural radiance fields, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recog- nition, pp. 5438–5448

  70. [2023]

    arXiv preprint arXiv:2309.06441

    Learning disentangled avatars with hybrid 3d representations. arXiv preprint arXiv:2309.06441 . Sen Peng et al.:Preprint submitted to Elsevier Page 9 of 11

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.