pith. sign in

arxiv: 2606.06685 · v1 · pith:QXAUPEYDnew · submitted 2026-06-04 · 💻 cs.CV · cs.GR

RigPAPR: Rig-Based Animation of Static Neural Point Clouds from a Fixed-Viewpoint Video

Shichong Peng , Yanshu Zhang , Ke Li This is my paper

Pith reviewed 2026-06-28 01:48 UTC · model grok-4.3

classification 💻 cs.CV cs.GR
keywords neural point renderinganimation from videolinear blend skinningjoint boundary artifactsauto-riggingnovel view synthesisPAPRfixed viewpoint driving
0
0 comments X

The pith

RigPAPR animates static neural point clouds from a fixed-viewpoint video by auto-rigging and applying direct linear blend skinning without mesh proxies or per-primitive corrections.

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

The paper seeks to turn a high-fidelity static neural point reconstruction into a rigged, re-posable 3D asset driven by a single monocular video. Existing Gaussian splatting and mesh-based methods produce gaps and spikes at joints because each primitive carries a fixed shape calibrated in the canonical pose. RigPAPR instead builds on proximity attention point rendering, which carries no per-primitive shape and recomposes every pixel at render time from the deformed positions of the primitives. This lets the surface reform naturally under rigid linear blend skinning, yielding cleaner boundaries and higher novel-view quality without templates or pose-dependent fixes.

Core claim

We present RigPAPR, which auto-rigs a static PAPR cloud and drives it under direct LBS from a single fixed-viewpoint video, without mesh proxy, pose-dependent correction, or category template. On synthetic subjects, RigPAPR matches the strongest baseline at the supervised view and exceeds mesh-based and Gaussian-splatting baselines at novel views by 3+dB PSNR, with cleaner joint-boundary renderings of both synthetic and real subjects.

What carries the argument

Proximity Attention Point Rendering (PAPR), which recomposes each pixel at render time from the positions of deformed primitives rather than storing per-primitive shapes that must tile in the canonical pose.

If this is right

  • A static PAPR reconstruction can be turned into a drivable asset from one fixed-view video without category templates.
  • Direct LBS on PAPR primitives produces fewer joint-boundary artifacts than the same skinning applied to Gaussian splats or meshes.
  • Novel-view PSNR improves by more than 3 dB over mesh and Gaussian baselines while matching supervised-view performance.
  • The same pipeline works for both synthetic and real captured subjects.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The absence of fixed per-primitive shapes may allow PAPR-style rendering to handle non-rigid deformations more gracefully than shape-carrying primitives.
  • Auto-rigging from a single view could be combined with image-to-video models to create 3D assets from generated driving sequences.
  • The method may reduce reliance on explicit mesh proxies in animation pipelines that start from neural point clouds.

Load-bearing premise

That recomposing pixels from the deformed positions of the primitives at render time will automatically restore surface continuity at joints under rigid linear blend skinning.

What would settle it

Rendering the rigged PAPR model under strong articulation and observing persistent gaps or spikes at joint boundaries in the output images.

Figures

Figures reproduced from arXiv: 2606.06685 by Ke Li, Shichong Peng, Yanshu Zhang.

Figure 1
Figure 1. Figure 1: Gaussian splats vs. an interpolation-based representation under LBS. Left: each splat carries its own fixed individual shape; under LBS it is reoriented by its bound bone but cannot bend with the surface. Neighbouring splats therefore stop tiling at the joint boundary even when their centers stay correctly on the deformed surface: the longer splat (red) protrudes past its neighbour as a spike, while the na… view at source ↗
Figure 2
Figure 2. Figure 2: RigPAPR overview. A mesh proxy extracted from the PAPR cloud is auto-rigged and the resulting skeleton and initial skinning weights are transferred back onto the points; the proxy is then discarded (§3.3). A two-phase optimization drives the rigged cloud from the fixed-viewpoint video, recovering per-frame joint rotations and refining the skinning weights (§3.4). At inference, LBS deforms the canonical poi… view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative comparison on synthetic T-Rex. Rows are methods, columns are frames; left block is the supervised Training View, right block is a held-out Novel View. Column t0 is the canonical (binding) frame; t1, t2 are progressively articulated frames. Red highlights mark Puppeteer’s joint￾rotation errors and Mani-GS’s articulation artifacts; we encourage zooming in for detail. Training View Novel View t0 (… view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison on the real-world robot capture. Rows are methods (Mani-GS, ours); columns are matched frames split into the supervised Training View (left) and a held-out Novel View (right). t0 is the canonical (binding) frame; t1, t2 are progressively articulated. Training-view ground truth is provided in the appendix ( [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Novel pose driving. Both methods are driven at test time by two pose sequences not seen during the Phase 2 fitting trajectory. Rows are methods; columns show the shared canonical (binding) frame t0 followed by two frames from each driving sequence. No ground truth is available for the driven poses. Mani-GS produces visible artifacts around the arms and legs at the new poses; ours generalizes cleanly. Best … view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative comparison on synthetic fox. Training View is on top and Novel View is on the bottom. Rows are methods; columns are matched frames, with t0 the canonical (binding) frame and t1, t2 progressively articulated. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative comparison on synthetic Simpsons. Rows are methods; columns are matched frames split into a supervised Training View (left) and a held-out Novel View (right). t0 is the canonical (binding) frame; t1, t2 are progressively articulated frames. 16 [PITH_FULL_IMAGE:figures/full_fig_p016_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative comparison on synthetic wolf. Training View is on top and Novel View is on the bottom. Rows are methods; columns are matched frames, with t0 the canonical (binding) frame and t1, t2 progressively articulated. 17 [PITH_FULL_IMAGE:figures/full_fig_p017_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative comparison on synthetic spider. Rows are methods; columns are matched frames split into a supervised Training View (left) and a held-out Novel View (right). t0 is the canonical (binding) frame and t1 is the articulated frame shown; the additional t2 is omitted to allow larger images. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Qualitative comparison on the real-world robot capture (with training-view ground truth). Rows are matched frames (t0 is the canonical/binding frame; t1, t2 are progressively artic￾ulated); columns are split into a supervised Training View (Mani-GS [7], ours, ground truth) and a held-out Novel View (Mani-GS, ours). 19 [PITH_FULL_IMAGE:figures/full_fig_p019_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Qualitative comparison on the real-world statue capture. Rows are matched frames (t0 is the canonical/binding frame; t1, t2 are progressively articulated); columns are split into a supervised Training View (Mani-GS [7], ours, ground truth) and a held-out Novel View (Mani-GS, ours). 20 [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Phase configurations on novel-pose driving (real-world robot). Rows are time steps: the canonical (binding) frame t0, then t1 and t2 from each of the two unseen driving sequences. Columns group the four configurations: Mani-GS at Phase 1 and Phase 2, then ours at Phase 1 and Phase 2. Mani-GS Phase 2 shows the per-splat overfit artifacts on the limbs visible in the main paper ( [PITH_FULL_IMAGE:figures/fu… view at source ↗
Figure 13
Figure 13. Figure 13: Two-phase optimization ablation on T-Rex. Training View is on top and Novel View is on the bottom. Rows are configurations (Phase 1 w/o L∆depth, Phase 1, Full, GT); columns are matched frames with t0 the canonical (binding) frame and t1, t2 progressively articulated. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Mani-GS adaptive-scaling ablation. Simpsons (top) and fox (bottom), each shown as two rows comparing the default Mani-GS configuration (per-triangle adaptive scaling on) with the configuration used in the main paper (scaling off). Both rows share our full two-phase optimization; only the scaling toggle differs. Columns are matched training-view frames, with t0 the canonical (binding) frame and t1, t2 prog… view at source ↗
Figure 15
Figure 15. Figure 15: Renderings under an oracle (ground-truth) mesh deformation. All three methods are driven by the same ground-truth mesh sequence; the full setup, including the shared subdivided mesh used for both baselines, is described in Section B.3. The figure is laid out as two side-by-side panels: T-Rex (top) above Fox (bottom) on the left, and Simpsons on the right; within each subject sub-block, rows are methods an… view at source ↗
read the original abstract

Static neural point reconstructions capture a subject at high fidelity from posed images. Given such a reconstruction, we aim to animate it to follow a monocular fixed-viewpoint driving video of the subject, whether captured or produced by image-to-video (I2V) generation, and to recover a rigged, re-posable 3D asset. Existing methods deform Gaussian splats through direct linear blend skinning (LBS) or mesh proxies, both of which are prone to joint-boundary artifacts under articulation, even with per-primitive corrections. We trace the artifact to the representation: each splat carries an individual shape calibrated in the canonical pose to tile with its neighbours. Under rigid LBS, each splat moves with its bone but cannot bend, so the canonical tiling breaks at joint boundaries into gaps and spikes. Proximity attention point rendering (PAPR) instead carries no per-primitive shape; each pixel is recomposed at render time from the deformed primitives' positions, so the surface re-forms naturally with the articulation. We present RigPAPR, which auto-rigs a static PAPR cloud and drives it under direct LBS from a single fixed-viewpoint video, without mesh proxy, pose-dependent correction, or category template. On synthetic subjects, RigPAPR matches the strongest baseline at the supervised view and exceeds mesh-based and Gaussian-splatting baselines at novel views by 3+dB PSNR, with cleaner joint-boundary renderings of both synthetic and real subjects.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The paper introduces RigPAPR, a pipeline that auto-rigs a static PAPR (Proximity Attention Point Rendering) point cloud and drives it via direct linear blend skinning (LBS) from a single fixed-viewpoint video. It contrasts PAPR's position-only primitives, whose surface is recomposed per-pixel at render time, against Gaussian splats and mesh proxies whose fixed per-primitive shapes produce joint-boundary gaps and spikes under articulation. On synthetic subjects the method matches the strongest baseline at the supervised view and exceeds mesh-based and Gaussian-splatting baselines by more than 3 dB PSNR at novel views, while producing qualitatively cleaner joint boundaries on both synthetic and real subjects.

Significance. If the reported gains and artifact reduction hold under a reproducible protocol, the work supplies a concrete, template-free alternative to per-primitive correction or mesh proxy methods for animating neural point reconstructions. The central representational distinction—removing fixed per-primitive geometry so that the surface re-forms naturally under LBS—is internally consistent and directly addresses the mechanism that generates the cited artifacts.

minor comments (3)
  1. [Abstract] The abstract states quantitative PSNR gains and qualitative improvements but supplies no evaluation protocol, list of baselines, number of subjects, or error analysis; these details are required to assess whether the 3 dB margin is robust or subject to post-hoc selection.
  2. The claim that PAPR 'carries no per-primitive shape' would benefit from an explicit equation or pseudocode fragment showing how the per-pixel recomposition is performed after LBS deformation (e.g., the weighting or attention mechanism).
  3. Figure captions and table headers should explicitly label which views are supervised versus novel and which baselines are mesh-based versus splatting-based to allow direct comparison with the text claims.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary, significance assessment, and recommendation of minor revision. The referee's description of the method, its contrast to Gaussian splats and mesh proxies, and the reported gains accurately reflect the manuscript.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper introduces RigPAPR as an empirical pipeline that auto-rigs a static PAPR point cloud and drives it via direct LBS from a fixed-viewpoint video. The core distinction (PAPR recomposes pixels from deformed positions at render time, unlike per-primitive shapes in Gaussian splats) is a representational property stated directly in the abstract and does not reduce any reported PSNR or artifact claim to a fitted parameter or self-citation by construction. Performance numbers are presented as measured outcomes on synthetic and real subjects against baselines; no equations, uniqueness theorems, or ansatzes are shown to collapse into the inputs. The method is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no explicit free parameters, axioms, or invented entities; the method is described as building directly on existing PAPR without additional postulated constructs.

pith-pipeline@v0.9.1-grok · 5806 in / 1257 out tokens · 48708 ms · 2026-06-28T01:48:21.520712+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

59 extracted references · 4 canonical work pages · 2 internal anchors

  1. [1]

    Automatic rigging and animation of 3D characters.ACM Transactions on Graphics, 26(3):72:1–72:8, 2007

    Ilya Baran and Jovan Popovi ´c. Automatic rigging and animation of 3D characters.ACM Transactions on Graphics, 26(3):72:1–72:8, 2007

    2007

  2. [2]

    Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets

    Andreas Blattmann, Tim Dockhorn, Sumith Kulal, Daniel Mendelevitch, Maciej Kilian, Do- minik Lorenz, Yam Levi, Zion English, Vikram V oleti, Adam Letts, Varun Jampani, and Robin Rombach. Stable video diffusion: Scaling latent video diffusion models to large datasets.arXiv preprint arXiv:2311.15127, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  3. [3]

    Video generation models as world simulators, 2024

    Tim Brooks, Bill Peebles, Connor Holmes, Will DePue, Yufei Guo, Li Jing, David Schnurr, Joe Taylor, Troy Luhman, Eric Luhman, Clarence Ng, Ricky Wang, and Aditya Ramesh. Video generation models as world simulators, 2024

    2024

  4. [4]

    SV-GS: Sparse view 4D reconstruction with skeleton-driven gaussian splatting, 2026

    Jun-Jee Chao and V olkan Isler. SV-GS: Sparse view 4D reconstruction with skeleton-driven gaussian splatting, 2026

    2026

  5. [5]

    Kling-Omni technical report, 2025

    Jialu Chen, Yuanzheng Ci, Xiangyu Du, Zipeng Feng, Kun Gai, Sainan Guo, Feng Han, Jing He, Kang He, Xiao Hu, Xiaohuan Hu, Boyuan Jiang, Fangyuan Kong, Hang Li, Jie Li, Qingyu Li, Shen Li, Xiaohan Li, Yan Qiang Li, Jiajun Liang, Borui Liao, Yiqiao Liao, Weihong Lin, Quande Liu, Xiaokun Liu, Yilun Liu, Yuliang Liu, Shun Lu, Hangyu Mao, Yu-Fei Mao, Haodong O...

    2025

  6. [6]

    Objaverse: A universe of annotated 3d objects

    Matt Deitke, Dustin Schwenk, Jordi Salvador, Luca Weihs, Oscar Michel, Eli VanderBilt, Ludwig Schmidt, Kiana Ehsani, Aniruddha Kembhavi, and Ali Farhadi. Objaverse: A universe of annotated 3D objects.arXiv preprint arXiv:2212.08051, 2022

    work page arXiv 2022

  7. [7]

    Mani-GS: Gaussian splatting manipulation with triangular mesh

    Xiangjun Gao, Xiaoyu Li, Yiyu Zhuang, Qi Zhang, Wenbo Hu, Chaopeng Zhang, Yao Yao, Ying Shan, and Long Quan. Mani-GS: Gaussian splatting manipulation with triangular mesh. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 21392– 21402, 2025

    2025

  8. [8]

    Diffusion as shader: 3D-aware video diffusion for versatile video generation control

    Zekai Gu, Rui Yan, Jiahao Lu, Peng Li, Zhiyang Dou, Chenyang Si, Zhen Dong, Qifeng Liu, Cheng Lin, Ziwei Liu, Wenping Wang, and Yuan Liu. Diffusion as shader: 3D-aware video diffusion for versatile video generation control. InACM SIGGRAPH 2025 Conference Papers, pages 62:1–62:12. ACM, 2025

    2025

  9. [9]

    SuGaR: Surface-aligned gaussian splatting for efficient 3D mesh reconstruction and high-quality mesh rendering

    Antoine Guédon and Vincent Lepetit. SuGaR: Surface-aligned gaussian splatting for efficient 3D mesh reconstruction and high-quality mesh rendering. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5354–5363, 2024

    2024

  10. [10]

    Real-time meshless deformation.Computer Animation and Virtual Worlds, 16(3-4):189–200, 2005

    Xiaohu Guo and Hong Qin. Real-time meshless deformation.Computer Animation and Virtual Worlds, 16(3-4):189–200, 2005

    2005

  11. [11]

    SC- GS: Sparse-controlled Gaussian splatting for editable dynamic scenes

    Yi-Hua Huang, Yang-Tian Sun, Ziyi Yang, Xiaoyang Lyu, Yan-Pei Cao, and Xiaojuan Qi. SC- GS: Sparse-controlled Gaussian splatting for editable dynamic scenes. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 4220–4230, 2024

    2024

  12. [12]

    NeRFshop: Interactive editing of neural radiance fields.Proceedings of the ACM on Computer Graphics and Interactive Techniques, 6(1), May 2023

    Clément Jambon, Bernhard Kerbl, Georgios Kopanas, Stavros Diolatzis, Thomas Leimkühler, and George Drettakis. NeRFshop: Interactive editing of neural radiance fields.Proceedings of the ACM on Computer Graphics and Interactive Techniques, 6(1), May 2023

    2023

  13. [13]

    CoTracker3: Simpler and better point tracking by pseudo-labelling real videos

    Nikita Karaev, Iurii Makarov, Jianyuan Wang, Natalia Neverova, Andrea Vedaldi, and Christian Rupprecht. CoTracker3: Simpler and better point tracking by pseudo-labelling real videos. In IEEE/CVF International Conference on Computer Vision (ICCV), 2025

    2025

  14. [14]

    Skinning with dual quaternions

    Ladislav Kavan, Steven Collins, Jiˇrí Žára, and Carol O’Sullivan. Skinning with dual quaternions. InProceedings of the 2007 Symposium on Interactive 3D Graphics and Games, pages 39–46, 2007. 10

    2007

  15. [15]

    Geometric skinning with approximate dual quaternion blending.ACM Transactions on Graphics, 27(4):1–23, 2008

    Ladislav Kavan, Steven Collins, Ji ˇrí Žára, and Carol O’Sullivan. Geometric skinning with approximate dual quaternion blending.ACM Transactions on Graphics, 27(4):1–23, 2008

    2008

  16. [16]

    3D Gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics, 42(4), July 2023

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler, and George Drettakis. 3D Gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics, 42(4), July 2023

    2023

  17. [17]

    Rigidity-aware 3D gaussian deformation from a single image

    Jinhyeok Kim, Jaehun Bang, Seunghyun Seo, and Kyungdon Joo. Rigidity-aware 3D gaussian deformation from a single image. InSIGGRAPH Asia 2025 Conference Papers, pages 179:1– 179:11. ACM, 2025

    2025

  18. [18]

    Mor- phGS: Morphology-adaptive articulated 3D motion transfer from videos, 2026

    Taeyeon Kim, Youngju Na, Jumin Lee, Sebin Lee, Minhyuk Sung, and Sung-Eui Yoon. Mor- phGS: Morphology-adaptive articulated 3D motion transfer from videos, 2026

    2026

  19. [19]

    HUGS: Human Gaussian splats

    Muhammed Kocabas, Jen-Hao Rick Chang, James Gabriel, Oncel Tuzel, and Anurag Ranjan. HUGS: Human Gaussian splats. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

    2024

  20. [20]

    Pulsar: Efficient sphere-based neural rendering

    Christoph Lassner and Michael Zollhöfer. Pulsar: Efficient sphere-based neural rendering. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 1440–1449, 2021

    2021

  21. [21]

    OffsetOPT: Explicit surface reconstruction without normals

    Huan Lei. OffsetOPT: Explicit surface reconstruction without normals. InIEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR), 2025

    2025

  22. [22]

    Pose space deformation: A unified approach to shape interpolation and skeleton-driven deformation

    John P Lewis, Matt Cordner, and Nickson Fong. Pose space deformation: A unified approach to shape interpolation and skeleton-driven deformation. InProceedings of the 27th Annual Conference on Computer Graphics and Interactive Techniques, pages 165–172, 2000

    2000

  23. [23]

    Black, Hao Li, and Javier Romero

    Tianye Li, Timo Bolkart, Michael J. Black, Hao Li, and Javier Romero. Learning a model of facial shape and expression from 4D scans.ACM Transactions on Graphics, 36:1–17, 2017

    2017

  24. [24]

    Taylor, Mathias Unberath, Ming-Yu Liu, and Chen-Hsuan Lin

    Zhaoshuo Li, Thomas Müller, Alex Evans, Russell H. Taylor, Mathias Unberath, Ming-Yu Liu, and Chen-Hsuan Lin. Neuralangelo: High-fidelity neural surface reconstruction. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023

    2023

  25. [25]

    Animatable Gaussians: Learning pose- dependent Gaussian maps for high-fidelity human avatar modeling

    Zhe Li, Zerong Zheng, Lizhen Wang, and Yebin Liu. Animatable Gaussians: Learning pose- dependent Gaussian maps for high-fidelity human avatar modeling. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

    2024

  26. [26]

    DreamMesh4D: Video-to-4D generation with sparse- controlled gaussian-mesh hybrid representation

    Zhiqi Li, Yiming Chen, and Peidong Liu. DreamMesh4D: Video-to-4D generation with sparse- controlled gaussian-mesh hybrid representation. InAdvances in Neural Information Processing Systems (NeurIPS), 2024

    2024

  27. [27]

    Depth Anything 3: Recovering the Visual Space from Any Views

    Haotong Lin, Sili Chen, Junhao Liew, Donny Y . Chen, Zhenyu Li, Guang Shi, Jiashi Feng, and Bingyi Kang. Depth anything 3: Recovering the visual space from any views.arXiv preprint arXiv:2511.10647, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  28. [28]

    RigAnything: Template-free autoregressive rigging for diverse 3D assets.ACM Transactions on Graphics, 44(4):1–12, 2025

    Isabella Liu, Zhan Xu, Wang Yifan, Hao Tan, Zexiang Xu, Xiaolong Wang, Hao Su, and Zifan Shi. RigAnything: Template-free autoregressive rigging for diverse 3D assets.ACM Transactions on Graphics, 44(4):1–12, 2025

    2025

  29. [29]

    Matthew Loper, Naureen Mahmood, Javier Romero, Gerard Pons-Moll, and Michael J. Black. SMPL: A skinned multi-person linear model.ACM Trans. Graphics (Proc. SIGGRAPH Asia), 34(6):248:1–248:16, October 2015

    2015

  30. [30]

    Joint-dependent local deformations for hand animation and object grasping

    Nadia Magnenat-Thalmann, Richard Laperrière, and Daniel Thalmann. Joint-dependent local deformations for hand animation and object grasping. InProceedings on Graphics Interface’88, pages 26–33, 1989

    1989

  31. [31]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. NeRF: Representing scenes as neural radiance fields for view synthesis. In European Conference on Computer Vision (ECCV), 2020. 11

    2020

  32. [32]

    Barron, Sofien Bouaziz, Dan B Goldman, Steven M

    Keunhong Park, Utkarsh Sinha, Jonathan T. Barron, Sofien Bouaziz, Dan B Goldman, Steven M. Seitz, and Ricardo Martin-Brualla. Nerfies: Deformable neural radiance fields. InIEEE/CVF International Conference on Computer Vision (ICCV), pages 5865–5874, 2021

    2021

  33. [33]

    D-NeRF: Neural radiance fields for dynamic scenes

    Albert Pumarola, Enric Corona, Gerard Pons-Moll, and Francesc Moreno-Noguer. D-NeRF: Neural radiance fields for dynamic scenes. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10318–10327, 2021

    2021

  34. [34]

    GaussianAvatars: Photorealistic head avatars with rigged 3D Gaussians

    Shenhan Qian, Tobias Kirschstein, Liam Schoneveld, Davide Davoli, Simon Giebenhain, and Matthias Nießner. GaussianAvatars: Photorealistic head avatars with rigged 3D Gaussians. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

    2024

  35. [35]

    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. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5020–5030, 2024

    2024

  36. [36]

    Rivers and Doug L

    Alec R. Rivers and Doug L. James. FastLSM: Fast lattice shape matching for robust real-time deformation. InACM SIGGRAPH 2007 Papers, pages 82:1–82:6, 2007

    2007

  37. [37]

    ADOP: Approximate differentiable one-pixel point rendering.ACM Transactions on Graphics, 41(4):99:1–99:14, 2022

    Darius Rückert, Linus Franke, and Marc Stamminger. ADOP: Approximate differentiable one-pixel point rendering.ACM Transactions on Graphics, 41(4):99:1–99:14, 2022

    2022

  38. [38]

    Meshless simulation for skeleton driven elastic deformation.Journal of Zhejiang University-Science A, 7(9):1596–1602, 2006

    Chao Song, Hong-xin Zhang, Jin Huang, and Hu-jun Bao. Meshless simulation for skeleton driven elastic deformation.Journal of Zhejiang University-Science A, 7(9):1596–1602, 2006

    2006

  39. [39]

    Puppeteer: Rig and animate your 3D models

    Chaoyue Song, Xiu Li, Fan Yang, Zhongcong Xu, Jiacheng Wei, Fayao Liu, Jiashi Feng, Guosheng Lin, and Jianfeng Zhang. Puppeteer: Rig and animate your 3D models. InAdvances in Neural Information Processing Systems (NeurIPS), 2025

    2025

  40. [40]

    As-rigid-as-possible surface modeling

    Olga Sorkine-Hornung and Marc Alexa. As-rigid-as-possible surface modeling. InEurographics Symposium on Geometry Processing, volume 4, pages 109–116, 2007

    2007

  41. [41]

    Template-free articulated neural point clouds for reposable view synthesis

    Lukas Uzolas, Elmar Eisemann, and Petr Kellnhofer. Template-free articulated neural point clouds for reposable view synthesis. InAdvances in Neural Information Processing Systems (NeurIPS), volume 36, pages 31621–31637, 2023

    2023

  42. [42]

    GaMeS: Mesh-based adapting and modification of Gaussian splatting.arXiv preprint arXiv:2402.01459, 2024

    Joanna Waczy´nska, Piotr Borycki, Sławomir Tadeja, Jacek Tabor, and Przemysław Spurek. GaMeS: Mesh-based adapting and modification of Gaussian splatting.arXiv preprint arXiv:2402.01459, 2024

    work page arXiv 2024

  43. [43]

    Template-free articulated Gaussian splatting for real-time reposable dynamic view synthesis

    Diwen Wan, Yuxiang Wang, Ruijie Lu, and Gang Zeng. Template-free articulated Gaussian splatting for real-time reposable dynamic view synthesis. InAdvances in Neural Information Processing Systems (NeurIPS), volume 37, pages 62000–62023, 2024

    2024

  44. [44]

    NeuS: Learning neural implicit surfaces by volume rendering for multi-view reconstruction

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. NeuS: Learning neural implicit surfaces by volume rendering for multi-view reconstruction. In Advances in Neural Information Processing Systems (NeurIPS), 2021

    2021

  45. [45]

    Bovik, Hamid R

    Zhou Wang, Alan C. Bovik, Hamid R. Sheikh, and Eero P. Simoncelli. Image quality assessment: from error visibility to structural similarity.IEEE Transactions on Image Processing, 13(4):600– 612, 2004

    2004

  46. [46]

    4D Gaussian Splatting for real-time dynamic scene rendering

    Guanjun Wu, Taoran Yi, Jiemin Fang, Lingxi Xie, Xiaopeng Zhang, Wei Wei, Wenyu Liu, Qi Tian, and Xinggang Wang. 4D Gaussian Splatting for real-time dynamic scene rendering. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 20310– 20320, 2024

    2024

  47. [47]

    SC4D: Sparse- controlled video-to-4D generation and motion transfer

    Zijie Wu, Chaohui Yu, Yanqin Jiang, Chenjie Cao, Fan Wang, and Xiang Bai. SC4D: Sparse- controlled video-to-4D generation and motion transfer. InEuropean Conference on Computer Vision (ECCV), pages 361–379. Springer, 2024

    2024

  48. [48]

    AnimaMimic: Imitating 3D animation from video priors, 2025

    Tianyi Xie, Yunuo Chen, Yaowei Guo, Yin Yang, Bolei Zhou, Demetri Terzopoulos, Ying Jiang, and Chenfanfu Jiang. AnimaMimic: Imitating 3D animation from video priors, 2025. 12

    2025

  49. [49]

    Point-NeRF: Point-based neural radiance fields

    Qiangeng Xu, Zexiang Xu, Julien Philip, Sai Bi, Zhixin Shu, Kalyan Sunkavalli, and Ulrich Neumann. Point-NeRF: Point-based neural radiance fields. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5438–5448, 2022

    2022

  50. [50]

    RigNet: Neural rigging for articulated characters.ACM Transactions on Graphics, 39(4):58:1–58:14, 2020

    Zhan Xu, Yang Zhou, Evangelos Kalogerakis, Chris Landreth, and Karan Singh. RigNet: Neural rigging for articulated characters.ACM Transactions on Graphics, 39(4):58:1–58:14, 2020

    2020

  51. [51]

    MoRig: Motion-aware rigging of character meshes from point clouds

    Zhan Xu, Yang Zhou, Li Yi, and Evangelos Kalogerakis. MoRig: Motion-aware rigging of character meshes from point clouds. InSIGGRAPH Asia 2022 Conference Papers, 2022

    2022

  52. [52]

    Deformable 3D Gaussians for high-fidelity monocular dynamic scene reconstruction

    Ziyi Yang, Xinyu Gao, Wen Zhou, Shaohui Jiao, Yuqing Zhang, and Xiaogang Jin. Deformable 3D Gaussians for high-fidelity monocular dynamic scene reconstruction. InIEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages 20331–20341, 2024

    2024

  53. [53]

    RigGS: Rigging of 3D gaussians for modeling articulated objects in videos

    Yuxin Yao, Zhi Deng, and Junhui Hou. RigGS: Rigging of 3D gaussians for modeling articulated objects in videos. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5592–5601, 2025

    2025

  54. [54]

    A3-GS: Animate any articulated objects with 3D Gaussian splatting, 2026

    Yuxin Yao and Junhui Hou. A3-GS: Animate any articulated objects with 3D Gaussian splatting, 2026

    2026

  55. [55]

    PhysRig: Differen- tiable physics-based skinning and rigging framework for realistic articulated object modeling

    Hao Zhang, Haolan Xu, Chun Feng, Varun Jampani, and Narendra Ahuja. PhysRig: Differen- tiable physics-based skinning and rigging framework for realistic articulated object modeling. InIEEE/CVF International Conference on Computer Vision (ICCV), 2025

    2025

  56. [56]

    One model to rig them all: Diverse skeleton rigging with UniRig.ACM Transactions on Graphics, 44(4):1–18, 2025

    Jia-Peng Zhang, Cheng-Feng Pu, Meng-Hao Guo, Yan-Pei Cao, and Shi-Min Hu. One model to rig them all: Diverse skeleton rigging with UniRig.ACM Transactions on Graphics, 44(4):1–18, 2025

    2025

  57. [57]

    Efros, Eli Shechtman, and Oliver Wang

    Richard Zhang, Phillip Isola, Alexei A. Efros, Eli Shechtman, and Oliver Wang. The unrea- sonable effectiveness of deep features as a perceptual metric. InIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 586–595, 2018

    2018

  58. [58]

    Motion blender gaussian splatting for dynamic reconstruction

    Xinyu Zhang, Haonan Chang, Yuhan Liu, and Abdeslam Boularias. Motion blender gaussian splatting for dynamic reconstruction. InProceedings of the 9th Conference on Robot Learning (CoRL), volume 305 ofProceedings of Machine Learning Research, pages 5107–5128. PMLR, 2025

    2025

  59. [59]

    decays a→0

    Yanshu Zhang, Shichong Peng, Alireza Moazeni, and Ke Li. PAPR: Proximity attention point rendering. InAdvances in Neural Information Processing Systems (NeurIPS), 2023. 13 This appendix is organized into four parts. Section A extends the qualitative comparison of the main paper to the four remaining synthetic subjects (fox, Simpsons, wolf, spider) and to ...

    2023