pith. sign in

arxiv: 2607.01663 · v1 · pith:BPETKD4Vnew · submitted 2026-07-02 · 💻 cs.CV

Unified Panoramic-Gaussian Representation for Monocular 4D Scene Synthesis

Yuankun Yang , Yi Wei , Wenyang Zhou , Li Zhang This is my paper

Pith reviewed 2026-07-03 16:37 UTC · model grok-4.3

classification 💻 cs.CV
keywords 4D scene synthesismonocular videoGaussian representationpanoramic representationnovel view synthesisdynamic scenesunseen regionscamera trajectory
0
0 comments X

The pith

PanoGaussian distills panoramic representations into dynamic Gaussians to enable consistent 4D scene synthesis from monocular videos under large viewpoint variations.

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

The paper reformulates monocular 4D scene synthesis to include synthesis of unseen regions beyond observed views. Existing camera-conditioned generation methods produce inconsistencies when camera paths deviate far from training trajectories because they lack explicit 3D priors. A panoramic trajectory guidance framework improves cross-view consistency but introduces scale and shape distortions when objects move. PanoGaussian addresses this by distilling the panoramic output into an explicit dynamic Gaussian representation that encodes physical motion priors. Experiments show the resulting representation maintains geometric consistency across wide viewpoint changes.

Core claim

By distilling a panoramic representation into an explicit dynamic Gaussian representation, PanoGaussian captures the dynamic physical priors of a 4D scene and removes the scale and shape distortions that arise from object motion in panoramic space, thereby achieving consistent 4D scene synthesis even under large viewpoint variations.

What carries the argument

PanoGaussian, a unified Panoramic-Gaussian representation that distills panoramic trajectory guidance into explicit dynamic 3D Gaussians to encode object motion priors.

If this is right

  • The method extends 4D synthesis beyond view interpolation to regions never observed in the input video.
  • Panoramic guidance improves cross-view consistency compared with random camera trajectories in video generation models.
  • Explicit dynamic Gaussians encode physical motion priors that panoramic representations alone cannot maintain.
  • The unified training and inference framework supports both panoramic guidance and Gaussian distillation in one pipeline.

Where Pith is reading between the lines

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

  • The hybrid panoramic-to-Gaussian distillation may apply to other dynamic scene tasks where pure implicit or explicit representations fall short.
  • If the Gaussian stage can be made faster, the approach could support interactive 4D scene editing from casual videos.
  • The same distortion problem may appear in other panoramic or equirectangular representations of moving scenes, suggesting a broader need for explicit 3D distillation steps.

Load-bearing premise

Distilling the panoramic representation into dynamic Gaussians removes motion-induced scale and shape distortions while preserving physical consistency and without creating new inconsistencies under large trajectory changes.

What would settle it

Generate a 4D scene with a camera trajectory that deviates substantially from the input video path and observe whether object shapes and scales remain geometrically consistent across novel views; visible distortions would falsify the claim.

Figures

Figures reproduced from arXiv: 2607.01663 by Li Zhang, Wenyang Zhou, Yi Wei, Yuankun Yang.

Figure 1
Figure 1. Figure 1: PanoGaussian. (a) Given a monocular video, (b) PanoGaussian constructs a panoramic representation for scene exploration under a progressive strategy. (c) This panoramic representation is unified with Gaussian representation to achieve geometry￾preserved dynamic scene modeling. (d) The unified Panoramic–Gaussian representation enables geometrically consistent 4D scene synthesis across unseen regions. To add… view at source ↗
Figure 2
Figure 2. Figure 2: Architecture Overview. (a) PanoGaussian begins by reprojecting the monocular video into panoramic coordinates for full-scene exploration (Sec. 4.1). Un￾der this panorama, tangent directions are constructed to progressively guide trajectory expansion for unseen regions (Sec. 4.2). (b) PanoGaussian employs Gaussian-rendered depth to warp the original video frames along these expanded trajectories. The warped… view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative comparison of PanoGaussian on DyCheck [ [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison on Kubric-4D [71]. Under large camera displace￾ment, baselines fail to synthesize coherent unseen regions, while PanoGaussian pre￾serves consistent structure and appearance [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Scene exploration under extreme camera transitions. [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Panoramic design is vital for unseen-region consistency. [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative ablation study under extreme camera transitions. [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Results on outward-facing scenes. PanoGaussian maintains visual con￾sistency on nearby outdoor structures; distant regions degrade gracefully rather than collapsing. A Future Works To our knowledge, PanoGaussian is the first 4D reconstruction attempt that extends view interpolation to synthesis with unseen regions. While it achieves consistent and geometry-aware 4D synthesis, several directions remain for … view at source ↗
Figure 9
Figure 9. Figure 9: Full 4D scene synthesis with PanoGaussian. [PITH_FULL_IMAGE:figures/full_fig_p023_9.png] view at source ↗
read the original abstract

4D scene synthesis from monocular videos has made significant progress in recent years. However, existing methods are typically constrained by view interpolation. As a result, they struggle to infer unseen regions beyond the observed views. In this paper, we reformulate the task as 4D scene synthesis with unseen regions, which extends beyond traditional interpolation settings. Camera-conditioned video generation enables unseen region synthesis by guiding generation along specified cameras. However, these methods lack explicit 3D priors and are optimized with random camera trajectories. This design leads to severe inconsistencies under large trajectory deviations. To address this limitation, we build a unified training and inference framework with panoramic trajectory guidance. While this design improves cross-view consistency, the panoramic representation alone fails to model dynamic content effectively. Object motion in panoramic space introduces scale and shape distortions. To address this, we propose PanoGaussian, a unified Panoramic-Gaussian representation that distills the panoramic representation into an explicit dynamic Gaussian representation to capture dynamic physical priors of the 4D scene. Experiments demonstrate that PanoGaussian achieves consistent 4D scene synthesis even under large viewpoint variations.

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

1 major / 0 minor

Summary. The paper claims that monocular 4D scene synthesis is limited by view interpolation and that camera-conditioned generation with random trajectories produces inconsistencies under large deviations. It introduces panoramic trajectory guidance to improve cross-view consistency but notes that panoramic representations distort dynamic object motion. To fix this, the authors propose PanoGaussian, which distills the panoramic representation into an explicit dynamic Gaussian representation that captures dynamic physical priors, enabling consistent 4D synthesis even under large viewpoint variations.

Significance. If the distillation step reliably removes scale/shape distortions while preserving physical priors and avoiding new inconsistencies, the method would meaningfully extend 4D synthesis beyond interpolation. The abstract, however, contains no equations, loss formulations, ablation results, or quantitative metrics, so the practical significance cannot be assessed from the provided text.

major comments (1)
  1. The central claim rests on the assertion that distilling the panoramic representation into dynamic Gaussians removes motion-induced scale and shape distortions while capturing physical priors and avoiding new inconsistencies under large trajectories. No mechanism, loss function, constraint, or training objective is described to show how this distillation enforces the required properties. This assumption is load-bearing for the entire improvement over camera-conditioned baselines.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their review and for highlighting the need for greater clarity on the distillation mechanism. We address the major comment below.

read point-by-point responses

  1. Referee: The central claim rests on the assertion that distilling the panoramic representation into dynamic Gaussians removes motion-induced scale and shape distortions while capturing physical priors and avoiding new inconsistencies under large trajectories. No mechanism, loss function, constraint, or training objective is described to show how this distillation enforces the required properties. This assumption is load-bearing for the entire improvement over camera-conditioned baselines.

    Authors: We agree that the abstract and the provided summary text do not include the specific loss functions or training objectives for the distillation step. The full manuscript expands on the PanoGaussian representation in the methods, but the description of how the distillation explicitly enforces distortion removal and physical priors (via particular constraints or objectives) is insufficiently detailed. We will revise the manuscript to add the key loss terms, constraints, and training objectives in both the abstract and the methods section to make the mechanism explicit. revision: yes

Circularity Check

0 steps flagged

No circularity: method is an independent design choice

full rationale

The abstract and described framework present panoramic trajectory guidance followed by distillation to dynamic Gaussians as a sequence of engineering decisions to address stated limitations (inconsistencies under large trajectories, scale/shape distortions). No equation, loss term, or parameter is shown to be fitted on a subset and then renamed as a 'prediction' of a closely related quantity. No self-citation is invoked as a load-bearing uniqueness theorem or to smuggle an ansatz. The central claim rests on the empirical success of the proposed representation rather than reducing to its own inputs by construction. This is the expected non-finding for a methods paper whose derivation chain is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

Based on abstract only; no explicit free parameters, axioms, or invented entities beyond the named representation are detailed.

invented entities (1)
  • PanoGaussian no independent evidence
    purpose: Unified representation that distills panoramic views into dynamic Gaussians to capture 4D physical priors
    Introduced in the paper as the core technical contribution to address distortions in panoramic space.

pith-pipeline@v0.9.1-grok · 5731 in / 1078 out tokens · 19981 ms · 2026-07-03T16:37:37.566543+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

95 extracted references · 19 canonical work pages · 3 internal anchors

  1. [1]

    IEEE transactions on visualization and computer graphics (2024)

    Ai, H., Cao, Z., Lu, H., Chen, C., Ma, J., Zhou, P., Kim, T.K., Hui, P., Wang, L.: Dream360: Diverse and immersive outdoor virtual scene creation via transformer- based 360 image outpainting. IEEE transactions on visualization and computer graphics (2024)

    2024

  2. [2]

    arXiv preprint (2025)

    Bai,J., Xia, M., Fu, X.,Wang, X.,Mu, L., Cao, J.,Liu, Z., Hu, H.,Bai, X., Wan, P., et al.: Recammaster: Camera-controlled generative rendering from a single video. arXiv preprint (2025)

    2025

  3. [3]

    In: 3DV (2025)

    Bai, J., Huang, L., Guo, J., Gong, W., Li, Y., Guo, Y.: 360-gs: Layout-guided panoramic gaussian splatting for indoor roaming. In: 3DV (2025)

    2025

  4. [4]

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

    Blattmann, A., Dockhorn, T., Kulal, S., Mendelevitch, D., Kilian, M., Lorenz, D., Levi, Y., English, Z., Voleti, V., Letts, A., et al.: 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

  5. [5]

    In: SIGGRAPH (2025)

    Cao, C., Zhou, J., Li, S., Liang, J., Yu, C., Wang, F., Xue, X., Fu, Y.: Uni3c: Unify- ing precisely 3d-enhanced camera and human motion controls for video generation. In: SIGGRAPH (2025)

    2025

  6. [6]

    In: NeurIPS (2025)

    Chen, K., Khurana, T., Ramanan, D.: Reconstruct, inpaint, finetune: Dynamic novel-view synthesis from monocular videos. In: NeurIPS (2025)

    2025

  7. [7]

    In: CVPR (2025)

    Chen, Z., Wu, C., Shen, Z., Zhao, C., Ye, W., Feng, H., Ding, E., Zhang, S.H.: Splatter-360: Generalizable 360 gaussian splatting for wide-baseline panoramic im- ages. In: CVPR (2025)

    2025

  8. [8]

    In: NeurIPS (2024)

    Chu, W.H., Ke, L., Fragkiadaki, K.: Dreamscene4d: Dynamic multi-object scene generation from monocular videos. In: NeurIPS (2024)

    2024

  9. [9]

    In: SIGGRAPH (2024)

    Chugunov, I., Joshi, A., Murthy, K., Bleibel, F., Heide, F.: Neural light spheres for implicit image stitching and view synthesis. In: SIGGRAPH (2024)

    2024

  10. [10]

    arXiv preprint arXiv:2312.01196 (2023)

    Das, D., Wewer, C., Yunus, R., Ilg, E., Lenssen, J.E.: Neural parametric gaussians for monocular non-rigid object reconstruction. arXiv preprint arXiv:2312.01196 (2023)

    work page arXiv 2023

  11. [11]

    ArXiv (2024)

    Duan, Y., Wei, F., Dai, Q., He, Y., Chen, W., Chen, B.: 4d gaussian splatting: Towards efficient novel view synthesis for dynamic scenes. ArXiv (2024)

    2024

  12. [12]

    arXiv preprint arXiv:2312.01089 (2023)

    Duisterhof, B., Xu, D., Chen, R., Chen, Y.C.: Mvsplat: Efficient 3d gaussian splat- ting from multi-view stereo. arXiv preprint arXiv:2312.01089 (2023)

    work page arXiv 2023

  13. [13]

    In: SIGGRAPH (2022)

    Fang, J., Yi, T., Wang, X., Xie, L., Zhang, X., Liu, W., Nießner, M., Tian, Q.: Fast dynamic radiance fields with time-aware neural voxels. In: SIGGRAPH (2022)

    2022

  14. [14]

    In: CVPR (2023)

    Fridovich-Keil, S., Meanti, G., Warburg, F.R., Recht, B., Kanazawa, A.: K-planes: Explicit radiance fields in space, time, and appearance. In: CVPR (2023)

    2023

  15. [15]

    In: Proceedings of the IEEE International Conference on Com- puter Vision (2021)

    Gao, C., Saraf, A., Kopf, J., Huang, J.B.: Dynamic view synthesis from dynamic monocular video. In: Proceedings of the IEEE International Conference on Com- puter Vision (2021)

    2021

  16. [16]

    In: ICCV (2021)

    Gao, C., Saraf, A., Kopf, J., Huang, J.B.: Dynamic view synthesis from dynamic monocular video. In: ICCV (2021)

    2021

  17. [17]

    In: NeurIPS (2022)

    Gao, H., Li, R., Tulsiani, S., Russell, B., Kanazawa, A.: Monocular dynamic view synthesis: A reality check. In: NeurIPS (2022)

    2022

  18. [18]

    In: ICME (2022)

    Gu, K., Maugey, T., Knorr, S., Guillemot, C.: Omni-nerf: neural radiance field from 360 image captures. In: ICME (2022)

    2022

  19. [19]

    ICLR (2026) Unified Panoramic–Gaussian for 4D Synthesis [ECCV 2026] 17

    Guo, F., Hsu, C.C., Ding, S., Zhang, C.: Uncertainty matters in dynamic gaussian splatting for monocular 4d reconstruction. ICLR (2026) Unified Panoramic–Gaussian for 4D Synthesis [ECCV 2026] 17

    2026

  20. [20]

    ArXiv (2024)

    Guo, Z., Zhou, W.g., Li, L., Wang, M., Li, H.: Motion-aware 3d gaussian splatting for efficient dynamic scene reconstruction. ArXiv (2024)

    2024

  21. [21]

    arXiv preprint (2024)

    He, H., Xu, Y., Guo, Y., Wetzstein, G., Dai, B., Li, H., Yang, C.: Cameractrl: Enabling camera control for text-to-video generation. arXiv preprint (2024)

    2024

  22. [22]

    arXiv preprint (2025)

    He, H., Yang, C., Lin, S., Xu, Y., Wei, M., Gui, L., Zhao, Q., Wetzstein, G., Jiang, L., Li, H.: Cameractrl ii: Dynamic scene exploration via camera-controlled video diffusion models. arXiv preprint (2025)

    2025

  23. [23]

    ACM TOG (2017)

    Hedman, P., Alsisan, S., Szeliski, R., Kopf, J.: Casual 3d photography. ACM TOG (2017)

    2017

  24. [24]

    ACM TOG (2018)

    Hedman, P., Kopf, J.: Instant 3d photography. ACM TOG (2018)

    2018

  25. [25]

    NeurIPS (2020)

    Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models. NeurIPS (2020)

    2020

  26. [26]

    ICLR (2022)

    Hong, W., Ding, M., Zheng, W., Liu, X., Tang, J.: Cogvideo: Large-scale pretrain- ing for text-to-video generation via transformers. ICLR (2022)

    2022

  27. [27]

    360Roam: Real- Time Indoor Roaming Using Geometry-Aware 360° Radiance Fields,

    Huang, H., Chen, Y., Zhang, T., Yeung, S.K.: 360roam: Real-time indoor roaming using geometry-aware 360 radiance fields. arXiv preprint arXiv:2208.02705 (2022)

    work page arXiv 2022

  28. [28]

    In: ICCV (2025)

    Huang, J., Miao, S., Yang, B., Ma, Y., Liao, Y.: Vivid4d: Improving 4d reconstruc- tion from monocular video by video inpainting. In: ICCV (2025)

    2025

  29. [29]

    ArXiv (2023)

    Katsumata, K., Vo, D.M., Nakayama, H.: An efficient 3d gaussian representation for monocular/multi-view dynamic scenes. ArXiv (2023)

    2023

  30. [30]

    ACM Transactions on Graphics (2023)

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G.: 3d gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics (2023)

    2023

  31. [31]

    NeurIPS (2024)

    Lee, S., Chung, J., Huh, J., Lee, K.M.: Odgs: 3d scene reconstruction from omni- directional images with 3d gaussian splattings. NeurIPS (2024)

    2024

  32. [32]

    In: CVPR (2025)

    Lei, J., Weng, Y., Harley, A.W., Guibas, L., Daniilidis, K.: Mosca: Dynamic gaus- sian fusion from casual videos via 4d motion scaffolds. In: CVPR (2025)

    2025

  33. [33]

    ArXiv (2024)

    Li, F., Zhang, H., Ahuja, N.: Self-calibrating 4d novel view synthesis from monoc- ular videos using gaussian splatting. ArXiv (2024)

    2024

  34. [34]

    In: WACV (2025)

    Li, L., Huang, H., Yeung, S.K., Cheng, H.: Omnigs: Fast radiance field reconstruc- tion using omnidirectional gaussian splatting. In: WACV (2025)

    2025

  35. [35]

    arXiv preprint arXiv:2406.13527 (2024)

    Li, R., Pan, P., Yang, B., Xu, D., Zhou, S., Zhang, X., Li, Z., Kadambi, A., Wang, Z., Tu, Z., et al.: 4k4dgen: Panoramic 4d generation at 4k resolution. arXiv preprint arXiv:2406.13527 (2024)

    work page arXiv 2024

  36. [36]

    arXiv preprint arXiv:2312.16812 (2023)

    Li, Z., Chen, Z., Li, Z., Xu, Y.: Spacetime gaussian feature splatting for real-time dynamic view synthesis. arXiv preprint arXiv:2312.16812 (2023)

    work page arXiv 2023

  37. [37]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2021)

    Li, Z., Niklaus, S., Snavely, N., Wang, O.: Neural scene flow fields for space-time view synthesis of dynamic scenes. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2021)

    2021

  38. [38]

    In: CVPR (2021)

    Li, Z., Niklaus, S., Snavely, N., Wang, O.: Neural scene flow fields for space-time view synthesis of dynamic scenes. In: CVPR (2021)

    2021

  39. [39]

    Li, Z., Wang, Q., Cole, F., Tucker, R., Snavely, N.: Dynibar: Neural dynamic image- basedrendering.In:ProceedingsoftheIEEE/CVFConferenceonComputerVision and Pattern Recognition (2023)

    2023

  40. [40]

    arXiv preprint (2024)

    Liang, H., Ren, J., Mirzaei, A., Torralba, A., Liu, Z., Gilitschenski, I., Fidler, S., Oztireli, C., Ling, H., Gojcic, Z., et al.: Feed-forward bullet-time reconstruction of dynamic scenes from monocular videos. arXiv preprint (2024)

    2024

  41. [41]

    ArXiv (2023) 18 Y

    Liang, Y., Khan, N., Li, Z., Nguyen-Phuoc, T., Lanman, D., Tompkin, J., Xiao, L.: Gaufre: Gaussian deformation fields for real-time dynamic novel view synthesis. ArXiv (2023) 18 Y. Yang et al

    2023

  42. [42]

    arXiv preprint arXiv:2507.10065 (2025)

    Lin, C., Lin, Y., Pan, P., Yu, Y., Yan, H., Fragkiadaki, K., Mu, Y.: Movies: Motion- aware 4d dynamic view synthesis in one second. arXiv preprint arXiv:2507.10065 (2025)

    work page arXiv 2025

  43. [43]

    arXiv preprint arXiv:2506.09997 (2025)

    Lin, C.H., Lv, Z., Wu, S., Xu, Z., Nguyen-Phuoc, T., Tseng, H.Y., Straub, J., Khan, N., Xiao, L., Yang, M.H., et al.: Dgs-lrm: Real-time deformable 3d gaussian reconstruction from monocular videos. arXiv preprint arXiv:2506.09997 (2025)

    work page arXiv 2025

  44. [44]

    arXiv:2312.03431 (2023)

    Lin, Y., Dai, Z., Zhu, S., Yao, Y.: Gaussian-flow: 4d reconstruction with dynamic 3d gaussian particle. arXiv:2312.03431 (2023)

    work page arXiv 2023

  45. [45]

    In: CVPR (2025)

    Liu, J., Lin, S., Li, Y., Yang, M.H.: Dynamicscaler: Seamless and scalable video generation for panoramic scenes. In: CVPR (2025)

    2025

  46. [46]

    In: AAAI (2024)

    Lu, Z., Zheng, Q., Shi, B., Jiang, X.: Pano-nerf: Synthesizing high dynamic range novel views with geometry from sparse low dynamic range panoramic images. In: AAAI (2024)

    2024

  47. [47]

    In: AAAI (2024)

    Lu,Z.,Hu,K.,Wang,C.,Bai,L.,Wang,Z.:Autoregressiveomni-awareoutpainting for open-vocabulary 360-degree image generation. In: AAAI (2024)

    2024

  48. [48]

    In: 3DV (2024)

    Luiten, J., Kopanas, G., Leibe, B., Ramanan, D.: Dynamic 3d gaussians: Tracking by persistent dynamic view synthesis. In: 3DV (2024)

    2024

  49. [49]

    arXiv preprint arXiv:2506.03140 (2025)

    Luo, Y., Bai, J., Shi, X., Xia, M., Wang, X., Wan, P., Zhang, D., Gai, K., Xue, T.: Camclonemaster: Enabling reference-based camera control for video generation. arXiv preprint arXiv:2506.03140 (2025)

    work page arXiv 2025

  50. [50]

    Luo,Z.,Ran,H.,Lu,L.:Instant4d:4dgaussiansplattinginminutes.arXivpreprint arXiv:2510.01119 (2025)

    work page arXiv 2025

  51. [51]

    PR (2024)

    Miao, X., Bai, Y., Duan, H., Wan, F., Huang, Y., Long, Y., Zheng, Y.: Ctnerf: Cross-time transformer for dynamic neural radiance field from monocular video. PR (2024)

    2024

  52. [52]

    OpenVid-1M: A Large-Scale High-Quality Dataset for Text-to-video Generation

    Nan, K., Xie, R., Zhou, P., Fan, T., Yang, Z., Chen, Z., Li, X., Yang, J., Tai, Y.: Openvid-1m: A large-scale high-quality dataset for text-to-video generation. arXiv preprint arXiv:2407.02371 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  53. [53]

    In: ICML (2021)

    Nichol, A.Q., Dhariwal, P.: Improved denoising diffusion probabilistic models. In: ICML (2021)

    2021

  54. [54]

    In: ICCV (2021)

    Park, K., Sinha, U., Barron, J.T., Bouaziz, S., Goldman, D.B., Seitz, S.M., Martin- Brualla, R.: Nerfies: Deformable neural radiance fields. In: ICCV (2021)

    2021

  55. [55]

    ACM TOG (2021)

    Park,K.,Sinha,U.,Hedman,P.,Barron,J.T.,Bouaziz,S.,Goldman,D.B.,Martin- Brualla, R., Seitz, S.M.: Hypernerf: a higher-dimensional representation for topo- logically varying neural radiance fields. ACM TOG (2021)

    2021

  56. [57]

    In: CVPR (2021)

    Pumarola, A., Corona, E., Pons-Moll, G., Moreno-Noguer, F.: D-nerf: Neural ra- diance fields for dynamic scenes. In: CVPR (2021)

    2021

  57. [58]

    In: AAAI (2024)

    Ramasinghe, S., Shevchenko, V., Avraham, G., van den Hengel, A.: Blirf: Band limited radiance fields for dynamic scene modeling. In: AAAI (2024)

    2024

  58. [59]

    In: CVPR (2025)

    Ren,X.,Shen,T.,Huang,J.,Ling,H.,Lu,Y.,Nimier-David,M.,Müller,T.,Keller, A., Fidler, S., Gao, J.: Gen3c: 3d-informed world-consistent video generation with precise camera control. In: CVPR (2025)

    2025

  59. [60]

    In: CVPR (2022)

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: CVPR (2022)

    2022

  60. [61]

    BulletGen: Improving 4D Reconstruction with Bullet-Time Generation

    Rozumnyi, D., Luiten, J., Khan, N., Schonberger, J., Kontschieder, P.: Bullet- gen: Improving 4d reconstruction with bullet-time generation. arXiv preprint arXiv:2506.18601 (2025) Unified Panoramic–Gaussian for 4D Synthesis [ECCV 2026] 19

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  61. [62]

    In: Proceedings of the 25th annual conference on Computer graphics and interactive techniques (1998)

    Shade, J., Gortler, S., He, L.w., Szeliski, R.: Layered depth images. In: Proceedings of the 25th annual conference on Computer graphics and interactive techniques (1998)

    1998

  62. [63]

    arXiv preprint arXiv:2502.03465 (2025)

    Shen, Q., Yi, X., Lin, M., Zhang, H., Yan, S., Wang, X.: Seeing world dynamics in a nutshell. arXiv preprint arXiv:2502.03465 (2025)

    work page arXiv 2025

  63. [64]

    In: Proceedings of the 26th annual conference on Computer graphics and interactive techniques (1999)

    Shum, H.Y., He, L.W.: Rendering with concentric mosaics. In: Proceedings of the 26th annual conference on Computer graphics and interactive techniques (1999)

    1999

  64. [65]

    IEEE Transactions on Visualization and Computer Graphics (2023)

    Song, L., Chen, A., Li, Z., Chen, Z., Chen, L., Yuan, J., Xu, Y., Geiger, A.: Nerf- player: A streamable dynamic scene representation with decomposed neural radi- ance fields. IEEE Transactions on Visualization and Computer Graphics (2023)

    2023

  65. [66]

    In: SIGGRAPH (2024)

    Stearns, C., Harley, A., Uy, M., Dubost, F., Tombari, F., Wetzstein, G., Guibas, L.: Dynamic gaussian marbles for novel view synthesis of casual monocular videos. In: SIGGRAPH (2024)

    2024

  66. [67]

    arXiv preprint arXiv:2403.01444 (2024)

    Sun, J., Jiao, H., Li, G., Zhang, Z., Zhao, L., Xing, W.: 3dgstream: On-the-fly training of 3d gaussians for efficient streaming of photo-realistic free-viewpoint videos. arXiv preprint arXiv:2403.01444 (2024)

    work page arXiv 2024

  67. [68]

    In: ICCV (2023)

    Tian, F., Du, S., Duan, Y.: Mononerf: Learning a generalizable dynamic radiance field from monocular videos. In: ICCV (2023)

    2023

  68. [69]

    In: ICCV (2021)

    Tretschk, E., Tewari, A., Golyanik, V., Zollhöfer, M., Lassner, C., Theobalt, C.: Non-rigid neural radiance fields: Reconstruction and novel view synthesis of a dy- namic scene from monocular video. In: ICCV (2021)

    2021

  69. [71]

    In: ECCV (2024)

    Van Hoorick, B., Wu, R., Ozguroglu, E., Sargent, K., Liu, R., Tokmakov, P., Dave, A., Zheng, C., Vondrick, C.: Generative camera dolly: Extreme monocular dynamic novel view synthesis. In: ECCV (2024)

    2024

  70. [72]

    In: ArXiv Preprint (2021)

    Wang, C., Eckart, B., Lucey, S., Gallo, O.: Neural trajectory fields for dynamic novel view synthesis. In: ArXiv Preprint (2021)

    2021

  71. [73]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    Wang,G.,Wang,P.,Chen,Z.,Wang,W.,Loy,C.C.,Liu,Z.:Perf:Panoramicneural radiance field from a single panorama. IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    2024

  72. [74]

    In: CVPR (2024)

    Wang, Q., Li, W., Mou, C., Cheng, X., Zhang, J.: 360dvd: Controllable panorama video generation with 360-degree video diffusion model. In: CVPR (2024)

    2024

  73. [75]

    In: ICCV (2025)

    Wang, Q., Ye, V., Gao, H., Zeng, W., Austin, J., Li, Z., Kanazawa, A.: Shape of motion: 4d reconstruction from a single video. In: ICCV (2025)

    2025

  74. [76]

    In: AAAI (2025)

    Wang, S., Yang, X., Shen, Q., Jiang, Z., Wang, X.: Gflow: Recovering 4d world from monocular video. In: AAAI (2025)

    2025

  75. [77]

    In: CVPR (2025)

    Wu, D., Liu, F., Hung, Y.H., Qian, Y., Zhan, X., Duan, Y.: 4d-fly: Fast 4d recon- struction from a single monocular video. In: CVPR (2025)

    2025

  76. [78]

    arXiv preprint arXiv:2310.08528 (2023)

    Wu, G., Yi, T., Fang, J., Xie, L., Zhang, X., Wei, W., Liu, W., Tian, Q., Wang, X.: 4d gaussian splatting for real-time dynamic scene rendering. arXiv preprint arXiv:2310.08528 (2023)

    work page arXiv 2023

  77. [79]

    In: CVPR (2024)

    Wu, G., Yi, T., Fang, J., Xie, L., Zhang, X., Wei, W., Liu, W., Tian, Q., Wang, X.: 4d gaussian splatting for real-time dynamic scene rendering. In: CVPR (2024)

    2024

  78. [80]

    arXiv preprint (2024)

    Wu, R., Gao, R., Poole, B., Trevithick, A., Zheng, C., Barron, J.T., Holynski, A.: Cat4d: Create anything in 4d with multi-view video diffusion models. arXiv preprint (2024)

    2024

  79. [81]

    arXiv preprint arXiv:2307.03177 (2023) 20 Y

    Wu, T., Zheng, C., Cham, T.J.: Panodiffusion: 360-degree panorama outpainting via diffusion. arXiv preprint arXiv:2307.03177 (2023) 20 Y. Yang et al

    work page arXiv 2023

  80. [82]

    arXiv preprint (2025)

    Wu, T., Yang, S., Po, R., Xu, Y., Liu, Z., Lin, D., Wetzstein, G.: Video world models with long-term spatial memory. arXiv preprint (2025)

    2025

Showing first 80 references.