Pith. sign in

REVIEW 3 major objections 6 minor 78 references

PanoSplatt3R: Leveraging Perspective Pretraining for Generalized Unposed Wide-Baseline Panorama Reconstruction

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read PanoSplatt3R claims that a single feed-forward network can reconstruct a 3D scene from two unposed panorama images and render novel views at quality above pose-dependent baselines.

desk verdict A well-ablated unposed panorama method whose headline NVS win depends on a ground-truth scale factor at test time — worth reviewing, but the SOTA claim needs rework. read the letter →

arxiv 2507.21960 v1 pith:Z4BIW5K5 submitted 2025-07-29 cs.CV

classification cs.CV
keywords panoramareconstructionunposed3DnovelviewsynthesisGaussianSplattingrotarypositionalembeddingRoPErollingdomaintransferwide-baselinestereo
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to show that a general-purpose 3D reconstruction model pretrained on ordinary perspective photos can be carried over to 360-degree panoramic images with only a small positional-embedding change, and that the resulting system reconstructs surrounding scenes and renders new views from two panoramas without being given camera poses. The authors report that PanoSplatt3R, after this transfer, beats current pose-dependent panorama reconstruction methods on both image quality and estimated depth: on HM3D, PSNR 28.938 and depth AbsRel 0.044 versus 28.308 and 0.078 for Splatter-360. If true, the practical consequence is that wide-baseline panorama reconstruction no longer needs pose estimation or calibration as a prerequisite, which matters for VR, real estate, robotics, and anywhere panoramas are captured casually.

What carries the argument

The load-bearing mechanism is RoPE rolling, a minimal modification of rotary positional embeddings for 2D vision transformers: for attention head m, the horizontal coordinate is shifted by W·m/M modulo W before computing the rotation matrix, so different heads encode different placements of the seam. This preserves RoPE's relative-position structure, so pretrained weights transfer, while giving the network a way to represent that the left and right edges of a panorama are adjacent. A second mechanism is the two-stage progressive training: first regress Gaussian center locations with L1 and L2 losses against projected depth, then train the full 2D Gaussian Splatting parameters with cubemap-rendered image and depth losses.

What would settle it

Re-run the Table 1 comparison on a fixed, publicly released evaluation set using the actual test splits from Splatter-360 or PanSplat, with the pose-dependent baselines given ground-truth poses; if Splatter-360's PSNR equals or exceeds 28.938 on HM3D or its AbsRel drops to 0.044 on the same split, the claim that PanoSplatt3R outperforms posed methods without pose input is not supported.

Watch

Extended reading notes

Core claim

The central claim is that the unposed setting—no camera poses as input—is not a handicap if the model starts from a strong perspective-domain pretrained stereo foundation and is adapted with a periodicity-aware positional encoding. In the paper's experiments, PanoSplatt3R outperforms pose-dependent baselines (PanoGRF, MVSplat, Splatter-360) in novel-view PSNR, SSIM, and LPIPS and in depth metrics on both HM3D and Replica, and when those baselines are fed poses estimated by SIFT plus the eight-point algorithm, their quality collapses while PanoSplatt3R is unchanged. The authors attribute the gain to transferring Mast3R and Dust3R knowledge and to RoPE rolling, which lets attention heads see different horizontal shifts of the equirectangular image instead of treating the left-right seam as a maximally distant boundary.

Load-bearing premise

The headline comparison rests on the authors' own re-rendered HM3D test set, generated by random simulated indoor trajectories and not released; if those random views are easier than the views used for the published baseline numbers, the state-of-the-art scores could be an artifact of test-set choice rather than of the method.

Editorial extensions

If this is right

  • Pose-free panorama reconstruction becomes practical: an unposed pair of wide-baseline panoramas can be turned into 3D geometry and photorealistic novel views in a single forward pass.
  • Perspective pretraining is enough as a source of generalization; no teacher-student projection pipeline or specialized panoramic cost volume is required.
  • Pose-dependent methods are brittle when poses are estimated: feeding them SIFT-plus-eight-point poses degrades reconstruction severely, while an unposed model is unaffected by pose error.
  • Modeling horizontal periodicity with RoPE rolling fixes seam discontinuities, as shown by the wall-continuity ablation and by the HM3D PSNR gap between the full model (28.938) and the no-rolling variant (27.916).
  • The same backbone also extrapolates beyond the training interval, retaining better depth and image metrics when novel frames are sampled outside the input frame range.

Reading between the lines

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

  • A testable extension is to apply the same RoPE-rolling recipe to other equirectangular-input transformers for depth estimation or segmentation, where horizontal wrap-around also matters; the authors only demonstrate reconstruction.
  • The method still recovers metric scale at test time by solving PnP and rescaling, so 'unposed' here means relative pose is learned while scale is handled externally; a fully self-supervised metric-scale variant would be a natural next step.
  • If randomly generated test trajectories turn out to produce easier views than the published baseline splits, the size of the advantage over Splatter-360 could shrink when all methods are evaluated on one fixed set; this is a test-set risk rather than a flaw in the mechanism itself.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper introduces PanoSplatt3R, an unposed two-view wide-baseline panorama reconstruction method that builds on the Dust3R/Mast3R architecture and predicts per-pixel 3D Gaussians without camera pose input at inference. The main technical contributions are a modified rotary positional embedding called RoPE rolling that models the horizontal periodicity of equirectangular images, a progressive training strategy, and 2DGS-based rendering. The method is evaluated on HM3D and Replica, and the authors report state-of-the-art novel-view synthesis and depth metrics compared with pose-dependent baselines such as Splatter-360 and PanoGRF.

Significance. If the claims hold, PanoSplatt3R would be a meaningful step toward practical wide-baseline panorama reconstruction without explicit pose estimation, and the RoPE-rolling idea could transfer to other panoramic transformer architectures. The paper's strengths include a clean adaptation of perspective pretraining, a useful multi-head positional-offset mechanism, thorough ablations (Tables 5) that support the contributions of RoPE rolling, pretrained weights, progressive training, and 2DGS, and strong unposed depth results (AbsRel 0.044 vs 0.078 on HM3D in Table 1) that do not depend on the test-time scaling step. However, the headline novel-view synthesis superiority is currently supported only through a test-time scale alignment that uses ground-truth translation magnitude, and the cross-method comparison relies on unpublished re-rendered test sets. The significance is therefore conditional on these issues being resolved.

major comments (3)
  1. [Section 4.1 (Scale Recovery) and Table 5]
  2. [Section 4.1 (Datasets) and Table 1]
  3. [Section 3.5 and Table 1]
minor comments (6)
  1. [Equation (1)]
  2. [Equation (2)]
  3. [Figure 2]
  4. [Section 3.5]
  5. [Table 4]
  6. [Section 4.4.1]

Circularity Check

0 steps flagged · score 2.0 of 10

No derivation-level circularity found; the paper is an empirical, externally evaluated feed-forward model, with only a minor same-group citation and a disclosed oracle-scale caveat that affects the headline NVS comparison but does not make the result circular.

full rationale

PanoSplatt3R's pipeline is a learned feed-forward model: Gaussian parameters are predicted from image pairs and supervised with external ground truth through Eqs. (3) and (4). There is no step where a predicted quantity is defined in terms of the claimed output, and no fitted parameter is renamed as a prediction. The perspective pretraining from Mast3R/Dust3R is external and independently validated by ablation Cfg.c in Table 5, and the RoPE-rolling variant is an internal design comparison rather than a circular validation. Reference [63] is a same-group citation used for the general idea of pose-free transformer reconstruction, but the actual pose-free mechanism is attributed to Dust3R/Mast3R and is not load-bearing. The main caveat is Section 4.1's 'Scale Recovery': the reported Table 1 numbers apply a ground-truth scale factor lambda = |t|/|t_hat| before computing NVS metrics. Table 5 shows that, without this scale, 'Full PanoSplatt3R' scores 28.200 PSNR on HM3D versus Splatter-360's 28.308 and 28.181 versus 29.888 on Replica, so the headline NVS superiority is not purely unposed. This is an experimental-validity risk, not circularity: the scale is a disclosed global normalization, not a modeled prediction, and the depth claims remain competitive without it. Similarly, the note in Section 4.1 that re-rendered HM3D metrics are 'highly consistent' with previous works without releasing the data is a comparability/reproducibility concern, not a circular argument. Overall, the derivation chain is self-contained; the score reflects the minor self-citation and the disclosed oracle-scale caveat, not a circular derivation.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim rests on a transfer-learning assumption about perspective pretraining, a dataset-comparability assumption for the re-rendered evaluation split, and a supervision-coverage assumption for cubemap rendering. The only explicit test-time fitted quantity is the scale factor computed from ground-truth translation magnitude.

free parameters (2)
  • Test-time scale factor lambda = computed as |t|/|t_hat| from ground-truth translation
    Used in evaluation to align the reconstruction's scale to metric units before computing PSNR/SSIM/LPIPS; without it, reported PSNR on HM3D drops from 28.938 to 28.200 and on Replica from 31.522 to 28.181 (Table 5). This is a ground-truth-derived quantity, not predicted by the model.
  • Stage-2 loss weights = MSE 1.0, depth L2 0.08, LPIPS 0.05
    Hand-chosen weights in Eq. (4) that balance image and depth supervision; they affect the trained model's quality but are conventional hyperparameters rather than fitted parameters.
assumptions (3)
  • domain assumption Mast3R/Dust3R perspective-domain pretrained weights transfer to panoramic images under the RoPE-rolling modification.
    Invoked in Sections 3.5 and 4.2 where the model is initialized with Mast3R weights and fine-tuned on panoramic data; the ablation (Cfg.c vs Full) supports the transfer benefit empirically.
  • domain assumption The re-rendered HM3D videos with random AI-Habitat trajectories are statistically similar to the Splatter-360 dataset splits.
    Invoked in Section 4.1, 'Datasets', where metrics are claimed to be 'highly consistent' with previous works without quantitative verification.
  • domain assumption Cubemap projection with six 90 degree FOV renders provides adequate supervision for the entire panoramic scene in stage 2.
    Invoked in Section 3.5, Eq. (4), where the loss is evaluated on six cube faces at a novel view; this assumes the six faces cover the full sphere with no significant gaps.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PanoSplatt3R: Leveraging Perspective Pretraining for Generalized Unposed Wide-Baseline Panorama Reconstruction." pith.science (2026). https://pith.science/paper/Z4BIW5K5

@misc{pith2026250721960,
  author       = {Pith},
  title        = {Pith review of: PanoSplatt3R: Leveraging Perspective Pretraining for Generalized Unposed Wide-Baseline Panorama Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z4BIW5K5}},
  note         = {Machine review of arXiv:2507.21960}
}
read the original abstract

Wide-baseline panorama reconstruction has emerged as a highly effective and pivotal approach for not only achieving geometric reconstruction of the surrounding 3D environment, but also generating highly realistic and immersive novel views. Although existing methods have shown remarkable performance across various benchmarks, they are predominantly reliant on accurate pose information. In real-world scenarios, the acquisition of precise pose often requires additional computational resources and is highly susceptible to noise. These limitations hinder the broad applicability and practicality of such methods. In this paper, we present PanoSplatt3R, an unposed wide-baseline panorama reconstruction method. We extend and adapt the foundational reconstruction pretrainings from the perspective domain to the panoramic domain, thus enabling powerful generalization capabilities. To ensure a seamless and efficient domain-transfer process, we introduce RoPE rolling that spans rolled coordinates in rotary positional embeddings across different attention heads, maintaining a minimal modification to RoPE's mechanism, while modeling the horizontal periodicity of panorama images. Comprehensive experiments demonstrate that PanoSplatt3R, even in the absence of pose information, significantly outperforms current state-of-the-art methods. This superiority is evident in both the generation of high-quality novel views and the accuracy of depth estimation, thereby showcasing its great potential for practical applications. Project page: https://npucvr.github.io/PanoSplatt3R

Figures

Figures reproduced from arXiv: 2507.21960 by the authors.

Figure 1
Figure 1. Overview of our proposed PanoSplatt3R. Given two panorama images as input and without knowing their relative pose, our model reconstructs the entire 3D scene using Gaussian Splats. The estimated geometry can provide photorealistic renderings at novel view points. We removed two walls from the final reconstruction to better display the room. producing feature representations that are more compatible with standard net… view at source ↗
Figure 2
Figure 2. Visual illustration of RoPE rolling. Multiple attention heads mark the location of A, B and C using different ‘RoPE’s to avoid always distancing A and C in positional embeddings. \mathbf {R}(n, 2t) = e^{i\theta _tp_n^x}, \quad \mathbf {R}(n, 2t+1) = e^{i\theta _tp_n^y}. (1) In vision transformers, the rotation frequencies are typi￾cally set as θt = 100−t/(dhead/4), for an attention head with dhead channels. Using pi… view at source ↗
Figure 3
Figure 3. Visual Comparison on reconstructed image (above) and depth (below). Each row comes from different pairs of input. comparisons are not entirely fair. We add Splatter-360 [10], who is trained on the same data as ours for a fair compari￾son. Our model outperforms Splatter-360 by a large margin, showcasing its strong generalization ability. Despite not being trained on Matterport3D [46] and be￾ing designed to handle mor… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Model without RoPE Rolling estimates broken geom [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Visual comparisons between PanoGRF, MVSplat, Splatter-360 and PanoSplatt3R(ours) on the HM3D dataset. [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: Visual comparisons between PanoGRF, MVSplat, Splatter-360 and PanoSplatt3R(ours) on the Replica dataset. [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

78 extracted references · 66 canonical work pages

  1. [1]

    Elite360d: Towards efficient 360 depth estimation via semantic-and distance-aware bi- projection fusion

    Hao Ai and Lin Wang. Elite360d: Towards efficient 360 depth estimation via semantic-and distance-aware bi- projection fusion. In IEEE Conf. Comput. Vis. Pattern Recog., pages 9926–9935, 2024. 3

  2. [2]

    Hrdfuse: Monocular 360deg depth estimation by collaboratively learning holistic-with-regional depth distri- butions

    Hao Ai, Zidong Cao, Yan-Pei Cao, Ying Shan, and Lin Wang. Hrdfuse: Monocular 360deg depth estimation by collaboratively learning holistic-with-regional depth distri- butions. In IEEE Conf. Comput. Vis. Pattern Recog., pages 13273–13282, 2023. 2

  3. [3]

    Virtual reality and 360 panorama technology: a media comparison to study changes in sense of presence, anxiety, and positive emotions

    Eleonora Brivio, Silvia Serino, Erica Negro Cousa, Andrea Zini, Giuseppe Riva, and Gianluca De Leo. Virtual reality and 360 panorama technology: a media comparison to study changes in sense of presence, anxiety, and positive emotions. Virtual Reality, 25:303–311, 2021. 1

  4. [4]

    pixelsplat: 3d gaussian splats from image pairs for scalable generalizable 3d reconstruction

    David Charatan, Sizhe Lester Li, Andrea Tagliasacchi, and Vincent Sitzmann. pixelsplat: 3d gaussian splats from image pairs for scalable generalizable 3d reconstruction. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 19457–19467,

  5. [5]

    VI3DRM:Towards meticulous 3D Reconstruction from Sparse Views via Photo-Realistic Novel View Synthesis

    Hao Chen, Jiafu Wu, Ying Jin, Jinlong Peng, Xiaofeng Mao, Mingmin Chi, Mufeng Yao, Bo Peng, Jian Li, and Yun Cao. Vi3drm: Towards meticulous 3d reconstruction from sparse views via photo-realistic novel view synthesis. arXiv preprint arXiv:2409.08207, 2024. 2

  6. [6]

    ZeroGS: Training 3D Gaussian Splatting from Unposed Images

    Yu Chen, Rolandos Alexandros Potamias, Evangelos Ververas, Jifei Song, Jiankang Deng, and Gim Hee Lee. Ze- rogs: Training 3d gaussian splatting from unposed images. arXiv preprint arXiv:2411.15779, 2024. 2

  7. [7]

    Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images

    Yuedong Chen, Haofei Xu, Chuanxia Zheng, Bohan Zhuang, Marc Pollefeys, Andreas Geiger, Tat-Jen Cham, and Jianfei Cai. Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images. In Eur. Conf. Comput. Vis., pages 370–

  8. [8]

    Panogrf: generalizable spherical radiance fields for wide-baseline panoramas

    Zheng Chen, Yan-Pei Cao, Yuan-Chen Guo, Chen Wang, Ying Shan, and Song-Hai Zhang. Panogrf: generalizable spherical radiance fields for wide-baseline panoramas. In Adv. Neural Inform. Process. Syst., pages 6961–6985, 2023. 1, 3, 4, 5, 6, 7, 13, 14

Show all 78 references
  1. [9]

    Pref3r: Pose- free feed-forward 3d gaussian splatting from variable-length image sequence

    Zequn Chen, Jiezhi Yang, and Heng Yang. Pref3r: Pose- free feed-forward 3d gaussian splatting from variable-length image sequence. arXiv preprint arXiv:2411.16877, 2024. 2

  2. [10]

    Splatter-360: Generalizable 360 gaussian splatting for wide- baseline panoramic images

    Zheng Chen, Chenming Wu, Zhelun Shen, Chen Zhao, We- icai Ye, Haocheng Feng, Errui Ding, and Song-Hai Zhang. Splatter-360: Generalizable 360 gaussian splatting for wide- baseline panoramic images. In IEEE Conf. Comput. Vis. Pat- tern Recog., pages 21590–21599, 2025. 1, 3, 4, 5...

  3. [11]

    Spherenet: Learning spherical representations for detection and classification in omnidirectional images

    Benjamin Coors, Alexandru Paul Condurache, and Andreas Geiger. Spherenet: Learning spherical representations for detection and classification in omnidirectional images. In Eur. Conf. Comput. Vis., pages 518–533, 2018. 2

  4. [12]

    Eliminating the blind spot: Adapting 3d object detection and monocular depth estimation to 360 panoramic imagery

    Greire Payen de La Garanderie, Amir Atapour Abarghouei, and Toby P Breckon. Eliminating the blind spot: Adapting 3d object detection and monocular depth estimation to 360 panoramic imagery. In Eur. Conf. Comput. Vis., pages 789– 807, 2018. 2

  5. [13]

    An image is worth 16x16 words: Trans- formers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, et al. An image is worth 16x16 words: Trans- formers for image recognition at scale. In Int. Conf. Le...

  6. [14]

    Pano popups: In- door 3d reconstruction with a plane-aware network

    Marc Eder, Pierre Moulon, and Li Guan. Pano popups: In- door 3d reconstruction with a plane-aware network. In Int. Conf. 3D Vision, pages 76–84, 2019. 2

  7. [15]

    Tangent images for mitigating spherical distortion

    Marc Eder, Mykhailo Shvets, John Lim, and Jan-Michael Frahm. Tangent images for mitigating spherical distortion. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 12426– 12434, 2020. 2

  8. [16]

    Desktop-based safety training using 360-degree panorama and static virtual reality techniques: A comparative exper- imental study

    Ricardo Eiris, Masoud Gheisari, and Behzad Esmaeili. Desktop-based safety training using 360-degree panorama and static virtual reality techniques: A comparative exper- imental study. Automation in construction , 109:102969,

  9. [17]

    Instantsplat: Un- bounded sparse-view pose-free gaussian splatting in 40 sec- onds

    Zhiwen Fan, Wenyan Cong, Kairun Wen, Kevin Wang, Jian Zhang, Xinghao Ding, Danfei Xu, Boris Ivanovic, Marco Pavone, Georgios Pavlakos, et al. Instantsplat: Un- bounded sparse-view pose-free gaussian splatting in 40 sec- onds. arXiv preprint arXiv:2403.20309, 2(3):4, 2024. 2

  10. [18]

    Large spatial model: End-to-end unposed images to semantic 3d

    Zhiwen Fan, Jian Zhang, Wenyan Cong, Peihao Wang, Renjie Li, Kairun Wen, Shijie Zhou, Achuta Kadambi, Zhangyang Wang, Danfei Xu, et al. Large spatial model: End-to-end unposed images to semantic 3d. In Adv. Neural Inform. Process. Syst., pages 40212–40229, 2025. 2

  11. [19]

    Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography.Communications of the ACM, 24(6):381–395, 1981

    Martin A Fischler and Robert C Bolles. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography.Communications of the ACM, 24(6):381–395, 1981. 6

  12. [20]

    Forward flow for novel view synthesis of dynamic scenes

    Xiang Guo, Jiadai Sun, Yuchao Dai, Guanying Chen, Xiao- qing Ye, Xiao Tan, Errui Ding, Yumeng Zhang, and Jingdong Wang. Forward flow for novel view synthesis of dynamic scenes. In Int. Conf. Comput. Vis. , pages 16022–16033,

  13. [21]

    Somsi: Spherical novel view synthesis with soft occlusion multi-sphere images

    Tewodros Habtegebrial, Christiano Gava, Marcel Rogge, Di- dier Stricker, and Varun Jampani. Somsi: Spherical novel view synthesis with soft occlusion multi-sphere images. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 15725– 15734, 2022. 8

  14. [22]

    In defense of the eight-point algorithm

    Richard I Hartley. In defense of the eight-point algorithm. IEEE Trans. Pattern Anal. Mach. Intell. , 19(6):580–593,

  15. [23]

    Rotary position embedding for vision transformer

    Byeongho Heo, Song Park, Dongyoon Han, and Sangdoo Yun. Rotary position embedding for vision transformer. In Eur. Conf. Comput. Vis., pages 289–305. Springer, 2024. 4

  16. [24]

    2d gaussian splatting for geometrically ac- curate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d gaussian splatting for geometrically ac- curate radiance fields. In ACM SIGGRAPH 2024 conference papers, pages 1–11, 2024. 3, 6

  17. [25]

    Unifuse: Unidirectional fusion for 360 panorama depth estimation

    Hualie Jiang, Zhe Sheng, Siyu Zhu, Zilong Dong, and Rui Huang. Unifuse: Unidirectional fusion for 360 panorama depth estimation. IEEE Robotics and Automation Letters, 6 (2):1519–1526, 2021. 2

  18. [26]

    Stereo4d: Learning how 9 things move in 3d from internet stereo videos

    Linyi Jin, Richard Tucker, Zhengqi Li, David Fouhey, Noah Snavely, and Aleksander Holynski. Stereo4d: Learning how 9 things move in 3d from internet stereo videos. InIEEE Conf. Comput. Vis. Pattern Recog., pages 10497–10509, 2025. 2

  19. [27]

    Selfsplat: Pose-free and 3d prior-free generalizable 3d gaussian splatting

    Gyeongjin Kang, Jisang Yoo, Jihyeon Park, Seungtae Nam, Hyeonsoo Im, Sangheon Shin, Sangpil Kim, and Eunbyung Park. Selfsplat: Pose-free and 3d prior-free generalizable 3d gaussian splatting. In IEEE Conf. Comput. Vis. Pattern Recog., pages 22012–22022, 2025. 2

  20. [28]

    3d gaussian splatting for real-time radiance field rendering

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

  21. [29]

    xformers: A modular and hackable trans- former modelling library

    Benjamin Lefaudeux, Francisco Massa, Diana Liskovich, Wenhan Xiong, Vittorio Caggiano, Sean Naren, Min Xu, Jieru Hu, Marta Tintore, Susan Zhang, Patrick Labatut, Daniel Haziza, Luca Wehrstedt, Jeremy Reizenstein, and Grigory Sizov. xformers: A modular and hackable trans- forme...

  22. [30]

    Slam with panoramic vision

    Thomas Lemaire and Simon Lacroix. Slam with panoramic vision. Journal of Field Robotics, 24(1-2):91–111, 2007. 1

  23. [31]

    Ep n p: An accurate o (n) solution to the p n p problem

    Vincent Lepetit, Francesc Moreno-Noguer, and Pascal Fua. Ep n p: An accurate o (n) solution to the p n p problem. Int. J. Comput. Vis., 81:155–166, 2009. 5

  24. [32]

    Ground- ing image matching in 3d with mast3r

    Vincent Leroy, Yohann Cabon, and J´erˆome Revaud. Ground- ing image matching in 3d with mast3r. InEur. Conf. Comput. Vis., pages 71–91, 2024. 1, 3, 4, 6

  25. [33]

    Re- conx: Reconstruct any scene from sparse views with video diffusion model

    Fangfu Liu, Wenqiang Sun, Hanyang Wang, Yikai Wang, Haowen Sun, Junliang Ye, Jun Zhang, and Yueqi Duan. Re- conx: Reconstruct any scene from sparse views with video diffusion model. arXiv preprint arXiv:2408.16767, 2024. 2

  26. [34]

    Augmented reality: a novel approach for navigating in panorama-based virtual environments (pbve)

    Peiran Liu, Xiaoyong Sun, Nicolas D Georganas, and Eric Dubois. Augmented reality: a novel approach for navigating in panorama-based virtual environments (pbve). In The 2nd IEEE Internatioal Workshop on Haptic, Audio and Visual Environments and Their Applications, 2003. HAVE 2...

  27. [35]

    Neural rays for occlusion-aware image-based ren- dering

    Yuan Liu, Sida Peng, Lingjie Liu, Qianqian Wang, Peng Wang, Christian Theobalt, Xiaowei Zhou, and Wenping Wang. Neural rays for occlusion-aware image-based ren- dering. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 7824–7833, 2022. 7

  28. [36]

    Slam3r: Real- time dense scene reconstruction from monocular rgb videos

    Yuzheng Liu, Siyan Dong, Shuzhe Wang, Yingda Yin, Yan- chao Yang, Qingnan Fan, and Baoquan Chen. Slam3r: Real- time dense scene reconstruction from monocular rgb videos. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 16651– 16662, 2025. 2

  29. [37]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 10012– 10022, 2021. 4

  30. [38]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017. 12

  31. [39]

    Distinctive image features from scale- invariant keypoints

    David G Lowe. Distinctive image features from scale- invariant keypoints. Int. J. Comput. Vis. , 60:91–110, 2004. 6

  32. [40]

    Align3r: Aligned monocular depth estima- tion for dynamic videos

    Jiahao Lu, Tianyu Huang, Peng Li, Zhiyang Dou, Cheng Lin, Zhiming Cui, Zhen Dong, Sai-Kit Yeung, Wenping Wang, and Yuan Liu. Align3r: Aligned monocular depth estima- tion for dynamic videos. In IEEE Conf. Comput. Vis. Pattern Recog., pages 22820–22830, 2025. 2

  33. [41]

    3d geometry-aware deformable gaussian splatting for dynamic view synthesis

    Zhicheng Lu, Xiang Guo, Le Hui, Tianrui Chen, Min Yang, Xiao Tang, Feng Zhu, and Yuchao Dai. 3d geometry-aware deformable gaussian splatting for dynamic view synthesis. In IEEE Conf. Comput. Vis. Pattern Recog., pages 8900–8910,

  34. [42]

    Fit: Flexible vision trans- former for diffusion model

    Zeyu Lu, Zidong Wang, Di Huang, Chengyue Wu, Xihui Liu, Wanli Ouyang, and Lei Bai. Fit: Flexible vision trans- former for diffusion model. InInt. Conf. Mach. Learn., pages 33160–33176. PMLR, 2024. 4

  35. [43]

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

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. In Eur. Conf. Comput. Vis., pages 405–421. Springer,

  36. [44]

    Epipolar-free 3d gaussian splatting for generalizable novel view synthesis

    Zhiyuan Min, Yawei Luo, Jianwen Sun, and Yi Yang. Epipolar-free 3d gaussian splatting for generalizable novel view synthesis. In Adv. Neural Inform. Process. Syst., pages 39573–39596, 2024. 2

  37. [45]

    High-resolution depth estimation for 360deg panoramas through perspective and panoramic depth images registration

    Chi-Han Peng and Jiayao Zhang. High-resolution depth estimation for 360deg panoramas through perspective and panoramic depth images registration. In IEEE/CVF Win- ter Conference on Applications of Computer Vision , pages 3116–3125, 2023. 2

  38. [46]

    Habitat-matterport 3d dataset (hm3d): 1000 large-scale 3d environments for embodied ai

    Santhosh K Ramakrishnan, Aaron Gokaslan, Erik Wijmans, Oleksandr Maksymets, Alex Clegg, John Turner, Eric Un- dersander, Wojciech Galuba, Andrew Westbury, Angel X Chang, et al. Habitat-matterport 3d dataset (hm3d): 1000 large-scale 3d environments for embodied ai. arXiv prepri...

  39. [47]

    Vi- sion transformers for dense prediction

    Ren ´e Ranftl, Alexey Bochkovskiy, and Vladlen Koltun. Vi- sion transformers for dense prediction. In Int. Conf. Comput. Vis., pages 12179–12188, 2021. 5

  40. [48]

    360monodepth: High-resolution 360deg monocular depth estimation

    Manuel Rey-Area, Mingze Yuan, and Christian Richardt. 360monodepth: High-resolution 360deg monocular depth estimation. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 3762–3772, 2022. 2

  41. [49]

    Habitat: A plat- form for embodied ai research

    Manolis Savva, Abhishek Kadian, Oleksandr Maksymets, Yili Zhao, Erik Wijmans, Bhavana Jain, Julian Straub, Jia Liu, Vladlen Koltun, Jitendra Malik, et al. Habitat: A plat- form for embodied ai research. In Int. Conf. Comput. Vis. , pages 9339–9347, 2019. 5, 12

  42. [50]

    Panoformer: Panorama transformer for indoor 360 depth estimation

    Zhijie Shen, Chunyu Lin, Kang Liao, Lang Nie, Zishuo Zheng, and Yao Zhao. Panoformer: Panorama transformer for indoor 360 depth estimation. In Eur. Conf. Comput. Vis., pages 195–211. Springer, 2022. 2

  43. [51]

    Splatt3r: Zero-shot gaussian splatting from uncalibrated image pairs

    Brandon Smart, Chuanxia Zheng, Iro Laina, and Vic- tor Adrian Prisacariu. Splatt3r: Zero-shot gaussian splatting from uncalibrated image pairs. arXiv preprint arXiv:2408.13912, 2024. 2

  44. [52]

    The replica dataset: A digital replica of indoor spaces

    Julian Straub, Thomas Whelan, Lingni Ma, Yufan Chen, Erik Wijmans, Simon Green, Jakob J Engel, Raul Mur-Artal, Carl Ren, Shobhit Verma, et al. The replica dataset: A digital replica of indoor spaces. arXiv preprint arXiv:1906.05797,

  45. [53]

    Roformer: Enhanced transformer with rotary position embedding

    Jianlin Su, Murtadha Ahmed, Yu Lu, Shengfeng Pan, Wen Bo, and Yunfeng Liu. Roformer: Enhanced transformer with rotary position embedding. Neurocomputing, 568:127063,

  46. [54]

    Hohonet: 360 indoor holistic understanding with latent horizontal fea- tures

    Cheng Sun, Min Sun, and Hwann-Tzong Chen. Hohonet: 360 indoor holistic understanding with latent horizontal fea- tures. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 2573–2582, 2021. 3

  47. [55]

    Hisplat: Hierarchical 3d gaus- sian splatting for generalizable sparse-view reconstruction

    Shengji Tang, Weicai Ye, Peng Ye, Weihao Lin, Yang Zhou, Tao Chen, and Wanli Ouyang. Hisplat: Hierarchical 3d gaus- sian splatting for generalizable sparse-view reconstruction. In Int. Conf. Learn. Represent., 2025. 5, 6

  48. [56]

    Mv-dust3r+: Single-stage scene reconstruction from sparse views in 2 seconds

    Zhenggang Tang, Yuchen Fan, Dilin Wang, Hongyu Xu, Rakesh Ranjan, Alexander Schwing, and Zhicheng Yan. Mv-dust3r+: Single-stage scene reconstruction from sparse views in 2 seconds. In IEEE Conf. Comput. Vis. Pattern Recog., pages 5283–5293, 2025. 2

  49. [57]

    Distortion-aware convolutional filters for dense prediction in panoramic images

    Keisuke Tateno, Nassir Navab, and Federico Tombari. Distortion-aware convolutional filters for dense prediction in panoramic images. In Eur. Conf. Comput. Vis., pages 707– 722, 2018. 2

  50. [58]

    Bifuse: Monocular 360 depth estimation via bi-projection fusion

    Fu-En Wang, Yu-Hsuan Yeh, Min Sun, Wei-Chen Chiu, and Yi-Hsuan Tsai. Bifuse: Monocular 360 depth estimation via bi-projection fusion. In IEEE Conf. Comput. Vis. Pattern Recog., pages 462–471, 2020. 2

  51. [59]

    3d reconstruction with spatial memory

    Hengyi Wang and Lourdes Agapito. 3d reconstruction with spatial memory. In Int. Conf. 3D Vision, 2025. 2

  52. [60]

    Ibr- net: Learning multi-view image-based rendering

    Qianqian Wang, Zhicheng Wang, Kyle Genova, Pratul P Srinivasan, Howard Zhou, Jonathan T Barron, Ricardo Martin-Brualla, Noah Snavely, and Thomas Funkhouser. Ibr- net: Learning multi-view image-based rendering. In IEEE Conf. Comput. Vis. Pattern Recog., pages 4690–4699, 2021. 7

  53. [61]

    Continuous 3d per- ception model with persistent state

    Qianqian Wang, Yifei Zhang, Aleksander Holynski, Alexei A Efros, and Angjoo Kanazawa. Continuous 3d per- ception model with persistent state. In IEEE Conf. Comput. Vis. Pattern Recog., pages 10510–10522, 2025. 2

  54. [62]

    Dust3r: Geometric 3d vi- sion made easy

    Shuzhe Wang, Vincent Leroy, Yohann Cabon, Boris Chidlovskii, and Jerome Revaud. Dust3r: Geometric 3d vi- sion made easy. In IEEE Conf. Comput. Vis. Pattern Recog., pages 20697–20709, 2024. 1, 2, 3, 4

  55. [63]

    Looprefine: Deep camera pose estimation with loop consistency

    Zhiwei Wang, Hui Deng, Jiawei Shi, Mochu Xiang, Zhicheng Lu, Qi Liu, and Yuchao Dai. Looprefine: Deep camera pose estimation with loop consistency. IEEE Robotics and Automation Letters, 2025. 2

  56. [64]

    Croco v2: Improved cross-view completion pre- training for stereo matching and optical flow

    Philippe Weinzaepfel, Thomas Lucas, Vincent Leroy, Yohann Cabon, Vaibhav Arora, Romain Br ´egier, Gabriela Csurka, Leonid Antsfeld, Boris Chidlovskii, and J ´erˆome Revaud. Croco v2: Improved cross-view completion pre- training for stereo matching and optical flow. In Int. Con...

  57. [65]

    Depthsplat: Connecting gaussian splatting and depth

    Haofei Xu, Songyou Peng, Fangjinhua Wang, Hermann Blum, Daniel Barath, Andreas Geiger, and Marc Polle- feys. Depthsplat: Connecting gaussian splatting and depth. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 16453– 16463, 2025. 5, 6

  58. [66]

    Freesplatter: Pose- free gaussian splatting for sparse-view 3d reconstruction

    Jiale Xu, Shenghua Gao, and Ying Shan. Freesplatter: Pose- free gaussian splatting for sparse-view 3d reconstruction. arXiv preprint arXiv:2412.09573, 2024. 2

  59. [67]

    Fast3r: Towards 3d reconstruction of 1000+ images in one forward pass

    Jianing Yang, Alexander Sax, Kevin J Liang, Mikael Henaff, Hao Tang, Ang Cao, Joyce Chai, Franziska Meier, and Matt Feiszli. Fast3r: Towards 3d reconstruction of 1000+ images in one forward pass. In IEEE Conf. Comput. Vis. Pattern Recog., pages 21924–21935, 2025. 2

  60. [68]

    No pose, no problem: Surprisingly simple 3d gaussian splats from sparse unposed images

    Botao Ye, Sifei Liu, Haofei Xu, Xueting Li, Marc Pollefeys, Ming-Hsuan Yang, and Songyou Peng. No pose, no problem: Surprisingly simple 3d gaussian splats from sparse unposed images. In Int. Conf. Learn. Represent., 2024. 2, 5

  61. [69]

    pixelnerf: Neural radiance fields from one or few images

    Alex Yu, Vickie Ye, Matthew Tancik, and Angjoo Kanazawa. pixelnerf: Neural radiance fields from one or few images. In IEEE Conf. Comput. Vis. Pattern Recog., pages 4578–4587,

  62. [70]

    Panelnet: Understanding 360 indoor environment via panel representation

    Haozheng Yu, Lu He, Bing Jian, Weiwei Feng, and Shan Liu. Panelnet: Understanding 360 indoor environment via panel representation. In IEEE Conf. Comput. Vis. Pattern Recog., pages 878–887, 2023. 3

  63. [71]

    Viewcrafter: Taming video diffusion models for high-fidelity novel view synthesis.arXiv preprint arXiv:2409.02048, 2024

    Wangbo Yu, Jinbo Xing, Li Yuan, Wenbo Hu, Xiaoyu Li, Zhipeng Huang, Xiangjun Gao, Tien-Tsin Wong, Ying Shan, and Yonghong Tian. Viewcrafter: Taming video diffusion models for high-fidelity novel view synthesis.arXiv preprint arXiv:2409.02048, 2024. 2

  64. [72]

    Egformer: Equirectangular geometry- biased transformer for 360 depth estimation

    Ilwi Yun, Chanyong Shin, Hyunku Lee, Hyuk-Jae Lee, and Chae Eun Rhee. Egformer: Equirectangular geometry- biased transformer for 360 depth estimation. In Int. Conf. Comput. Vis., pages 6101–6112, 2023. 2

  65. [73]

    Pansplat: 4k panorama synthesis with feed-forward gaussian splatting

    Cheng Zhang, Haofei Xu, Qianyi Wu, Camilo Cruz Gam- bardella, Dinh Phung, and Jianfei Cai. Pansplat: 4k panorama synthesis with feed-forward gaussian splatting. In IEEE Conf. Comput. Vis. Pattern Recog. , pages 11437– 11447, 2025. 1, 3, 4, 7

  66. [74]

    Bending reality: Distortion-aware transformers for adapting to panoramic se- mantic segmentation

    Jiaming Zhang, Kailun Yang, Chaoxiang Ma, Simon Reiß, Kunyu Peng, and Rainer Stiefelhagen. Bending reality: Distortion-aware transformers for adapting to panoramic se- mantic segmentation. In IEEE Conf. Comput. Vis. Pattern Recog., pages 16917–16927, 2022. 2

  67. [75]

    Monst3r: A simple approach for estimating geometry in the presence of motion

    Junyi Zhang, Charles Herrmann, Junhwa Hur, Varun Jam- pani, Trevor Darrell, Forrester Cole, Deqing Sun, and Ming- Hsuan Yang. Monst3r: A simple approach for estimating geometry in the presence of motion. In Int. Conf. Learn. Represent., 2025. 2

  68. [76]

    Acdnet: Adaptively combined dilated con- volution for monocular panorama depth estimation

    Chuanqing Zhuang, Zhengda Lu, Yiqun Wang, Jun Xiao, and Ying Wang. Acdnet: Adaptively combined dilated con- volution for monocular panorama depth estimation. InAAAI, pages 3653–3661, 2022. 2

  69. [77]

    Omnidepth: Dense depth estimation for indoors spherical panoramas

    Nikolaos Zioulis, Antonis Karakottas, Dimitrios Zarpalas, and Petros Daras. Omnidepth: Dense depth estimation for indoors spherical panoramas. In Eur. Conf. Comput. Vis. , pages 448–465, 2018. 2 11 PanoSplatt3R: Leveraging Perspective Pretraining for Generalized Unposed Wide-B...

  70. [386]

    2, 5, 6, 7, 13, 14

    Springer, 2024. 2, 5, 6, 7, 13, 14

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.