Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

FastMap: Revisiting Structure from Motion through First-Order Optimization

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read FastMap claims that global structure-from-motion does not need second-order bundle adjustment or explicit 3D points to be both fast and accurate on dense, high-overlap image collections.

desk verdict Fast global SfM with first-order-only optimizers: the speed claim is solid, the 'comparable accuracy' claim is conditional on relaxed metrics; worth publishing after recalibration and code release. read the letter →

arxiv 2505.04612 v3 pith:QVNRWRUS submitted 2025-05-07 cs.CV

Jiahao Li , Haochen Wang , Muhammad Zubair Irshad , Igor Vasiljevic , Matthew R. Walter , Vitor Campagnolo Guizilini , Greg Shakhnarovich This is my paper · ORCID
classification cs.CV
keywords structurefrommotionglobalSfMfirst-orderoptimizationbundleadjustmentepipolargeometryiterativereweightedleastsquarescameraposeestimationGPUkernelfusion
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

FastMap claims that global structure-from-motion does not need second-order bundle adjustment or explicit 3D points to be both fast and accurate on dense, high-overlap image collections. The paper builds a global SfM pipeline whose optimization steps are all first-order, so each iteration costs time linear in the number of image pairs rather than in the number of matched keypoints. Across eight real-world datasets from roughly 200 to 6,000 images, it reports being up to 10 times faster than GPU-accelerated COLMAP and GLOMAP while matching their pose accuracy at relaxed thresholds and producing competitive novel-view-synthesis quality. The practical stake is that generating camera-pose pseudo-ground-truth for large scene collections, currently an hours-to-days bottleneck, could drop to minutes on a single GPU.

What carries the argument

The load-bearing mechanism is reweighted epipolar adjustment (Eq. 9). For each image pair, the squared epipolar error of all point pairs is reorganized as $e_n^\top \hat W_n e_n$, where $e_n = \operatorname{flatten}(E_n) \in \mathbb{R}^9$ is the flattened essential matrix and $\hat W_n \in \mathbb{R}^{9\times 9}$ is a precomputed matrix formed from point pairs weighted by the inverse absolute epipolar error $1/|\hat{\epsilon}_{nm}|$ measured at initialization. This gives an approximation to a robust $\ell^1$ epipolar loss whose evaluation and gradient require only small matrix products per image pair, independent of the number of 3D points. A second mechanism, kernel fusion, packs the forward and backward passes of these small-matrix operations into a single custom CUDA kernel, eliminating kernel-launch and data-movement overhead; the paper reports speedups of roughly 20 to 90 times over an autograd-based implementation of the same update.

What would settle it

Take a dense scene whose translation alignment already succeeds, then deliberately perturb the initial camera poses by rotations and translations of growing size before running epipolar adjustment; if the final pose error stays low for perturbations up to several degrees the proximity assumption is mild, but if accuracy abruptly degrades past a small perturbation radius, the assumption is confirmed as the load-bearing condition. A second direct check is running FastMap on a straight-line or SLAM-style trajectory dataset: the paper predicts strong failure because relative-motion-only constraints cannot disambiguate baseline distances.

Watch

Extended reading notes

Core claim

The paper's central claim is that a global SfM pipeline can be built entirely from first-order optimizers and still match the pose quality of standard second-order SfM systems on dense scenes. The key move is replacing bundle adjustment with a point-free epipolar adjustment: the absolute epipolar error is rewritten, via iterative reweighted least squares, as a weighted $\ell^2$ cost over one small precomputed matrix per image pair, so no 3D point enters the optimization and the per-step cost is linear in the number of image pairs. FastMap is reported to be up to 10 times faster than GPU-accelerated COLMAP and GLOMAP, with comparable RTA@3 accuracy on most datasets but a clearer gap at stricter one-degree metrics. The paper openly documents that the method degrades on sparse-view scenes, repetitive or symmetric structures, and degenerate straight-line motions, where some of its components fail or become unreliable.

Load-bearing premise

Everything downstream of translation alignment depends on the initial poses being close enough to the optimum that the point weights frozen at initialization correctly approximate an $\ell^1$ epipolar loss; on sparse views or degenerate camera motion the paper's own results show this assumption fails and accurate poses are not recovered.

Editorial extensions

If this is right

  • On dense high-overlap scenes, camera-pose estimation would drop from hours or days to minutes on a single GPU, since every stage avoids second-order solvers and point variables.
  • Applications that consume SfM poses at relaxed tolerances, such as NeRF and Gaussian-splatting training, would see similar view-synthesis quality with a fraction of the backend time.
  • Scaling behavior would be governed by the number of image pairs rather than the number of matched keypoints, making very large image collections more tractable.
  • The documented gap at strict one-degree pose metrics and the failures on sparse, repetitive, or degenerate scenes would delineate exactly where incremental bundle adjustment remains necessary.

Reading between the lines

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

  • An editor-level extension: the point-free IRLS epipolar objective is agnostic to how correspondences are obtained, so it could in principle consume dense or learned correspondences without changing the optimization machinery; the paper does not test this.
  • Because the proximity assumption on initialization is the bottleneck, an adaptive controller that switches in a few second-order iterations only when epipolar residuals fail to decrease would likely extend FastMap to sparse or degenerate scenes while preserving most of the speedup; this is not in the paper.
  • Track completion turns matching transitivity into pairwise constraints without triangulation, which suggests that explicit 3D reconstruction may be unnecessary for pose recovery; one testable consequence is that pose quality should be largely unchanged when triangulated point clouds are omitted from downstream use, something the paper does not isolate.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. FastMap is a global structure-from-motion pipeline that estimates camera poses using only first-order optimization. It avoids explicit bundle adjustment over 3D points by solving rotation alignment, translation alignment, and a reweighted epipolar adjustment whose per-iteration cost is linear in the number of image pairs, and it implements the gradient computation as a single fused CUDA kernel. The paper reports 3–10x speedups over GPU-accelerated COLMAP and GLOMAP on eight datasets, with FastMap using only 2 CPU threads while baselines use 48, and claims comparable pose accuracy and downstream novel-view-synthesis quality. The appendix includes ablations for kernel fusion, distortion estimation, track completion, multiple translation initializations, and epipolar adjustment, along with a candid limitations section documenting failures on ETH3D and LaMAR.

Significance. If the speed claims are accepted, this is a significant practical contribution: it demonstrates that a carefully engineered first-order pipeline can compete on wall-clock time with second-order solvers in the dense high-overlap regime, and the point-free problem formulations with precomputed per-pair 9x9 quadratic forms are elegant and useful. The kernel-fusion ablation (Table 4) is clean and makes the speedup mechanism concrete. The pose-accuracy half of the claim is, however, conditional: the method is competitive at relaxed thresholds (RTA@3) on many scenes, but the abstract's unqualified 'comparable pose accuracy' is not supported across strict metrics and difficult scenes. The paper is honest about these limitations, which strengthens the credibility of the speed results but does not eliminate the need to scope the accuracy claim.

major comments (3)
  1. [Section 3.3, Eq. (9)] The reweighted epipolar adjustment is load-bearing for the pose-accuracy claim, but its validity depends on the initialization being close to the optimum, as the paper itself states. In Eq. (9) the per-point weights are computed once from residuals at the translation-alignment output and then frozen while the weighted L2 problem is optimized; the approximation z^2 ≈ z^2/|z_hat| is only local, and the periodic outlier filter can delete large-residual inliers that are actually needed to escape a bad basin. The consequences are visible in the paper's own tables: on Tanks-and-Temples Advanced (Table 1; per-scene Table 8), FastMap's RTA@1 is 42.3 versus 77.5 for GLOMAP and 97.0 for COLMAP, and Courthouse collapses to RTA@3 40.9 versus 97.3 for GLOMAP; Appendix C additionally reports widespread failures on ETH3D (Table 15) and LaMAR (Table 16). Section C's disclosure is welcome, but it does not fix the mismatch between the abstract's 'comparable pose accuracy' and the conditional behavior of Eq. (9). Please either restrict the central claim to the regime where the initialization condition holds, or add a measurable diagnostic for that condition and show it is satisfied on the claimed-accuracy datasets.
  2. [Abstract and Section 4.2 / Table 1] The 'comparable pose accuracy' claim is too broad even outside the acknowledged failure regimes. The strict relative pose metrics show consistent gaps: Tanks-and-Temples Advanced AUC@1 is 16.7 versus 69.8 (GLOMAP) and 90.0 (COLMAP); Eyeful Tower apartment RTA@1 is 6.4 versus 18.2/21.9; and tnttraining RTA@3 is 87.8 versus 99.9 for COLMAP. Section 4.2 acknowledges that 'the difference is more prominent for stricter metrics,' but the abstract and the opening of Section 5 state 'comparable pose accuracy' without this qualification. The authors should either revise the claim to specify relaxed thresholds on non-degenerate dense scenes, or provide evidence that the strict-metric gaps do not matter for the intended downstream tasks. This is not a request for perfection, but for the claim to match the reported evidence.
  3. [Appendix B.8] The evaluation of pose accuracy is partially self-referential: for MipNeRF360, ZipNeRF, and NeRF-OSR, the reference poses are generated by COLMAP, which is one of the two baselines compared against. COLMAP's own output therefore cannot be wrong relative to the reference, and any error in COLMAP is absorbed into the ground truth used to judge all methods. A similar situation holds for the Tanks-and-Temples reference poses taken from Kulhanek and Sattler [31]. I do not consider this a fatal flaw, since author-provided or externally published references are common, but for a claim of 'comparable pose accuracy' the paper should either add datasets with independent ground truth (e.g., the references already used for Mill-19 and Urbanscene3D) or quantify the possible bias, for instance by comparing relative errors on a subset against an independent reference.
minor comments (4)
  1. [Section 3.4] There is a typo in the paragraph on kernel design: 'his leads to substantial inefficiency' should read 'This leads to substantial inefficiency.'
  2. [Throughout] The method name is inconsistently formatted as 'FastMap', 'FAST MAP', and 'FAST MAP' across the title, abstract, and body; please choose one convention and use it consistently.
  3. [Section 3.1, Eq. (2) and Section 3.3, Eq. (9)] The temperature parameter tau in Eq. (2) and the IRLS outlier threshold schedule are not given concrete default values or a sensitivity study; please report the default values and add a short sensitivity analysis or ablation.
  4. [Table 4] The hardware notation in Table 4 (e.g., '4.05GHz', '2.2GHz', '2080 Ti') would benefit from explicit CPU/GPU model names and a note on how these configurations relate to the main experiments in Table 1.

Circularity Check

1 steps flagged · score 2.0 of 10

Derivation is self-contained first-order SfM; the only self-reference is that several evaluation datasets use COLMAP-generated poses as ground truth while COLMAP is a comparison baseline.

  1. other [Appendix B.8 (Data Ground Truth)]
    "With the exception of Tanks and Temples, each of the datasets we evaluate on includes author-provided reference camera poses. These reference poses are obtained through different means, including COLMAP (for MipNeRF360, ZipNeRF, NeRF-OSR), PixSfM [35] (for Mill-19 and Urbanscene3D), and commercial software (for DroneDeploy and Eyeful Tower)."

    This is an evaluation-level self-reference rather than a derivation-level circularity. The pose-accuracy comparisons in Table 1 evaluate FastMap, GLOMAP, and COLMAP against reference poses that, for three of the eight datasets, are themselves COLMAP reconstructions. Thus COLMAP's 'accuracy' on those datasets is measured against its own output, and FastMap's claim of 'comparable pose accuracy' is assessed against a target generated by one of the compared baselines. This can bias the comparison in COLMAP's favor, but it does not make FastMap's pose estimates depend on COLMAP's poses, and it does not force FastMap's reported numbers by construction.

full rationale

No derivation-level circularity was found. The core optimization chain is self-contained and built on classical, externally cited formulations: global rotation averaging (Eqs. 3-4) with a Martinec-Pajdla SVD initialization; global translation alignment (Eqs. 5-6) minimizing normalized relative-translation error; and epipolar adjustment (Eqs. 7-9) with precomputed per-pair quadratic forms attributed to Rodriguez et al. The IRLS weighting in Eq. 9 is an explicitly stated approximation ('if we have an initialization close to the optimum... z^2 ≈ z^2/|z_hat|'), and the paper documents regimes where that initialization condition fails (Appendix C: ETH3D, LaMAR). That is a robustness limitation, not a circular reduction. There are no fitted constants disguised as predictions, no load-bearing self-citations, no imported uniqueness theorems, and no ansatz smuggled in via citation. The only self-referential element is evaluative: several datasets use COLMAP-generated poses as ground truth while COLMAP is one of the comparison baselines (Appendix B.8). This is a benchmark-level bias, noted in the flagged step, but it does not enter the derivation of FastMap's poses. Accordingly, the paper is assigned a low circularity score of 2.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

FastMap introduces no new physical entities. It relies on two domain assumptions (shared camera grouping and dense coverage) plus an approximation assumption for the IRLS-weighted epipolar loss. The free parameters are hyperparameters of the search and optimization loops; their values are not fully specified in the paper.

free parameters (4)
  • tau (focal length validity temperature) = not stated
    Introduced in Eq 2 to control how sharply the essential-matrix singular value ratio is penalized; the value is not given in the main text.
  • IRLS outlier threshold schedule = not stated
    Periodic filtering of high-epipolar-error point pairs uses a threshold that starts large and decreases to a minimum; the values and schedule are not specified.
  • Number of translation initializations m = not stated in final configuration (ablation shows m=1..8)
    Multiple random initializations are merged; the final number used is not stated, and Table 13 shows the choice affects RTE@30.
  • First-order optimizer settings (step size, iterations) = not stated
    The paper says 'gradient descent' but does not specify optimizer, learning rate, or iteration counts for rotation, translation, or epipolar stages.
assumptions (4)
  • domain assumption All images are taken with a small number of distinct cameras, and the grouping is known.
    Section 3.1 'Camera Assumptions'; the distortion and focal length search procedures require at least one pair sharing intrinsics, and fail when every image has different intrinsics.
  • domain assumption The input images densely cover the 3D scene so that outlier pairs average out.
    Section C 'Limitations' states this assumption explicitly; sparse coverage leads to sensitivity to outliers and the method underperforms GLOMAP on ETH3D.
  • domain assumption The IRLS approximation z^2 ~ z^2/|z_hat| is valid because the initialization from translation alignment is close to the optimum.
    Section 3.3 'Epipolar Adjustment' relies on this to justify reweighted L2 as a robust L1 surrogate; when initialization is poor (sparse or degenerate scenes), refinement fails.
  • standard math A 3x3 matrix is an essential matrix if and only if its singular values satisfy lambda1 = lambda2 and lambda3 = 0.
    Standard result from Hartley and Zisserman [20], used in Section 3.1 for focal length estimation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FastMap: Revisiting Structure from Motion through First-Order Optimization." pith.science (2026). https://pith.science/paper/QVNRWRUS

@misc{pith2026250504612,
  author       = {Pith},
  title        = {Pith review of: FastMap: Revisiting Structure from Motion through First-Order Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QVNRWRUS}},
  note         = {Machine review of arXiv:2505.04612}
}
read the original abstract

We propose FastMap, a new global structure from motion method focused on speed and simplicity. Previous methods like COLMAP and GLOMAP are able to estimate high-precision camera poses, but suffer from poor scalability when the number of matched keypoint pairs becomes large, mainly due to the time-consuming process of second-order Gauss-Newton optimization. Instead, we design our method solely based on first-order optimizers. To obtain maximal speedup, we identify and eliminate two key performance bottlenecks: computational complexity and the kernel implementation of each optimization step. Through extensive experiments, we show that FastMap is up to 10 times faster than COLMAP and GLOMAP with GPU acceleration and achieves comparable pose accuracy.

Figures

Figures reproduced from arXiv: 2505.04612 by the authors.

Figure 1
Figure 1. Timing of FASTMAP compared to COLMAP and GLOMAP (all with GPU acceleration on a single A6000) on scenes from eight datasets, excluding the matching stage for all methods. Note the logarithmic time scale. Lines represent a least squares power function fit to timing across multiple datasets, as a function of the number of matched keypoint pairs. quires a fast and high-quality ground-truth annotator. A typical SfM pipe… view at source ↗
Figure 2
Figure 2. An overview of FASTMAP. Input images are processed using feature extraction and matching. Given the matching results, FASTMAP estimates the intrinsics and extrinsics of the cameras. Finally a sparse point cloud is generated by triangulation. construction. Many recent works [36, 63, 70] improve upon DUSt3R in various ways, such as efficiently process￾ing more input images [70], using diffusion models [36], and predic… view at source ↗
Figure 3
Figure 3. Effect of distortion on focal length estimation. Curves in [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Rig3R: Rig-Aware Conditioning for Learned 3D Reconstruction

    cs.CV 2025-06 conditional novelty 7.0 of 10

    Rig3R conditions learned 3D reconstruction on optional rig metadata and predicts rig-relative raymaps, enabling state-of-the-art pose estimation and rig calibration discovery from images.

Reference graph

Works this paper leans on

75 extracted references · 61 canonical work pages · cited by 1 Pith paper

  1. [31]

    NeRFBaselines: Con- sistent and reproducible evaluation of novel view synthesis methods

    Jonas Kulhanek and Torsten Sattler. NeRFBaselines: Con- sistent and reproducible evaluation of novel view synthesis methods. arXiv preprint arXiv:2406.17345, 2024. 7, 16

    arXiv 2024

  2. [1]

    Ceres Solver, 2023

    Sameer Agarwal, Keir Mierle, and The Ceres Solver Team. Ceres Solver, 2023. 2, 4, 6

    work page 2023

  3. [2]

    A mini- mal solution for two-view focal-length estimation using two affine correspondences

    Daniel Barath, Tekla Toth, and Levente Hajder. A mini- mal solution for two-view focal-length estimation using two affine correspondences. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2017. 3

    work page 2017

  4. [3]

    Fundamental ma- trix for cameras with radial distortion

    Jo ˜ao Pedro Barreto and Kostas Daniilidis. Fundamental ma- trix for cameras with radial distortion. In Proceedings of the International Conference on Computer Vision (ICCV), 2005. 3

    work page 2005

  5. [4]

    Barron, Ben Mildenhall, Dor Verbin, Pratul P

    Jonathan T. Barron, Ben Mildenhall, Dor Verbin, Pratul P. Srinivasan, and Peter Hedman. Mip-NeRF 360: Unbounded anti-aliased neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 6, 7, 16

    work page 2022

  6. [5]

    Barron, Ben Mildenhall, Dor Verbin, Pratul P

    Jonathan T. Barron, Ben Mildenhall, Dor Verbin, Pratul P. Srinivasan, and Peter Hedman. Zip-NeRF: Anti-aliased grid- based neural radiance fields. In Proceedings of the Interna- tional Conference on Computer Vision (ICCV) , 2023. 6, 8, 16

    work page 2023

  7. [6]

    Zoedepth: Zero-shot trans- fer by combining relative and metric depth

    Shariq Farooq Bhat, Reiner Birkl, Diana Wofk, Peter Wonka, and Matthias M ¨uller. Zoedepth: Zero-shot trans- fer by combining relative and metric depth. arXiv preprint arXiv:2302.12288, 2023. 1

    arXiv 2023

  8. [7]

    Scene coordinate reconstruction: Pos- ing of image collections via incremental learning of a relo- calizer

    Eric Brachmann, Jamie Wynn, Shuai Chen, Tommaso Cav- allari, ´Aron Monszpart, Daniyar Turmukhambetov, and Vic- tor Adrian Prisacariu. Scene coordinate reconstruction: Pos- ing of image collections via incremental learning of a relo- calizer. In Proceedings of the European Conference on Com- puter Vision (ECCV), 2024. 2, 7, 8

    work page 2024

Show all 75 references
  1. [8]

    HSfM: Hybrid structure-from-motion

    Hainan Cui, Xiang Gao, Shuhan Shen, and Zhanyi Hu. HSfM: Hybrid structure-from-motion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2017. 2

    2017

  2. [9]

    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. In Proceedings of the IEEE/CVF conference on computer vision and pattern rec...

    2023

  3. [10]

    Close-range camera calibration

    C Brown Duane. Close-range camera calibration. Pho- togramm. Eng, 37(8), 1971. 3

    1971

  4. [11]

    MASt3R-SfM: A fully-integrated solution for unconstrained structure-from-motion

    Bardienus Duisterhof, Lojze Zust, Philippe Weinzaepfel, Vincent Leroy, Yohann Cabon, and Jerome Revaud. MASt3R-SfM: A fully-integrated solution for unconstrained structure-from-motion. arXiv preprint arXiv:2409.19152 ,

    arXiv

  5. [12]

    A review of the one-parameter division undistortion model

    Bastian Erdn ¨uß. A review of the one-parameter division undistortion model. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences , 1, 2021. 3

    2021

  6. [13]

    Three-dimensional computer vision: A ge- ometric viewpoint

    Olivier Faugeras. Three-dimensional computer vision: A ge- ometric viewpoint. MIT Press, 1993. 3

    1993

  7. [14]

    Simultaneous linear estimation of multiple view geometry and lens distortion

    Andrew W Fitzgibbon. Simultaneous linear estimation of multiple view geometry and lens distortion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), 2001. 3

    2001

  8. [15]

    All for one, and one for all: Urbansyn dataset, the third musketeer of synthetic driv- ing scenes

    Jose L G ´omez, Manuel Silva, Antonio Seoane, Agn `es Borr´as, Mario Noriega, Germ ´an Ros, Jose A Iglesias- Guitian, and Antonio M L ´opez. All for one, and one for all: Urbansyn dataset, the third musketeer of synthetic driv- ing scenes. Neurocomputing, 637:130038, 2025. 1

    2025

  9. [16]

    Combining two-view constraints for motion estimation

    Venu Madhav Govindu. Combining two-view constraints for motion estimation. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) ,

  10. [17]

    Kubric: A scalable dataset generator

    Klaus Greff, Francois Belletti, Lucas Beyer, Carl Doersch, Yilun Du, Daniel Duckworth, David J Fleet, Dan Gnanapra- gasam, Florian Golemo, Charles Herrmann, et al. Kubric: A scalable dataset generator. In Proceedings of the IEEE/CVF conference on computer vision and pattern re...

    2022

  11. [18]

    Alltracker: Effi- cient dense point tracking at high resolution

    Adam W Harley, Yang You, Xinglong Sun, Yang Zheng, Nikhil Raghuraman, Yunqi Gu, Sheldon Liang, Wen-Hsuan Chu, Achal Dave, Pavel Tokmakov, et al. Alltracker: Effi- cient dense point tracking at high resolution. arXiv preprint arXiv:2506.07310, 2025. 1

    2025 arXiv

  12. [19]

    Extraction of focal lengths from the funda- mental matrix

    Richard Hartley. Extraction of focal lengths from the funda- mental matrix. Unpublished manuscript, 2, 1993. 3

    1993

  13. [20]

    Multiple view ge- ometry in computer vision

    Richard Hartley and Andrew Zisserman. Multiple view ge- ometry in computer vision . Cambridge University Press,

  14. [21]

    Rotation averaging

    Richard Hartley, Jochen Trumpf, Yuchao Dai, and Hongdong Li. Rotation averaging. International Journal on Computer Vision, 103, 2013. 2

    2013

  15. [22]

    Radial distortion self-calibration

    Jose Henrique Brito, Roland Angst, Kevin Koser, and Marc Pollefeys. Radial distortion self-calibration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), 2013. 3

    2013

  16. [23]

    A global linear method for camera pose registration

    Nianjuan Jiang, Zhaopeng Cui, and Ping Tan. A global linear method for camera pose registration. In Proceedings of the International Conference on Computer Vision (ICCV), 2013. 2, 5

    2013

  17. [24]

    Closed-form ex- pression for focal lengths from the fundamental matrix

    Kenichi Kanatani and Chikara Matsunaga. Closed-form ex- pression for focal lengths from the fundamental matrix. In Proceedings of the Asian Conference on Computer Vision ,

  18. [25]

    Co- tracker: It is better to track together

    Nikita Karaev, Ignacio Rocco, Benjamin Graham, Natalia Neverova, Andrea Vedaldi, and Christian Rupprecht. Co- tracker: It is better to track together. In European conference on computer vision, pages 18–35. Springer, 2024. 1

    2024

  19. [26]

    Repurpos- ing diffusion-based image generators for monocular depth estimation

    Bingxin Ke, Anton Obukhov, Shengyu Huang, Nando Met- zger, Rodrigo Caye Daudt, and Konrad Schindler. Repurpos- ing diffusion-based image generators for monocular depth estimation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9492–9502,

  20. [27]

    3D Gaussian splatting for real-time 9 radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3D Gaussian splatting for real-time 9 radiance field rendering. ACM Transactions on Graphics, 42 (4), 2023. 7, 8

    2023

  21. [28]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 ,

    arXiv

  22. [29]

    Tanks and temples: Benchmarking large-scale scene reconstruction

    Arno Knapitsch, Jaesik Park, Qian-Yi Zhou, and Vladlen Koltun. Tanks and temples: Benchmarking large-scale scene reconstruction. ACM Transactions on Graphics (ToG) , 36 (4), 2017. 6, 7

    2017

  23. [30]

    Ro- bust self-calibration of focal lengths from the fundamental matrix

    Viktor Kocur, Daniel Kyselica, and Zuzana Kukelova. Ro- bust self-calibration of focal lengths from the fundamental matrix. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 3

    2024

  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. In Proceedings of the European Conference on Computer Vision (ECCV) ,

  25. [33]

    Instant3d: Fast text-to-3d with sparse-view generation and large reconstruction model

    Jiahao Li, Hao Tan, Kai Zhang, Zexiang Xu, Fujun Luan, Yinghao Xu, Yicong Hong, Kalyan Sunkavalli, Greg Shakhnarovich, and Sai Bi. Instant3d: Fast text-to-3d with sparse-view generation and large reconstruction model. arXiv preprint arXiv:2311.06214, 2023. 1

    2023 arXiv

  26. [34]

    Capturing, reconstructing, and simulating: the UrbanScene3D dataset

    Liqiang Lin, Yilin Liu, Yue Hu, Xingguang Yan, Ke Xie, and Hui Huang. Capturing, reconstructing, and simulating: the UrbanScene3D dataset. In Proceedings of the European Conference on Computer Vision (ECCV), 2022. 6

    2022

  27. [35]

    Pixel-perfect structure-from-motion with featuremetric refinement

    Philipp Lindenberger, Paul-Edouard Sarlin, Viktor Larsson, and Marc Pollefeys. Pixel-perfect structure-from-motion with featuremetric refinement. In Proceedings of the Inter- national Conference on Computer Vision (ICCV), 2021. 16

    2021

  28. [36]

    Matrix3d: Large photogrammetry model all-in-one

    Yuanxun Lu, Jingyang Zhang, Tian Fang, Jean-Daniel Nah- mias, Yanghai Tsin, Long Quan, Xun Cao, Yao Yao, and Shi- wei Li. Matrix3d: Large photogrammetry model all-in-one. arXiv preprint arXiv:2502.07685, 2025. 3

    2025 arXiv

  29. [37]

    Deeper understanding of the homography decomposition for vision-based control

    Ezio Malis and Manuel Vargas. Deeper understanding of the homography decomposition for vision-based control . PhD thesis, Inria, 2007. 4

    2007

  30. [38]

    Robust rotation and translation estimation in multiview reconstruction

    Daniel Martinec and Tomas Pajdla. Robust rotation and translation estimation in multiview reconstruction. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2007. 2, 5, 12

    2007

  31. [39]

    Spring: A high-resolution high- detail dataset and benchmark for scene flow, optical flow and stereo

    Lukas Mehl, Jenny Schmalfuss, Azin Jahedi, Yaroslava Nali- vayko, and Andr ´es Bruhn. Spring: A high-resolution high- detail dataset and benchmark for scene flow, optical flow and stereo. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pa...

    2023

  32. [40]

    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. Communications of the ACM, 65(1), 2021. 7

    2021

  33. [41]

    OpenMVG: Open multiple view geometry

    Pierre Moulon, Pascal Monasse, Romuald Perrot, and Re- naud Marlet. OpenMVG: Open multiple view geometry. In Proceedings of the International Workshop on Reproducible Research in Pattern Recognition, 2017. 2

    2017

  34. [42]

    Instant neural graphics primitives with a mul- tiresolution hash encoding

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a mul- tiresolution hash encoding. ACM Transactions on Graphics (TOG), 41(4), 2022. 8

    2022

  35. [43]

    Robust camera location estimation by convex programming

    Onur Ozyesil and Amit Singer. Robust camera location estimation by convex programming. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2015. 2, 5

    2015

  36. [44]

    Global structure-from-motion re- visited

    Linfei Pan, Daniel Barath, Marc Pollefeys, and Jo- hannes Lutz Sch ¨onberger. Global structure-from-motion re- visited. In Proceedings of the European Conference on Com- puter Vision (ECCV), 2024. 1, 2, 4, 5, 6, 14

    2024

  37. [45]

    Barron, and Ricardo Martin-Brualla

    Keunhong Park, Philipp Henzler, Ben Mildenhall, Jonathan T. Barron, and Ricardo Martin-Brualla. CamP: Camera preconditioning for neural radiance fields. ACM Transactions on Graphics, 2023. 8

    2023

  38. [46]

    PyTorch: An imperative style, high-performance deep learning li- brary

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zem- ing Lin, Natalia Gimelshein, Luca Antiga, et al. PyTorch: An imperative style, high-performance deep learning li- brary. In Advances in Neural Information Processing Sys- ...

    2019

  39. [47]

    DroneDeploy NeRF dataset

    Nicholas Pilkington. DroneDeploy NeRF dataset. https://github.com/nickponline/dd- nerf- dataset, 2022. 6

    2022

  40. [48]

    Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer

    Ren ´e Ranftl, Katrin Lasinger, David Hafner, Konrad Schindler, and Vladlen Koltun. Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer. IEEE transactions on pattern analysis and machine intelligence, 44(3):1623–1637, 2020. 1

    2020

  41. [49]

    GEA optimization for live structureless motion estima- tion

    Antonio L Rodriguez, Pedro E L ´opez-de Teruel, and Alberto Ruiz. GEA optimization for live structureless motion estima- tion. In Proceedings of the IEEE International Conference on Computer Vision Workshops (ICCV Workshops), 2011. 5

    2011

  42. [50]

    Reduced epipolar cost for accelerated incremental sfm

    Antonio L Rodr ´ıguez, Pedro E L´opez-de Teruel, and Alberto Ruiz. Reduced epipolar cost for accelerated incremental sfm. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2011. 5

    2011

  43. [51]

    Nerf for outdoor scene relighting

    Viktor Rudnev, Mohamed Elgharib, William Smith, Lingjie Liu, Vladislav Golyanik, and Christian Theobalt. Nerf for outdoor scene relighting. In Proceedings of the European Conference on Computer Vision (ECCV), 2022. 6

    2022

  44. [52]

    LaMAR: Benchmarking localization and mapping for augmented reality

    Paul-Edouard Sarlin, Mihai Dusmanu, Johannes L Sch¨onberger, Pablo Speciale, Lukas Gruber, Viktor Larsson, Ondrej Miksik, and Marc Pollefeys. LaMAR: Benchmarking localization and mapping for augmented reality. In Proceedings of the European Conference on Computer Vision (ECCV)...

    2022

  45. [53]

    Structure-from-motion revisited

    Johannes Lutz Sch ¨onberger and Jan-Michael Frahm. Structure-from-motion revisited. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2016. 1, 2, 3, 6, 14

    2016

  46. [54]

    A multi-view stereo benchmark with high- resolution images and multi-camera videos

    Thomas Schops, Johannes L Schonberger, Silvano Galliani, Torsten Sattler, Konrad Schindler, Marc Pollefeys, and An- dreas Geiger. A multi-view stereo benchmark with high- resolution images and multi-camera videos. In Proceedings of the IEEE/CVF Conference on Computer Vision an...

    2017

  47. [55]

    Mvdream: Multi-view diffusion for 3d gen- eration

    Yichun Shi, Peng Wang, Jianglong Ye, Mai Long, Kejie Li, and Xiao Yang. Mvdream: Multi-view diffusion for 3d gen- eration. arXiv preprint arXiv:2308.16512, 2023. 1

    2023 arXiv

  48. [56]

    FlowMap: High-quality camera poses, in- trinsics, and depth via gradient descent

    Cameron Smith, David Charatan, Ayush Tewari, and Vin- cent Sitzmann. FlowMap: High-quality camera poses, in- trinsics, and depth via gradient descent. arXiv preprint arXiv:2404.15259, 2024. 2

    2024 arXiv

  49. [57]

    Theia: A fast and scalable structure-from-motion library

    Christopher Sweeney, Tobias Hollerer, and Matthew Turk. Theia: A fast and scalable structure-from-motion library. In Proceedings of the ACM International Conference on Multi- media MM, 2015. 2

    2015

  50. [58]

    Optimizing the viewing graph for structure-from-motion

    Chris Sweeney, Torsten Sattler, Tobias Hollerer, Matthew Turk, and Marc Pollefeys. Optimizing the viewing graph for structure-from-motion. In Proceedings of the International Conference on Computer Vision (ICCV), 2015. 3

    2015

  51. [59]

    Droid-slam: Deep visual slam for monocular, stereo, and rgb-d cameras

    Zachary Teed and Jia Deng. Droid-slam: Deep visual slam for monocular, stereo, and rgb-d cameras. Advances in neu- ral information processing systems, 34:16558–16569, 2021. 1

    2021

  52. [60]

    Bundle adjustment—a modern syn- thesis

    Bill Triggs, Philip F McLauchlan, Richard I Hartley, and Andrew W Fitzgibbon. Bundle adjustment—a modern syn- thesis. In Vision Algorithms: Theory and Practice: In- ternational Workshop on Vision Algorithms Corfu, Greece, September 21–22, 1999 Proceedings, 2000. 2, 4, 5, 14

    1999

  53. [61]

    Mega-NeRF: Scalable construction of large- scale NeRFs for virtual fly-throughs

    Haithem Turki, Deva Ramanan, and Mahadev Satya- narayanan. Mega-NeRF: Scalable construction of large- scale NeRFs for virtual fly-throughs. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 6, 7, 16

    2022

  54. [62]

    VGGSfM: Visual geometry grounded deep structure from motion

    Jianyuan Wang, Nikita Karaev, Christian Rupprecht, and David Novotny. VGGSfM: Visual geometry grounded deep structure from motion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 2

    2024

  55. [63]

    Vggt: Vi- sual geometry grounded transformer

    Jianyuan Wang, Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, and David Novotny. Vggt: Vi- sual geometry grounded transformer. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 5294–5306, 2025. 1, 3

    2025

  56. [64]

    DUSt3R: Geometric 3D vision made easy

    Shuzhe Wang, Vincent Leroy, Yohann Cabon, Boris Chidlovskii, and Jerome Revaud. DUSt3R: Geometric 3D vision made easy. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) ,

  57. [65]

    Tartanair: A dataset to push the limits of visual slam

    Wenshan Wang, Delong Zhu, Xiangwei Wang, Yaoyu Hu, Yuheng Qiu, Chen Wang, Yafei Hu, Ashish Kapoor, and Se- bastian Scherer. Tartanair: A dataset to push the limits of visual slam. In 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 4909–49...

    2020

  58. [66]

    On the distribution of min- ima in intrinsic-metric rotation averaging

    Kyle Wilson and David Bindel. On the distribution of min- ima in intrinsic-metric rotation averaging. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 2

    2020

  59. [67]

    Robust global translations with 1DSfM

    Kyle Wilson and Noah Snavely. Robust global translations with 1DSfM. In Proceedings of the European Conference on Computer Vision (ECCV), 2014. 2, 5

    2014

  60. [68]

    Structured 3d latents for scalable and versatile 3d gen- eration

    Jianfeng Xiang, Zelong Lv, Sicheng Xu, Yu Deng, Ruicheng Wang, Bowen Zhang, Dong Chen, Xin Tong, and Jiaolong Yang. Structured 3d latents for scalable and versatile 3d gen- eration. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 21469–21480, 2025. 1

    2025

  61. [69]

    VR-NeRF: High-fidelity virtualized walkable spaces

    Linning Xu, Vasu Agrawal, William Laney, Tony Garcia, Aayush Bansal, Changil Kim, Samuel Rota Bul `o, Lorenzo Porzi, Peter Kontschieder, Aljaˇz Boˇziˇc, Dahua Lin, Michael Zollh¨ofer, and Christian Richardt. VR-NeRF: High-fidelity virtualized walkable spaces. In SIGGRAPH Asia ...

    2023

  62. [70]

    Fast3r: Towards 3d reconstruction of 1000+ images in one forward pass.arXiv preprint arXiv:2501.13928, 2025

    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.arXiv preprint arXiv:2501.13928, 2025. 3

    2025 arXiv

  63. [71]

    Pointodyssey: A large-scale synthetic dataset for long-term point tracking

    Yang Zheng, Adam W Harley, Bokui Shen, Gordon Wet- zstein, and Leonidas J Guibas. Pointodyssey: A large-scale synthetic dataset for long-term point tracking. In Proceed- ings of the IEEE/CVF International Conference on Com- puter Vision, pages 19855–19865, 2023. 1

    2023

  64. [72]

    On the continuity of rotation representations in neural networks

    Yi Zhou, Connelly Barnes, Jingwan Lu, Jimei Yang, and Hao Li. On the continuity of rotation representations in neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019. 5

    2019

  65. [73]

    Baseline desensitizing in translation averaging

    Bingbing Zhuang, Loong-Fah Cheong, and Gim Hee Lee. Baseline desensitizing in translation averaging. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2018. 2, 5 11 A. More Results Additional speed benchmarking with different hardwa...

    2018

  66. [74]

    both images correspond to that camera; or

  67. [75]

    We estimate the distortion parameters for these cameras one by one

    only one of the images corresponds to that camera, but the distortion parameter of the other image is already es- timated. We estimate the distortion parameters for these cameras one by one. Each time, among the cameras whose distortion have not been estimated, we pick the one...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.