Pith. sign in

REVIEW 4 major objections 6 minor 62 references

GeoFlow: Efficient Driving Video Generation via Geometry-Aligned Priors

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Starting from a geometry-aligned prior instead of Gaussian noise matches 40-step driving-video quality in 8 steps.

desk verdict GeoFlow's central efficiency claim is credible under oracle control signals, but the paper never tests what happens when future poses and boxes are imperfect; that's the main thing to probe in review. read the letter →

arxiv 2608.12203 v1 pith:2R4ISA7A submitted 2026-08-12 cs.CV

Jiazheng Liu , Hang Li , Jiawei Zhang , Jiahe Li , Xiaohan Yu , Shengyin Fan , Jin Zheng , Xiao Bai This is my paper · ORCID
classification cs.CV
keywords drivingvideogenerationflowmatchinggeometry-alignedpriorfew-stepsamplinglatent3Dwarpingmulti-viewgeometryspatially-adaptivenoiseinjectionautonomoussimulation
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

GeoFlow argues that flow-matching and diffusion models waste most of their sampling steps because they start every driving frame from independent Gaussian noise, even though the next few frames are almost determined by known camera motion and scene geometry. Its fix is to start the generative process from a Geometry-Aligned Prior—a coarse prediction of each future frame made by lifting the reference frame into a latent 3D point cloud, warping it to the future camera pose, and rendering it back—with a per-pixel mask that adds noise exactly where the geometry is unreliable. On the paper's benchmark, this shrinks the number of steps needed for state-of-the-art quality from about 40 to about 8, a fivefold inference speedup at equal FVD, and adapting an existing model requires less than 30 H100 GPU-hours of fine-tuning. A sympathetic reader should care because the method changes only the starting distribution, not the network, so the same speedup may transfer to any geometry-conditioned video generator.

What carries the argument

The load-bearing object is the Geometry-Aligned Prior (GAP), a source distribution defined in the latent space of the video autoencoder rather than in RGB space. Construction: (1) estimate metric depth and an uncertainty map for the reference frame; (2) encode the frame into a latent and unproject it into a 3D feature point cloud; (3) apply the relative pose from future control signals to the cloud; (4) render it onto target views with a Z-buffer feature splatting that keeps the nearest point per pixel; and (5) blend the rendered latent with Gaussian noise according to a pixel-wise mask $M = \max(M_{\mathrm{occ}}, M_{\mathrm{dyn}}, M_{\mathrm{unc}})$, where occlusion, dynamic-agent boxes, and depth uncertainty mark where the geometry should be trusted. This mask is what makes the prior stochastic where it is wrong and deterministic where it is right, so the model learns residual refinement instead of regenerating static structure.

What would settle it

Run GeoFlow at 8 steps while adding controlled noise to the future camera poses or randomly jittering a fraction of the 3D boxes used to build the prior, then compare FVD against the unperturbed run. If FVD climbs to the baseline's 8-step level, the gain depends on exact future control knowledge; if it degrades gracefully, the prior is robust to plan errors.

Watch

Extended reading notes

Core claim

The central claim is that the starting distribution of a flow-matching model, not the network or the solver, is the main obstacle to few-step driving-video generation. GeoFlow builds a Geometry-Aligned Prior (GAP) distribution by unprojecting the encoded reference frame into a 3D feature point cloud with metric depth, transforming that cloud by the known future camera poses, and splatting it back into latent maps for the new frames. Because warping is imperfect, a continuous per-pixel reliability mask, formed from occlusion, dynamic-object, and depth-uncertainty signals, controls a linear blend between the warped latent and Gaussian noise; reliable static regions keep the warped features while uncertain regions fall back to noise. Fine-tuning the base model on this new source distribution with the same flow-matching objective gives, on the paper's reported benchmark, an FVD of 38.6 at 8 sampling steps, better than the baseline's 38.8 at 40 steps, with FID also lower at 8.3 versus 14.7. The paper interprets this as straightening and shortening the transport path between source and data distributions, turning generation from full scene synthesis into residual refinement of a geometrically grounded prediction.

Load-bearing premise

The prior only helps because the future camera poses and 3D object boxes used to warp the reference frame are known exactly and match the content being generated; if a system plans a different trajectory than the one rendered, the warped start is systematically misaligned.

Editorial extensions

If this is right

  • At 8 steps GeoFlow's FVD of 38.6 beats the baseline's 40-step FVD of 38.8, making the promised 5x step reduction an equality-of-quality result rather than just a speed claim.
  • Fine-tuning saturates at roughly 10,000 iterations and costs less than 30 H100 GPU-hours, so existing pretrained driving world models can adopt the prior without distillation or retraining.
  • The improvement holds across base architectures with different latent video autoencoders and across a stronger baseline, with few-step FVD dropping 38.6 to 62.5 percent in the generality tests.
  • A 16-frame clip at high quality is generated about 4.2x faster than the baseline's 40-step run, because geometric reconstruction and feature rendering add roughly 6 percent of the per-clip latency.
  • At the high-step end the prior does not hurt: GeoFlow's 40-step FVD of 34.0 is still below the baseline's 38.8.
  • The paper's robustness test shows swapping the metric depth estimator in a zero-shot manner preserves the few-step gains, indicating the method does not depend on one specific depth model.

Reading between the lines

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

  • A natural extension is to apply the same warped-prior strategy to other geometry-dominated generation tasks such as novel-view synthesis or camera-only simulation, where the reliability mask would need an analogous dynamic-object signal; the paper's depth-model swap suggests the mask absorbs much of the geometry error.
  • The paper assumes oracle future poses and boxes; perturbing those controls at inference would reveal whether the gain is genuine geometric alignment or simply conditioning on the target, and I would expect the 8-step advantage to shrink in proportion to the perturbation.
  • GAP should stack with distillation and fast ODE solvers rather than compete with them: a straighter, shorter trajectory is exactly the input those methods need to reach one- or two-step generation.
  • The authors' own ablations show that a deterministic warped prior causes error accumulation at high step counts, so the noise mask is not a minor detail but the component that makes the geometric start usable as a distribution.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes GeoFlow, a plug-and-play framework for efficient driving-video generation that replaces the standard Gaussian source distribution in latent flow matching with a Geometry-Aligned Prior (GAP). The prior is constructed by unprojecting the reference-frame VAE latent into a 3D point cloud using metric depth, warping it with the future ego-pose, splatting it into the target view, and then injecting noise with a spatially adaptive mask that combines occlusion, dynamic-object, and depth-uncertainty maps. The baseline flow-matching model is fine-tuned for a few thousand iterations with the standard conditional flow-matching objective but with training trajectories interpolating between this prior and the target latent. Experiments on NuScenes report that GeoFlow achieves FVD 38.6 in 8 steps versus FVD 38.8 for the 40-step baseline, with similar benefits across three base architectures and under 30 H100 GPU hours of fine-tuning.

Significance. If the claims hold, GeoFlow's source-distribution perspective is a useful complement to sampler- and distillation-based acceleration, and the reported cross-architecture gains (Table 1b) with a parameter-free prior construction are notable. The controlled comparison against OpenDWM, in which only the source distribution is changed, is the right experimental design for the central claim, and the paper is careful to ablate the noise-injection components. The prior is built from control signals and an external depth estimator rather than the model's own outputs, so there is no obvious circularity in the training objective. The main reservations concern the breadth of the efficiency claim and the lack of uncertainty quantification in the headline numbers.

major comments (4)
  1. [§4.3, Table 2] The headline claim that 'our model with only 8 sampling steps attains an FVD of 38.6, surpassing the Baseline with 40 steps (FVD 38.8). This represents a 5× steps reduction without compromising visual quality' is only partly supported by the reported numbers. At 8 steps GeoFlow has FID 8.3, which is substantially worse than the baseline at 40 steps (FID 6.8), so image-level quality is compromised even if FVD improves by 0.2. Moreover, the 38.6 vs 38.8 FVD gap is below any reasonable noise floor; the paper reports no confidence intervals or multiple-seed results. Please provide error bars or repeated-seed statistics and either drop or carefully qualify the 'without compromising visual quality' wording.
  2. [§3.3–3.4, §4.3] The efficiency claim is only demonstrated under oracle control signals. The prior is built from T_rel, 'derived from the ego-vehicle's control commands or trajectory planning' (Section 3.3), and from 3D bounding boxes that define M_dyn (Section 3.4). If the planned trajectory is not exactly realized, or boxes are misdetected, the warped prior is systematically misaligned and the model must correct a long, curved residual path—precisely what the method claims to remove. The paper evaluates robustness to different depth estimators (Table 5) but not to perturbations of T_rel or the boxes. Please add a sensitivity study (e.g., pose noise of 0.5–1 m / 2–3 degrees and box jitter) and state the operating regime of the 5× speedup.
  3. [§4.2, Table 1a] The caption 'GeoFlow achieves SOTA quality with significantly less steps' is not fully supported by Table 1a. GeoFlow has the best FVD (32.5) but a worse FID (6.8) than UniMLVG (5.8), and the comparison mixes different step counts, reference-frame numbers (UniMLVG uses three reference frames, as indicated by the dagger), and model backbones. Please qualify the claim as 'best FVD among the compared methods at substantially fewer steps' and, if possible, re-run the strongest competitors under matched conditions.
  4. [Table 2] GeoFlow's FVD is non-monotonic in the number of steps: it improves from 38.6 at 8 steps to 32.5 at 15 steps, then degrades to 32.6 at 20 steps and 34.0 at 40 steps. The paper explains why global uniform noise degrades at high steps but does not explain why its adaptive-injection variant also degrades at 40 steps. Please discuss or investigate this behavior, as it bears on the stability of the prior at high step counts.
minor comments (6)
  1. [§4.3] 'Within the first thousands iterations' followed by 'only hundreds of training steps' is inconsistent; please clarify the exact iteration counts reported in Fig. 5.
  2. [Table 4] The first row with no check marks is ambiguous; specify whether it corresponds to deterministic warping with no noise injection or to global uniform noise, and align it with Fig. 6a.
  3. [Fig. 3 caption] 'An Spatially-Adaptive Noise Injection' should be 'a Spatially-Adaptive Noise Injection'.
  4. [§3.1, Eq. (8)] The model is written as Fθ(xt,t,C), but the task is conditional on both I_ref and C; clarify how I_ref enters the network (e.g., through condition tokens or through the initialization xt).
  5. [§4.3, Table 3] The 4.2× speedup is derived from 5-step latency, while the headline comparison is at 8 steps; provide a wall-clock breakdown for the 8-step setting to make the 4.2× figure directly verifiable.
  6. [Table 1a] DriveDreamer-2 has no step count listed; state whether the step count is unknown or not applicable.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the Geometry-Aligned Prior is built from external depth estimates and given control signals, the training objective is the standard flow-matching regression, and the reported gains are empirical comparisons against external benchmarks.

full rationale

The derivation chain is self-contained and does not reduce to its inputs. The Geometry-Aligned Prior is constructed in Eqs. (2), (5), and (6) from the reference frame, an external metric depth estimator, Z-buffer rendering information, and explicitly given future control signals (relative pose T_rel and 3D bounding boxes). No method parameter is fitted to the evaluation metrics FVD or FID; the flow-matching objective in Eq. (8) is the standard conditional flow-matching loss applied to this new source distribution, and the claimed efficiency gains are empirical results measured against an external baseline (OpenDWM) with external benchmarks. The self-citations in the reference list (e.g., Sparsesurf [12], DNGaussian [20,21], GeoSVR [22], Cor-GS [54], EVE3D [55]) are prior geometry/surface-reconstruction works used for background context; none is invoked as a load-bearing premise, uniqueness theorem, or source of the central efficiency claim. The skeptical concern about robustness to mis-specified control signals is a real boundary condition on the claimed efficiency regime, but it is not an internal circularity: the paper explicitly states that T_rel is 'derived from the ego-vehicle’s control commands or trajectory planning' and that dynamic-object masks come from given 3D boxes, which is a stated assumption rather than a fitted quantity. A missing robustness study does not make the derivation circular. Honest non-finding: no significant circularity.

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

The method introduces no new fitted numerical parameters: the prior is constructed from the reference frame, external depth estimates, camera poses, and 3D boxes, with a hand-designed mask. The main load-bearing external inputs are the future control signals and the depth model, both treated as given.

assumptions (4)
  • domain assumption Future camera poses and 3D object boxes used for warping are accurate and available at inference time.
    Eq. (2) transforms the reference point cloud with T_rel derived from the planned trajectory, and M_dyn in Section 3.4 is projected from supplied boxes; the entire GAP construction collapses if these signals are wrong.
  • domain assumption The metric depth model's estimates and uncertainty map are reliable enough to guide the noise mask.
    Section 3.4 uses the normalized uncertainty map M_unc as one of three mask components; miscalibration would inject noise into reliable regions or leave artifacts unmasked.
  • domain assumption VAE latent features can be meaningfully unprojected into 3D and splatted back into 2D.
    Section 3.3 encodes the reference frame to latent Z_ref and builds a 3D feature point cloud from it; if latent channels do not respect 3D geometry, the warped prior is corrupted before noise injection.
  • standard math Conditional flow matching with a non-Gaussian source distribution still learns the correct vector field.
    The paper relies on the flow matching framework of Lipman et al. [24] in Eq. (1) and Eq. (8), which supports arbitrary source-target pairs; this is a standard result and not re-derived.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GeoFlow: Efficient Driving Video Generation via Geometry-Aligned Priors." pith.science (2026). https://pith.science/paper/2R4ISA7A

@misc{pith2026260812203,
  author       = {Pith},
  title        = {Pith review of: GeoFlow: Efficient Driving Video Generation via Geometry-Aligned Priors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2R4ISA7A}},
  note         = {Machine review of arXiv:2608.12203}
}
read the original abstract

Generative models like Diffusion Models and Flow Matching have demonstrated remarkable capabilities in synthesizing high-fidelity driving videos, but are severely constrained by high inference latency due to the requirement of extensive sampling steps. We argue that this inefficiency stems from the prevailing reliance on a standard Gaussian source distribution, where consecutive frames are initialized as independent Gaussian noise. This paradigm disregards the rich spatiotemporal correlations inherent in driving videos, compelling the model to regenerate deterministic scene structures existing in previous frames from noise, which is both computationally redundant and prone to geometric inconsistency. To address this problem, we propose GeoFlow, a novel framework designed to achieve efficient driving video generation by harnessing explicit geometric priors. Instead of sampling from standard Gaussian noise, we leverage multi-view geometry and spatially-adaptive noise injection to construct a Geometry-Aligned Prior (GAP) distribution as starting point. This initialization bridges the gap between source distribution and data distribution, yielding a significantly straighter and shorter sampling trajectory. Extensive experiments demonstrate that GeoFlow can achieve remarkable efficiency of both training and inference: merely several hours of fine-tuning on baseline models can significantly boost few-step generation quality, while fully converged training drastically reduces number of inference steps required for state-of-the-art video generation.

Figures

Figures reproduced from arXiv: 2608.12203 by the authors.

Figure 1
Figure 1. Efficient driving video generation with structure-aware priors. (a) [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Comparison between Vanilla Driving Video Generation and GeoFlow. The former uses standard Gaussian as source distribution, while GeoFlow leverages multi￾view geometry to build an informative source distribution in latent space. tion environments [45, 47], serving as interactive platforms for the closed-loop evaluation and training of autonomous driving algorithms. However, this high fidelity comes at a steep computa… view at source ↗
Figure 3
Figure 3. Schematic of the proposed GeoFlow framework. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (4 more)
Figure 3
Figure 3. Figure 3: Given the relative pose transformation Trel derived from the ego-vehicle’s control commands or trajectory planning, we transform the point cloud to the target coordinate system: \mathcal {P}_{target} = \mathbf {T}_{rel} \cdot \mathcal {P}_{ref}. (2) We then render Ptar…
Figure 4
Figure 4. Figure 4: 5-step generation visual quality comparison between GeoFlow and Baseline model. All these videos are generated with single reference frame. verse architectures. We validate this by experimenting with various models within the OpenDWM codebase, including those equipped …
Figure 5
Figure 5. Figure 5: Training Convergence. Inference Efficiency. We further evaluate the inference efficiency. As demonstrated in Sec. 3.5, although our method employs a shorter autoregressive chunk size (L = 6) compared to the baseline (L = 19) to ensure geometric validity, the total gene…
Figure 6
Figure 6. Figure 6: Qualitative and quantitative ablations. (a) [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 22 canonical work pages

  1. [1]

    Opendwm: Open driving world models (2025), https://github.com/SenseTime- FVG/OpenDWM

    work page 2025

  2. [2]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Caesar, H., Bankiti, V., Lang, A.H., Vora, S., Liong, V.E., Xu, Q., Krishnan, A., Pan, Y., Baldan, G., Beijbom, O.: nuscenes: A multimodal dataset for autonomous driving. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 11621–11631 (2020)

    2020

  3. [3]

    arXiv preprint arXiv:2505.22421 (2025)

    Chen, A., Zheng, W., Wang, Y., Zhang, X., Zhan, K., Jia, P., Keutzer, K., Zhang, S.: Geodrive: 3d geometry-informed driving world model with precise action con- trol. arXiv preprint arXiv:2505.22421 (2025)

    arXiv 2025

  4. [4]

    arXiv preprint arXiv:2412.04842 (2024)

    Chen, R., Wu, Z., Liu, Y., Guo, Y., Ni, J., Xia, H., Xia, S.: Unimlvg: Unified frame- work for multi-view long video generation with comprehensive control capabilities for autonomous driving. arXiv preprint arXiv:2412.04842 (2024)

    arXiv 2024

  5. [5]

    arXiv preprint arXiv:2502.03500 (2025) 16 J

    Cohen,E.,Achituve,I.,Diamant,I.,Netzer,A.,Habi,H.V.:Efficientimagerestora- tion via latent consistency flow matching. arXiv preprint arXiv:2502.03500 (2025) 16 J. Liu et al

    arXiv 2025

  6. [6]

    arXiv preprint arXiv:2512.23180 (2025)

    Deng,T.,Chen,X.,Chen,Y.,Chen,Q.,Xu,Y.,Yang,L.,Xu,L.,Zhang,Y.,Zhang, B., Huang, W., et al.: Gaussiandwm: 3d gaussian driving world model for unified scene understanding and multi-modal generation. arXiv preprint arXiv:2512.23180 (2025)

    arXiv 2025

  7. [7]

    Advances in Neural Information Processing Systems 38, 32551–32576 (2026)

    Gao, H., Chen, S., Jiang, B., Liao, B., Shi, Y., Guo, X., Pu, Y., Li, X., Liu, W., Zhang, Q., et al.: Rad: Training an end-to-end driving policy via large-scale 3dgs- based reinforcement learning. Advances in Neural Information Processing Systems 38, 32551–32576 (2026)

    work page 2026

  8. [8]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Gao, R., Chen, K., Xiao, B., Hong, L., Li, Z., Xu, Q.: Magicdrive-v2: High- resolution long video generation for autonomous driving with adaptive control. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 28135–28144 (2025)

    work page 2025

Show all 62 references
  1. [9]

    arXiv preprint arXiv:2310.02601 (2023)

    Gao,R.,Chen,K.,Xie,E.,Hong,L.,Li,Z.,Yeung,D.Y.,Xu,Q.:Magicdrive:Street view generation with diverse 3d geometry control. arXiv preprint arXiv:2310.02601 (2023)

    2023 arXiv

  2. [10]

    arXiv preprint arXiv:2503.18108 (2025)

    Ge, J., Liu, Z., Fan, L., Jiang, Y., Su, J., Li, Y., Zhang, Z., Chen, S.: Unraveling the effects of synthetic data on end-to-end autonomous driving. arXiv preprint arXiv:2503.18108 (2025)

    2025 arXiv

  3. [11]

    Communications of the ACM63(11), 139–144 (2020)

    Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y.: Generative adversarial networks. Communications of the ACM63(11), 139–144 (2020)

    2020

  4. [12]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Gu, M., Zhang, J., Li, J., Yu, X., Luo, H., Zheng, J., Bai, X.: Sparsesurf: Sparse- view 3d gaussian splatting for surface reconstruction. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 40, pp. 4311–4319 (2026)

    2026

  5. [13]

    In: Pro- ceedings of the AAAI Conference on Artificial Intelligence

    Han, Z., Zhang, B., Zhang, L., Feng, S., Lin, K., Liang, G., Ye, Y., et al.: Asyncdsb: Schedule-asynchronous diffusion schrödinger bridge for image inpainting. In: Pro- ceedings of the AAAI Conference on Artificial Intelligence. vol. 39, pp. 3374–3382 (2025)

    2025

  6. [14]

    He, H., Xu, Y., Guo, Y., Wetzstein, G., Dai, B., Li, H., Yang, C.: Cameractrl: En- ablingcameracontrolfortext-to-videogeneration.arXivpreprintarXiv:2404.02101 (2024)

    2024 arXiv

  7. [15]

    Advances in neural information processing systems33, 6840–6851 (2020)

    Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models. Advances in neural information processing systems33, 6840–6851 (2020)

    2020

  8. [16]

    Hou,C.,Chen,Z.:Training-freecameracontrolforvideogeneration.arXivpreprint arXiv:2406.10126 (2024)

    2024 arXiv

  9. [17]

    arXiv preprint arXiv:2506.08009 (2025)

    Huang, X., Li, Z., He, G., Zhou, M., Shechtman, E.: Self forcing: Bridging the train-test gap in autoregressive video diffusion. arXiv preprint arXiv:2506.08009 (2025)

    2025 arXiv

  10. [18]

    arXiv preprint arXiv:2509.13414 (2025)

    Keetha, N., Müller, N., Schönberger, J., Porzi, L., Zhang, Y., Fischer, T., Knapitsch, A., Zauss, D., Weber, E., Antunes, N., et al.: Mapanything: Univer- sal feed-forward metric 3d reconstruction. arXiv preprint arXiv:2509.13414 (2025)

    2025 arXiv

  11. [19]

    ACM Trans

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G.: 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph.42(4), 139–1 (2023)

    2023

  12. [20]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Li, J., Zhang, J., Bai, X., Zheng, J., Ning, X., Zhou, J., Gu, L.: Dngaussian: Opti- mizing sparse-view 3d gaussian radiance fields with global-local depth normaliza- tion. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 20775–20785 (2024)

    2024

  13. [21]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2026) GeoFlow 17

    Li, J., Zhang, J., Yu, X., Bai, X., Zheng, J., Ning, X., Gu, L.: Dngaussian++: Improving sparse-view gaussian radiance fields with depth normalization. IEEE Transactions on Pattern Analysis and Machine Intelligence (2026) GeoFlow 17

    2026

  14. [22]

    Advances in Neural Information Processing Systems38, 108809–108837 (2026)

    Li, J., Zhang, J., Zhang, Y., Bai, X., Zheng, J., Yu, X., Gu, L.: Geosvr: Taming sparse voxels for geometrically accurate surface reconstruction. Advances in Neural Information Processing Systems38, 108809–108837 (2026)

    2026

  15. [23]

    arXiv preprint arXiv:2511.10647 (2025)

    Lin, H., Chen, S., Liew, J., Chen, D.Y., Li, Z., Shi, G., Feng, J., Kang, B.: Depth anything 3: Recovering the visual space from any views. arXiv preprint arXiv:2511.10647 (2025)

    2025 arXiv

  16. [24]

    arXiv preprint arXiv:2210.02747 (2022)

    Lipman, Y., Chen, R.T., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. arXiv preprint arXiv:2210.02747 (2022)

    2022 arXiv

  17. [25]

    arXiv preprint arXiv:2503.06364 (2025)

    Liu, C., Ritschel, T.: Generative video bi-flow. arXiv preprint arXiv:2503.06364 (2025)

    2025 arXiv

  18. [26]

    arXiv preprint arXiv:2209.03003 (2022)

    Liu, X., Gong, C., Liu, Q.: Flow straight and fast: Learning to generate and transfer data with rectified flow. arXiv preprint arXiv:2209.03003 (2022)

    2022 arXiv

  19. [27]

    Advances in neural information processing systems35, 5775–5787 (2022)

    Lu, C., Zhou, Y., Bao, F., Chen, J., Li, C., Zhu, J.: Dpm-solver: A fast ode solver for diffusion probabilistic model sampling in around 10 steps. Advances in neural information processing systems35, 5775–5787 (2022)

    2022

  20. [28]

    In: European conference on computer vi- sion

    Lugmayr, A., Danelljan, M., Van Gool, L., Timofte, R.: Srflow: Learning the super- resolution space with normalizing flow. In: European conference on computer vi- sion. pp. 715–732. Springer (2020)

    2020

  21. [29]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Mao, X., Jiang, Z., Wang, F.Y., Zhang, J., Chen, H., Chi, M., Wang, Y., Luo, W.: Osv: One step is enough for high-quality image to video generation. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 12585–12594 (2025)

    2025

  22. [30]

    arXiv preprint arXiv:2410.02423 (2024)

    Martin, S., Gagneux, A., Hagemann, P., Steidl, G.: Pnp-flow: Plug-and-play image restoration with flow matching. arXiv preprint arXiv:2410.02423 (2024)

    2024 arXiv

  23. [31]

    arXiv preprint arXiv:2409.04003 (2024)

    Mei, J., Hu, T., Yang, X., Wen, L., Yang, Y., Wei, T., Ma, Y., Dou, M., Shi, B., Liu, Y.: Dreamforge: Motion-aware autoregressive video generation for multi-view driving scenes. arXiv preprint arXiv:2409.04003 (2024)

    2024 arXiv

  24. [32]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Ni, C., Zhao, G., Wang, X., Zhu, Z., Qin, W., Huang, G., Liu, C., Chen, Y., Wang, Y., Zhang, X., et al.: Recondreamer: Crafting world models for driving scene reconstruction via online restoration. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. ...

    2025

  25. [33]

    In: Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence

    Peng, J., Li, M., Wang, H.: Stabilizing holistic semantics in diffusion bridge for im- age inpainting. In: Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence. pp. 1756–1764 (2025)

    2025

  26. [34]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Qin, H., Luo, W., Wang, L., Zheng, D., Chen, J., Yang, M., Li, B., Hu, W.: Re- versing flow for image restoration. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 7545–7558 (2025)

    2025

  27. [35]

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

    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: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 6121–61...

    2025

  28. [36]

    arXiv preprint arXiv:2202.00512 (2022)

    Salimans, T., Ho, J.: Progressive distillation for fast sampling of diffusion models. arXiv preprint arXiv:2202.00512 (2022)

    2022 arXiv

  29. [37]

    arXiv preprint arXiv:2010.02502 (2020)

    Song, J., Meng, C., Ermon, S.: Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502 (2020)

    2020 arXiv

  30. [38]

    Song, Y., Dhariwal, P., Chen, M., Sutskever, I.: Consistency models (2023)

    2023

  31. [39]

    In: European conference on computer vision

    Wang, C., Zhu, Y., Yuan, C.: Diverse image inpainting with normalizing flow. In: European conference on computer vision. pp. 53–69. Springer (2022)

    2022

  32. [40]

    arXiv preprint arXiv:2510.23116 (2025) 18 J

    Wang, H., Zhang, J., Chen, H., Guo, H., Wang, D., Ma, J., Du, B.: Residual dif- fusion bridge model for image restoration. arXiv preprint arXiv:2510.23116 (2025) 18 J. Liu et al

    2025 arXiv

  33. [41]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Wang, J., Chen, M., Karaev, N., Vedaldi, A., Rupprecht, C., Novotny, D.: Vggt: Visual geometry grounded transformer. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 5294–5306 (2025)

    2025

  34. [42]

    arXiv preprint arXiv:2312.09109 (2023)

    Wang, X., Zhang, S., Zhang, H., Liu, Y., Zhang, Y., Gao, C., Sang, N.: Videolcm: Video latent consistency model. arXiv preprint arXiv:2312.09109 (2023)

    2023 arXiv

  35. [43]

    In: European conference on computer vision

    Wang, X., Zhu, Z., Huang, G., Chen, X., Zhu, J., Lu, J.: Drivedreamer: Towards real-world-drive world models for autonomous driving. In: European conference on computer vision. pp. 55–72. Springer (2024)

    2024

  36. [44]

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

    Wang, Y., He, J., Fan, L., Li, H., Chen, Y., Zhang, Z.: Driving into the future: Mul- tiview visual forecasting and planning with world model for autonomous driving. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 14749–14759 (2024)

    2024

  37. [45]

    In: Proceed- ings of the Computer Vision and Pattern Recognition Conference

    Yan, T., Wu, D., Han, W., Jiang, J., Zhou, X., Zhan, K., Xu, C.z., Shen, J.: Driv- ingsphere: Building a high-fidelity 4d world for closed-loop simulation. In: Proceed- ings of the Computer Vision and Pattern Recognition Conference. pp. 27531–27541 (2025)

    2025

  38. [46]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Yang, L., Kang, B., Huang, Z., Xu, X., Feng, J., Zhao, H.: Depth anything: Un- leashing the power of large-scale unlabeled data. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10371–10381 (2024)

    2024

  39. [47]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Yang,X.,Wen,L.,Wei,T.,Ma,Y.,Mei,J.,Li,X.,Lei,W.,Fu,D.,Cai,P.,Dou,M., et al.: Drivearena: A closed-loop generative simulation platform for autonomous driving. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 26933–26943 (2025)

    2025

  40. [48]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Yin, T., Gharbi, M., Zhang, R., Shechtman, E., Durand, F., Freeman, W.T., Park, T.: One-step diffusion with distribution matching distillation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 6613– 6623 (2024)

    2024

  41. [49]

    In: Pro- ceedings of the Computer Vision and Pattern Recognition Conference

    Yin, T., Zhang, Q., Zhang, R., Freeman, W.T., Durand, F., Shechtman, E., Huang, X.: From slow bidirectional to fast autoregressive video diffusion models. In: Pro- ceedings of the Computer Vision and Pattern Recognition Conference. pp. 22963– 22974 (2025)

    2025

  42. [50]

    arXiv preprint arXiv:2409.02048 (2024)

    Yu, W., Xing, J., Yuan, L., Hu, W., Li, X., Huang, Z., Gao, X., Wong, T.T., Shan, Y., Tian, Y.: Viewcrafter: Taming video diffusion models for high-fidelity novel view synthesis. arXiv preprint arXiv:2409.02048 (2024)

    2024 arXiv

  43. [51]

    In: ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

    Yue, C., Peng, Z., Ma, J., Zhang, D.: Enhanced control for diffusion bridge in image restoration. In: ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). pp. 1–5. IEEE (2025)

    2025

  44. [52]

    Advances in Neural Information Processing Systems 36, 13294–13307 (2023)

    Yue, Z., Wang, J., Loy, C.C.: Resshift: Efficient diffusion model for image super- resolution by residual shifting. Advances in Neural Information Processing Systems 36, 13294–13307 (2023)

    2023

  45. [53]

    In: ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

    Yun, J.H., Kim, S.B., Lee, S.W.: Flowhigh: Towards efficient and high-quality audio super-resolution with single-step flow matching. In: ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). pp. 1–5. IEEE (2025)

    2025

  46. [54]

    In: European conference on computer vision

    Zhang, J., Li, J., Yu, X., Huang, L., Gu, L., Zheng, J., Bai, X.: Cor-gs: sparse-view 3d gaussian splatting via co-regularization. In: European conference on computer vision. pp. 335–352. Springer (2024)

    2024

  47. [55]

    Advances in Neural Information Processing Systems38, 67934–67960 (2026) GeoFlow 19

    Zhang, J., Zhang, Y., Tosi, F., Gu, M., Li, J., Yu, X., Zheng, J., Bai, X., Poggi, M.: Eve3d: Elevating vision models for enhanced 3d surface reconstruction via gaussian splatting. Advances in Neural Information Processing Systems38, 67934–67960 (2026) GeoFlow 19

    2026

  48. [56]

    Advances in Neural Information Processing Systems37, 103599–103618 (2024)

    Zhang, Z., Li, Y., Wu, Y., Kag, A., Skorokhodov, I., Menapace, W., Siarohin, A., Cao, J., Metaxas, D., Tulyakov, S., et al.: Sf-v: Single forward video generation model. Advances in Neural Information Processing Systems37, 103599–103618 (2024)

    2024

  49. [57]

    In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision

    Zhang, Z., Chen, D., Liao, J.: I2v3d: Controllable image-to-video generation with 3d guidance. In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision. pp. 13360–13371 (2025)

    2025

  50. [58]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Zhao, G., Ni, C., Wang, X., Zhu, Z., Zhang, X., Wang, Y., Huang, G., Chen, X., Wang, B., Zhang, Y., et al.: Drivedreamer4d: World models are effective data machines for 4d driving scene representation. In: Proceedings of the Computer Vision and Pattern Recognition Conference. ...

    2025

  51. [59]

    arXiv preprint arXiv:2503.18438 (2025)

    Zhao, G., Wang, X., Ni, C., Zhu, Z., Qin, W., Huang, G., Wang, X.: Recon- dreamer++: Harmonizing generative and reconstructive models for driving scene representation. arXiv preprint arXiv:2503.18438 (2025)

    2025 arXiv

  52. [60]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Zhao, G., Wang, X., Zhu, Z., Chen, X., Huang, G., Bao, X., Wang, X.: Drivedreamer-2: Llm-enhanced world models for diverse driving video generation. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 39, pp. 10412–10420 (2025)

    2025

  53. [61]

    Advances in Neural Information Processing Systems36, 49842–49869 (2023)

    Zhao, W., Bai, L., Rao, Y., Zhou, J., Lu, J.: Unipc: A unified predictor-corrector framework for fast sampling of diffusion models. Advances in Neural Information Processing Systems36, 49842–49869 (2023)

    2023

  54. [62]

    arXiv preprint arXiv:2309.16948 (2023)

    Zhou, L., Lou, A., Khanna, S., Ermon, S.: Denoising diffusion bridge models. arXiv preprint arXiv:2309.16948 (2023)

    2023 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.