DriveWeaver: Point-Conditioned Video Inpainting for Controllable Vehicle Insertion in Autonomous Driving Simulation

Junzhe Jiang; Li Zhang; Zijie Pan; Zipei Ma

REVIEW 2 major objections 1 minor 65 references

DriveWeaver: Point-Conditioned Video Inpainting for Controllable Vehicle Insertion in Autonomous Driving Simulation

T0 review · 2 major / 1 minor · reviewed 2026-07-03 · grok-4.3

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

Pith's one-line read DriveWeaver inserts vehicles into driving videos through point-cloud-conditioned inpainting to achieve seamless blending and temporal consistency without pre-reconstructed 3D assets.

desk verdict DriveWeaver proposes point-cloud conditioned video inpainting plus a global-to-local hierarchy for vehicle insertion in driving sims, which logically sidesteps 3D asset lighting issues, but the abstract supplies zero numbers or ablations to check whether it actually works. read the letter →

arxiv 2606.31918 v2 pith:O4PR7Y4J submitted 2026-06-30 cs.CV

Junzhe Jiang , Zipei Ma , Zijie Pan , Li Zhang This is my paper · ORCID

classification cs.CV

keywords videoinpaintingvehicleinsertionautonomousdrivingsimulationpointcloudconditioning3DGaussianrepresentationsceneaugmentationtemporalconsistency

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

The paper proposes DriveWeaver to insert vehicles along predefined trajectories into autonomous driving simulation scenes. Existing approaches depend on limited manually created 3D models that often produce lighting mismatches with the background. DriveWeaver instead masks the insertion region and runs video inpainting guided by the vehicle's point cloud data. A global-to-local hierarchical strategy preserves vehicle identity over extended sequences. The output is converted to explicit 3D Gaussian representations for real-time rendering. This setup supports larger-scale generation of varied test scenes.

What carries the argument

Point-cloud-conditioned video inpainting using a global-to-local hierarchical strategy that processes masked regions frame by frame while enforcing identity consistency.

What would settle it

A long video sequence in which the inpainted vehicle exhibits noticeable changes in color, shape, or lighting despite fixed point cloud input and consistent masking.

Watch

Extended reading notes

Core claim

For a masked target insertion area, DriveWeaver performs video inpainting conditioned on vehicle point clouds to generate high-quality, temporally consistent vehicles. This video-inpainting-based approach ensures seamless blending between the foreground and background, while the readily available point cloud conditions enable superior generalization. A global-to-local hierarchical inpainting strategy maintains consistent vehicle identity and appearance across long sequences, and an urban reconstruction pipeline extracts explicit 3D Gaussian representations of the inserted vehicles to enable real-time rendering.

Load-bearing premise

Point cloud data supplies enough detail to keep the inserted vehicle's appearance and identity stable across many frames without visible drift or lighting shifts.

Editorial extensions

If this is right

Outperforms baselines in visual realism and geometric consistency across diverse datasets.
Enables scalable augmentation of driving scenes without reliance on manually curated 3D assets.
Supports real-time rendering of inserted vehicles through extracted 3D Gaussian models.
Facilitates creation of corner cases for testing autonomous driving models at larger volumes.

Reading between the lines

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

The method could extend to inserting other dynamic objects if suitable point cloud sources become available.
Simulation pipelines might reduce dependence on offline asset libraries by adopting similar conditioning approaches.
Real-time rendering gains could allow interactive scenario editing during model evaluation.

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, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

read the letter

The core idea is to mask an insertion spot in a driving video, feed in a vehicle point cloud as conditioning, run inpainting to generate the vehicle, use a hierarchical global-to-local pass to hold identity over long clips, and then run an urban reconstruction step to turn the result into explicit 3D Gaussians for rendering. This replaces reliance on curated 3D assets that often mismatch lighting and don't scale.

What stands out is the direct use of point clouds for structure and blending, which standard 2D inpainting lacks, and the hierarchy that targets identity drift without extra learned modules. The pipeline is internally consistent: point clouds give explicit 3D cues, inpainting handles seamless integration, and the post-processing step is presented as a utility rather than the main learned component.

The clear limitation is that the abstract asserts outperformance on visual realism and geometric consistency across datasets but reports no metrics, no baselines with numbers, no ablations, and no dataset details. Without those, the claims about superior generalization and long-term consistency cannot be checked. The assumption that the reconstruction pipeline reliably produces usable Gaussians is also untested in the given text.

This is aimed at researchers building scalable scene augmentation for autonomous driving simulation. A reader working on corner-case generation would find the framing useful if the experiments hold. The work deserves peer review so the quantitative results and implementation can be examined; the construction itself has no obvious internal contradictions.

Referee Report

2 major / 1 minor

Summary. The paper proposes DriveWeaver, a framework for controllable vehicle insertion into autonomous driving scenes. For a masked insertion area, it performs point-cloud-conditioned video inpainting to produce temporally consistent vehicles with seamless foreground-background blending. A global-to-local hierarchical inpainting strategy is introduced to maintain vehicle identity over long sequences, and an urban reconstruction pipeline extracts explicit 3D Gaussian representations of the inserted vehicles to support real-time rendering. The abstract claims that extensive experiments across diverse datasets show outperformance over baselines in visual realism and geometric consistency.

Significance. If the empirical claims hold, the work could meaningfully advance scalable scene augmentation for autonomous driving simulation by replacing reliance on limited pre-reconstructed 3D assets with a point-cloud-conditioned inpainting pipeline that improves generalization and blending. The hierarchical strategy and 3D Gaussian extraction address practical needs for long-horizon consistency and real-time use.

major comments (2)

[Abstract] Abstract: The central claim that 'extensive experiments across diverse datasets demonstrate that our method outperforms existing baselines in visual realism and geometric consistency' is unsupported by any quantitative metrics, tables, figures, error bars, dataset specifications, or ablation studies in the provided manuscript text. This absence prevents evaluation of the primary assertion of superiority and is load-bearing for the paper's contribution.
[Abstract] Abstract: The global-to-local hierarchical inpainting strategy is asserted to ensure 'consistent identity and appearance of the inserted vehicles' across long sequences, yet no implementation details, loss formulations, or quantitative evaluation of identity preservation (e.g., appearance consistency metrics over sequence length) are supplied, leaving the weakest assumption unverified.

minor comments (1)

[Abstract] Abstract: The phrase 'readily available point cloud conditions' would benefit from a brief clarification of the source and preprocessing of the point clouds, even if only at a high level.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on the abstract claims. We agree that stronger support and details are needed and will revise the manuscript accordingly.

read point-by-point responses

Referee: [Abstract] Abstract: The central claim that 'extensive experiments across diverse datasets demonstrate that our method outperforms existing baselines in visual realism and geometric consistency' is unsupported by any quantitative metrics, tables, figures, error bars, dataset specifications, or ablation studies in the provided manuscript text. This absence prevents evaluation of the primary assertion of superiority and is load-bearing for the paper's contribution.

Authors: We agree that the provided manuscript text does not include the supporting quantitative results, tables, or dataset details referenced in the abstract. The full paper contains Section 4 with experiments across datasets, but since these are absent from the version reviewed, we will revise by adding explicit references to the metrics (e.g., visual realism and consistency scores), tables, figures with error bars, dataset specifications, and ablation studies directly into or adjacent to the abstract. This addresses the load-bearing issue. revision: yes
Referee: [Abstract] Abstract: The global-to-local hierarchical inpainting strategy is asserted to ensure 'consistent identity and appearance of the inserted vehicles' across long sequences, yet no implementation details, loss formulations, or quantitative evaluation of identity preservation (e.g., appearance consistency metrics over sequence length) are supplied, leaving the weakest assumption unverified.

Authors: We agree that the abstract asserts the benefit of the global-to-local strategy without sufficient supporting details in the provided text. We will revise by expanding the method description (Section 3) with implementation specifics, loss formulations for identity preservation, and by adding quantitative evaluations such as appearance consistency metrics tracked over sequence length in the experiments. This will verify the claim. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity

full rationale

The paper describes a new methodological pipeline (point-cloud-conditioned video inpainting + global-to-local hierarchy + urban 3D Gaussian extraction) rather than any derivation chain, equations, or fitted parameters. No self-definitional steps, fitted inputs renamed as predictions, or load-bearing self-citations appear in the abstract or described construction. The central claims rest on the explicit 3D structure supplied by point clouds and the design choices for consistency, which are presented as independent engineering contributions without reduction to prior fitted quantities or author-only theorems.

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

The framework rests on the domain assumption that point clouds are readily available and that existing video inpainting architectures can be conditioned effectively on them; no free parameters or invented entities are mentioned.

assumptions (2)

domain assumption Point clouds for target vehicles are readily available as conditioning input
Abstract states 'the readily available point cloud conditions enable superior generalization'
domain assumption Video inpainting models can produce temporally consistent output when conditioned on point clouds
Central to the proposed generation process

how reviews work

0 comments

Cite this review

Pith. "Pith review of DriveWeaver: Point-Conditioned Video Inpainting for Controllable Vehicle Insertion in Autonomous Driving Simulation." pith.science (2026). https://pith.science/paper/O4PR7Y4J

@misc{pith2026260631918,
  author       = {Pith},
  title        = {Pith review of: DriveWeaver: Point-Conditioned Video Inpainting for Controllable Vehicle Insertion in Autonomous Driving Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O4PR7Y4J}},
  note         = {Machine review of arXiv:2606.31918}
}

read the original abstract

A pivotal step in autonomous driving simulation involves inserting foreground vehicles with predefined trajectories into simulated scenes. This process enhances scene diversity and facilitates the creation of various corner cases for testing and improving autonomous driving models. However, existing methods often rely on pre-reconstructed 3D assets, which frequently lead to lighting inconsistencies between the inserted foreground and the background. Moreover, the reliance on limited, manually-curated 3D assets hinders large-scale deployment. To address these challenges, we propose DriveWeaver, a novel framework for controllable vehicle insertion in autonomous driving simulation. Specifically, for a masked target insertion area, DriveWeaver performs video inpainting conditioned on vehicle point clouds to generate high-quality, temporally consistent vehicles. This video-inpainting-based approach ensures seamless blending between the foreground and background, while the readily available point cloud conditions enable superior generalization. To support long-term generation, we further design a global-to-local hierarchical inpainting strategy, ensuring the consistent identity and appearance of the inserted vehicles. Meanwhile, we extract explicit 3D Gaussian representations of the inserted vehicles through an urban reconstruction pipeline to enable real-time rendering for autonomous driving simulation. Extensive experiments across diverse datasets demonstrate that our method outperforms existing baselines in visual realism and geometric consistency, providing a robust tool for scalable autonomous driving scene augmentation.

Figures

Figures reproduced from arXiv: 2606.31918 by the authors.

**Figure 2.** Pipeline of DriveWeaver. Given an input driving video, a vehicle point cloud, and a target trajectory, DriveWeaver generates realistic scenarios with the vehicle seamlessly inserted. The pipeline leverages a frozen video diffusion backbone (Wan-DiT) and introduces a lightweight, trainable PointAdapter. Specifically, the processed RGB conditions and masks are channel-wise concatenated and fed into the PointAdapter,… view at source ↗

**Figure 3.** Long-term generation. To prevent error accumulation and semantic drift common in autoregressive methods, DriveWeaver decouples the generation into two stages. (a) Global anchoring stage: The model processes a temporally subsampled sparse sequence in a single pass to establish a consistent vehicle identity and lighting across the entire trajectory. (b) Local interpolation stage: The synthesized sparse anchor frames … view at source ↗

Figures from the paper (5 more)

**Figure 3.** Figure 3: Long-term generation. To prevent error accumulation and semantic drift common in autoregressive methods, DriveWeaver decouples the generation into two stages. (a) Global anchoring stage: The model processes a temporally subsampled sparse sequence in a single pass to es…

**Figure 4.** Figure 4: Qualitative comparison on the Waymo [38] dataset. [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

**Figure 4.** Figure 4: Qualitative comparison on the Waymo [38] dataset. Given that pre-reconstructed background 3DGS models serve as a standard prerequisite in practical autonomous driving simulation pipelines, this marginal offline distillation cost is readily justifiable. 4 Experiments W…

**Figure 5.** Figure 5: Qualitative comparison on the PandaSet [46] dataset. [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

**Figure 6.** Figure 6: Ablation study on architectural designs. [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 65 canonical work pages

[1]

In: ICLR (2023)

Chen, X., Wang, Y., Zhang, L., Zhuang, S., Ma, X., Yu, J., Wang, Y., Lin, D., Qiao, Y., Liu, Z.: Seine: Short-to-long video diffusion model for generative transition and prediction. In: ICLR (2023)

work page 2023
[2]

arXiv preprint (2023)

Chen, Y., Gu, C., Jiang, J., Zhu, X., Zhang, L.: Periodic vibration gaussian: Dy- namic urban scene reconstruction and real-time rendering. arXiv preprint (2023)

work page 2023
[3]

IEEE TPAMI (2025)

Chen, Y., Zhang, J., Xie, Z., Li, W., Zhang, F., Lu, J., Zhang, L.: S-nerf++: Autonomous driving simulation via neural reconstruction and generation. IEEE TPAMI (2025)

work page 2025
[4]

arXiv preprint (2024)

Chen, Z., Yang, J., Huang, J., de Lutio, R., Esturo, J.M., Ivanovic, B., Litany, O., Gojcic, Z., Fidler, S., Pavone, M., Song, L., Wang, Y.: Omnire: Omni urban scene reconstruction. arXiv preprint (2024)

work page 2024
[5]

arXiv preprint (2023)

Chung, H.W., Constant, N., Garcia, X., Roberts, A., Tay, Y., Narang, S., Firat, O.: Unimax: Fairer and more effective language sampling for large-scale multilingual pretraining. arXiv preprint (2023)

work page 2023
[6]

In: CoRL (2017)

Dosovitskiy, A., Ros, G., Codevilla, F., Lopez, A., Koltun, V.: Carla: An open urban driving simulator. In: CoRL (2017)

work page 2017
[7]

In: ICCV (2025)

Du, X., Wang, Y., Sun, H., Wu, Z., Sheng, H., Wang, S., Ying, J., Lu, M., Zhu, T., Zhan, K., Yu, X.: 3drealcar: An in-the-wild rgb-d car dataset with 360-degree views. In: ICCV (2025)

work page 2025
[8]

In: ICML (2024)

Esser, P., Kulal, S., Blattmann, A., Entezari, R., Müller, J., Saini, H., Levi, Y., Lorenz, D., Sauer, A., Boesel, F., et al.: Scaling rectified flow transformers for high-resolution image synthesis. In: ICML (2024)

work page 2024

Show all 65 references

[9]

arXiv preprint (2025)

Gao, C., Ding, L., Cai, X., Huang, Z., Wang, Z., Xue, T.: Lora-edit: Controllable first-frame-guided video editing via mask-aware lora fine-tuning. arXiv preprint (2025)

2025
[10]

arXiv preprint (2025)

Gao, H., Chen, S., Jiang, B., Liao, B., Shi, Y., Guo, X., Pu, Y., Yin, H., Li, X., Zhang, X., Zhang, Y., Liu, W., Zhang, Q., Wang, X.: Rad: Training an end-to-end driving policy via large-scale 3dgs-based reinforcement learning. arXiv preprint (2025)

2025
[11]

In: NIPS (2024)

Gao, S., Yang, J., Chen, L., Chitta, K., Qiu, Y., Geiger, A., Zhang, J., Li, H.: Vista: A generalizable driving world model with high fidelity and versatile controllability. In: NIPS (2024)

2024
[12]

arXiv preprint (2023)

Guo, Y., Yang, C., Rao, A., Wang, Y., Qiao, Y., Lin, D., Dai, B.: Animatediff: Animate your personalized text-to-image diffusion models without specific tuning. arXiv preprint (2023)

2023
[13]

Caesar, J

H. Caesar, J. Kabzan, K.T.e.a.: Nuplan: A closed-loop ml-based planning bench- mark for autonomous vehicles. In: CVPR ADP3 workshop (2021)

2021
[14]

In: CVPR (2025)

Henschel, R., Khachatryan, L., Hayrapetyan, D., Poghosyan, H., Tadevosyan, V., Wang, Z., Navasardyan, S., Shi, H.: Streamingt2v: Consistent, dynamic, and ex- tendable long video generation from text. In: CVPR (2025)

2025
[15]

In: NIPS (2017)

Heusel,M.,Ramsauer,H.,Unterthiner,T.,Nessler,B.,Hochreiter,S.:Ganstrained by a two time-scale update rule converge to a local nash equilibrium. In: NIPS (2017)

2017
[16]

Ho,J.,Jain,A.,Abbeel,P.:Denoisingdiffusionprobabilisticmodels.arXivpreprint (2020)

2020
[17]

arXiv preprint (2022)

Ho, J., Salimans, T.: Classifier-free diffusion guidance. arXiv preprint (2022)

2022
[18]

In: ECCV (2020) DriveWeaver 17

Hu, Y.T., Wang, H., Ballas, N., Grauman, K., Schwing, A.G.: Proposal-based video completion. In: ECCV (2020) DriveWeaver 17

2020
[19]

arXiv preprint (2025)

Jiang, J., Song, N., Li, J., Zhu, X., Zhang, L.: Realengine: Simulating autonomous driving in realistic context. arXiv preprint (2025)

2025
[20]

arXiv preprint (2025)

Jiang, Z., Han, Z., Mao, C., Zhang, J., Pan, Y., Liu, Y.: Vace: All-in-one video creation and editing. arXiv preprint (2025)

2025
[21]

In: ACM TOG (2023)

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

2023
[22]

arXiv preprint (2024)

Kong, W., Tian, Q., Zhang, Z., Min, R., Dai, Z., Zhou, J., Xiong, J., Li, X., Wu, B., Zhang, J., et al.: Hunyuanvideo: A systematic framework for large video generative models. arXiv preprint (2024)

2024
[23]

arXiv preprint (2025)

Li, J., Jiang, J., Miao, J., Long, M., Wen, T., Jia, P., Liu, S., Yu, C., Liu, M., Cai, Y., et al.: Realistic and controllable 3d gaussian-guided object editing for driving video generation. arXiv preprint (2025)

2025
[24]

In: ECCV (2024)

Liang, R., Gojcic, Z., Nimier-David, M., Acuna, D., Vijaykumar, N., Fidler, S., Wang, Z.: Photorealistic object insertion with diffusion-guided inverse rendering. In: ECCV (2024)

2024
[25]

In: AAAI (2025)

Liang, Y., Yan, Z., Chen, L., Zhou, J., Yan, L., Zhong, S., Zou, X.: Driveeditor: A unified 3d information-guided framework for controllable object editing in driving scenes. In: AAAI (2025)

2025
[26]

In: ICLR (2023)

Lipman, Y., Chen, R.T.Q., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. In: ICLR (2023)

2023
[27]

In: CVPR Workshops (2026)

Ljungbergh, W., Taveira, B., Zheng, W., Tonderski, A., Peng, C., Kahl, F., Peters- son, C., Felsberg, M., Keutzer, K., Tomizuka, M., et al.: R3d2: Realistic 3d asset insertion via diffusion for autonomous driving simulation. In: CVPR Workshops (2026)

2026
[28]

In: ECCV (2024)

Ljungbergh, W., Tonderski, A., Johnander, J., Caesar, H., Åström, K., Felsberg, M., Petersson, C.: Neuroncap: Photorealistic closed-loop safety testing for au- tonomous driving. In: ECCV (2024)

2024
[29]

In: CVPR (2025)

Lu, Y., Cai, Y., Zhang, S., Zhou, H., Hu, H., Yu, H., Geiger, A., Liao, Y.: Ur- bancad: Towards highly controllable and photorealistic 3d vehicles for urban scene simulation. In: CVPR (2025)

2025
[30]

In: ICCV (2025)

Ma, Z., Jiang, J., Chen, Y., Zhang, L.: Béziergs: Dynamic urban scene reconstruc- tion with bézier curve gaussian splatting. In: ICCV (2025)

2025
[31]

In: ECCV (2020)

Mildenhall, B., Srinivasan, P.P., Tancik, M., Barron, J.T., Ramamoorthi, R., Ng, R.: Nerf: Representing scenes as neural radiance fields for view synthesis. In: ECCV (2020)

2020
[32]

In: CVPR (2021)

Ost, J., Mannan, F., Thuerey, N., Knodt, J., Heide, F.: Neural scene graphs for dynamic scenes. In: CVPR (2021)

2021
[33]

In: ICCV (2023)

Peebles, W., Xie, S.: Scalable diffusion models with transformers. In: ICCV (2023)

2023
[34]

arXiv preprint (2025)

Qiu, H., Yu, N., Huang, Z., Debevec, P., Liu, Z.: Cinescale: Free lunch in high- resolution cinematic visual generation. arXiv preprint (2025)

2025
[35]

In: CVPR (2022)

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

2022
[36]

In: FSR (2017)

Shah, S., Dey, D., Lovett, C., Kapoor, A.: Airsim: High-fidelity visual and physical simulation for autonomous vehicles. In: FSR (2017)

2017
[37]

arXiv preprint (2024)

Singh, B., Kulharia, V., Yang, L., Ravichandran, A., Tyagi, A., Shrivastava, A.: Genmm: Geometrically and temporally consistent multimodal data generation for video and lidar. arXiv preprint (2024)

2024
[38]

In: CVPR (2020) 18 J

Sun, P., Kretzschmar, H., Dotiwalla, X., Chouard, A., Patnaik, V., Tsui, P., Guo, J., Zhou, Y., Chai, Y., Caine, B., et al.: Scalability in perception for autonomous driving: Waymo open dataset. In: CVPR (2020) 18 J. Jiang et al

2020
[39]

arXiv preprint (2023)

Turki, H., Zhang, J.Y., Ferroni, F., Ramanan, D.: Suds: Scalable urban dynamic scenes. arXiv preprint (2023)

2023
[40]

arXiv preprint (2018)

Unterthiner, T., van Steenkiste, S., Kurach, K., Marinier, R., Michalski, M., Gelly, S.: Towards accurate generative models of video: A new metric & challenges. arXiv preprint (2018)

2018
[41]

arXiv preprint (2025)

Wan, T., Wang, A., Ai, B., Wen, B., Mao, C., Xie, C.W., Chen, D., Yu, F., Zhao, H., Yang, J., Zeng, J., Wang, J., Zhang, J., Zhou, J., Wang, J., Chen, J., Zhu, K., Zhao, K., Yan, K., Huang, L., Feng, M., Zhang, N., Li, P., Wu, P., Chu, R., Feng, R., Zhang, S., Sun, S., Fang, T...

2025
[42]

In: AAAI (2019)

Wang, C., Huang, H., Han, X., Wang, J.: Video inpainting by jointly learning temporal structure and spatial details. In: AAAI (2019)

2019
[43]

In: ICLR (2025)

Wang, Q., Fan, L., Wang, Y., Chen, Y., Zhang, Z.: Freevs: Generative view syn- thesis on free driving trajectory. In: ICLR (2025)

2025
[44]

TIP (2004)

Wang, Z., Bovik, A.C., Sheikh, H.R., Simoncelli, E.P.: Image quality assessment: from error visibility to structural similarity. TIP (2004)

2004
[45]

In: CVPR (2023)

Wang, Z., Shen, T., Gao, J., Huang, S., Munkberg, J., Hasselgren, J., Gojcic, Z., Chen, W., Fidler, S.: Neural fields meet explicit geometric representations for inverse rendering of urban scenes. In: CVPR (2023)

2023
[46]

In: ITSC (2021)

Xiao, P., Shao, Z., Hao, S., Zhang, Z., Chai, X., Jiao, J., Li, Z., Wu, J., Sun, K., Jiang, K., Wang, Y., Yang, D.: Pandaset: Advanced sensor suite dataset for autonomous driving. In: ITSC (2021)

2021
[47]

In: ICLR (2023)

Xie, Z., Zhang, J., Li, W., Zhang, F., Zhang, L.: S-nerf: Neural radiance fields for street views. In: ICLR (2023)

2023
[48]

In: CVPR (2019)

Xu, R., Li, X., Zhou, B., Loy, C.C.: Deep flow-guided video inpainting. In: CVPR (2019)

2019
[49]

In: ECCV (2024)

Yan, Y., Lin, H., Zhou, C., Wang, W., Sun, H., Zhan, K., Lang, X., Zhou, X., Peng, S.: Street gaussians: Modeling dynamic urban scenes with gaussian splatting. In: ECCV (2024)

2024
[50]

In: CVPR (2025)

Yan, Y., Xu, Z., Lin, H., Jin, H., Guo, H., Wang, Y., Zhan, K., Lang, X., Bao, H., Zhou, X., Peng, S.: Streetcrafter: Street view synthesis with controllable video diffusion models. In: CVPR (2025)

2025
[51]

arXiv preprint (2024)

Yang, J., Huang, J., Chen, Y., Wang, Y., Li, B., You, Y., Sharma, A., Igl, M., Karkus, P., Xu, D., et al.: Storm: Spatio-temporal reconstruction model for large- scale outdoor scenes. arXiv preprint (2024)

2024
[52]

arXiv preprint (2023)

Yang, J., Ivanovic, B., Litany, O., Weng, X., Kim, S.W., Li, B., Che, T., Xu, D., Fidler, S., Pavone, M., Wang, Y.: Emernerf: Emergent spatial-temporal scene decomposition via self-supervision. arXiv preprint (2023)

2023
[53]

In: CVPR (2023)

Yang, Z., Chen, Y., Wang, J., Manivasagam, S., Ma, W.C., Yang, A.J., Urtasun, R.: Unisim: A neural closed-loop sensor simulator. In: CVPR (2023)

2023
[54]

In: ICCV (2025)

Yang, Z., Pan, Z., Yang, Y., Zhu, X., Zhang, L.: Driving scene synthesis on free- form trajectories with generative prior. In: ICCV (2025)

2025
[55]

arXiv preprint (2024)

Yang, Z., Teng, J., Zheng, W., Ding, M., Huang, S., Xu, J., Yang, Y., Hong, W., Zhang, X., Feng, G., et al.: Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint (2024)

2024
[56]

In: CVPR (2018) DriveWeaver 19

Zhang, R., Isola, P., Efros, A.A., Shechtman, E., Wang, O.: The unreasonable effectiveness of deep features as a perceptual metric. In: CVPR (2018) DriveWeaver 19

2018
[57]

In: ICCV (2025)

Zhang, Y., Cao, C., Yu, C., Zhu, J.: Lion-lora: Rethinking lora fusion to unify controllable spatial and temporal generation for video diffusion. In: ICCV (2025)

2025
[58]

In: CVPR (2025)

Zhao, G., Ni, C., Wang, X., Zhu, Z., Zhang, X., Wang, Y., Huang, G., Chen, X., Wang, B., Zhang, Y., Mei, W., Wang, X.: Drivedreamer4d: World models are effective data machines for 4d driving scene representation. In: CVPR (2025)

2025
[59]

arXiv preprint (2026)

Zheng, S., Yin, M., Hu, W., Li, X., Shan, Y., Fu, Y.: Versecrafter: Dynamic realistic video world model with 4d geometric control. arXiv preprint (2026)

2026
[60]

arXiv preprint (2024)

Zheng, Z., Peng, X., Yang, T., Shen, C., Li, S., Liu, H., Zhou, Y., Li, T., You, Y.: Open-sora: Democratizing efficient video production for all. arXiv preprint (2024)

2024
[61]

arXiv preprint (2024)

Zhou, H., Lin, L., Wang, J., Lu, Y., Bai, D., Liu, B., Wang, Y., Geiger, A., Liao, Y.: Hugsim: A real-time, photo-realistic and closed-loop simulator for autonomous driving. arXiv preprint (2024)

2024
[62]

In: CVPR (2024)

Zhou, H., Shao, J., Xu, L., Bai, D., Qiu, W., Liu, B., Wang, Y., Geiger, A., Liao, Y.: Hugs: Holistic urban 3d scene understanding via gaussian splatting. In: CVPR (2024)

2024
[63]

In: ICCV (2023)

Zhou, S., Li, C., Chan, K.C., Loy, C.C.: ProPainter: Improving propagation and transformer for video inpainting. In: ICCV (2023)

2023
[64]

In: ICCV (2025)

Zhou, Y., Ye, N., Ljungbergh, W., Li, T., Yang, J., Yang, Z., Zhu, H., Petersson, C., Li, H.: Decoupled diffusion sparks adaptive scene generation. In: ICCV (2025)

2025
[65]

In: CVPR (2021)

Zou, X., Yang, L., Liu, D., Lee, Y.J.: Progressive temporal feature alignment net- work for video inpainting. In: CVPR (2021)

2021

Pith tools

Reviewed July 3, 2026 · model on record in the stance chip above.

[1] [1]

In: ICLR (2023)

Chen, X., Wang, Y., Zhang, L., Zhuang, S., Ma, X., Yu, J., Wang, Y., Lin, D., Qiao, Y., Liu, Z.: Seine: Short-to-long video diffusion model for generative transition and prediction. In: ICLR (2023)

work page 2023

[2] [2]

arXiv preprint (2023)

Chen, Y., Gu, C., Jiang, J., Zhu, X., Zhang, L.: Periodic vibration gaussian: Dy- namic urban scene reconstruction and real-time rendering. arXiv preprint (2023)

work page 2023

[3] [3]

IEEE TPAMI (2025)

Chen, Y., Zhang, J., Xie, Z., Li, W., Zhang, F., Lu, J., Zhang, L.: S-nerf++: Autonomous driving simulation via neural reconstruction and generation. IEEE TPAMI (2025)

work page 2025

[4] [4]

arXiv preprint (2024)

Chen, Z., Yang, J., Huang, J., de Lutio, R., Esturo, J.M., Ivanovic, B., Litany, O., Gojcic, Z., Fidler, S., Pavone, M., Song, L., Wang, Y.: Omnire: Omni urban scene reconstruction. arXiv preprint (2024)

work page 2024

[5] [5]

arXiv preprint (2023)

Chung, H.W., Constant, N., Garcia, X., Roberts, A., Tay, Y., Narang, S., Firat, O.: Unimax: Fairer and more effective language sampling for large-scale multilingual pretraining. arXiv preprint (2023)

work page 2023

[6] [6]

In: CoRL (2017)

Dosovitskiy, A., Ros, G., Codevilla, F., Lopez, A., Koltun, V.: Carla: An open urban driving simulator. In: CoRL (2017)

work page 2017

[7] [7]

In: ICCV (2025)

Du, X., Wang, Y., Sun, H., Wu, Z., Sheng, H., Wang, S., Ying, J., Lu, M., Zhu, T., Zhan, K., Yu, X.: 3drealcar: An in-the-wild rgb-d car dataset with 360-degree views. In: ICCV (2025)

work page 2025

[8] [8]

In: ICML (2024)

Esser, P., Kulal, S., Blattmann, A., Entezari, R., Müller, J., Saini, H., Levi, Y., Lorenz, D., Sauer, A., Boesel, F., et al.: Scaling rectified flow transformers for high-resolution image synthesis. In: ICML (2024)

work page 2024