Pith. sign in

REVIEW 3 major objections 8 minor 1 cited by

SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors

T0 review · 3 major / 8 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 Satellite views plus cuboid priors produce detailed 3D buildings

desk verdict A genuinely new conditioning mechanism and a useful dataset, but the headline numbers are measured with ground-truth-derived priors, so real-world generalization is asserted rather than shown. read the letter →

arxiv 2508.18531 v1 pith:G3ZLCPIP submitted 2025-08-25 cs.CV cs.AI

Zhangyu Jin , Andrew Feng This is my paper · ORCID
classification cs.CVcs.AI
keywords 3DbuildinggenerationsatelliteimagerycoarsegeometricpriorsflowtransformercosineinterpolationSkylines-50Kdatasetrectifiedtop-downview
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 SatSkylines, a generative model that turns a top-down satellite image and a very coarse geometric prior—like a simple box or a footprint-based shape—into a detailed, textured 3D building. The central trick is to feed the model an interpolation between the latent code of the coarse prior and random Gaussian noise, so the model learns a one-to-many mapping from simple priors to many plausible detailed buildings. On a 500-building test set the method reaches a geometry IoU of 0.938, far above the 0.442 scored by Trellis with image-only input, and does so in about 15 seconds per building. If the result holds up in the wild, it would let city-scale 3D modeling be driven by globally available satellite imagery and OpenStreetMap footprints, without needing detailed voxel models or ground-level photos.

What carries the argument

The load-bearing mechanism is cosine geometric interpolation between a channel-wise normalized SS-VAE latent of the coarse geometric prior and Gaussian noise, done before the sparse-structure Rectified Flow transformer. The normalization forces the prior latent to look like noise (roughly $\mathcal{N}(0,1)$), and the identity $\cos^2(\lambda\pi/2)+\sin^2(\lambda\pi/2)=1$ guarantees the interpolated input stays Gaussian, so the model is trained on a smooth path from 'mostly prior' to 'pure noise' and at inference λ acts as a geometric-guidance dial.

What would settle it

Take a set of GPS coordinates covered by OpenStreetMap, run the full pipeline to get OSM-derived priors, then compare the generated geometry against airborne LiDAR scans of the same buildings: if the median voxel IoU against the LiDAR ground truth falls well below the 0.938 reported on clean priors (say below 0.7), then the clean-prior assumption is the limiting factor and the headline accuracy does not carry over to in-the-wild data.

Watch

Extended reading notes

Core claim

SatSkylines establishes that a single sparse-structure flow transformer, trained on a large set of stylized building assets, can generate detailed and appearance-consistent 3D buildings conditioned on a pure top-down satellite image and a coarse geometric prior as simple as one or two stacked cuboids. The authors achieve this by encoding the coarse prior into the SS-VAE latent space, normalizing that latent channel-wise so it matches the statistics of Gaussian noise, and then cosine-interpolating between the normalized prior latent and pure noise with a strength parameter λ. This keeps the input distribution Gaussian because $\cos^2(\lambda\pi/2)+\sin^2(\lambda\pi/2)=1$, lets the same training recipe as Trellis apply, and gives the user a dial between fidelity to the prior and creative freedom. In their experiments the model reaches 0.9381 IoU and 0.8684 F-score on a 500-instance test set, outperforming image-only Trellis (0.4415 IoU) and voxel-hungry CLAY (0.6859 IoU on a 20-instance subset), with roughly 15-second per-building inference.

Load-bearing premise

The reported accuracy is measured on coarse priors generated from the ground-truth building shapes themselves, not from noisy or incomplete real-world OpenStreetMap data, so the in-the-wild performance depends on real priors being similar enough to those clean priors.

Editorial extensions

If this is right

  • With only a cuboid prior (LOD 0), SatSkylines still achieves 0.9515 IoU on the test set, showing the method degrades gracefully as priors get coarser.
  • Richer priors monotonically improve reconstruction fidelity (Chamfer Distance drops from 0.0222 at LOD 0 to 0.0141 at LOD 2), so the same model benefits from any extra footprint or height information OpenStreetMap can provide.
  • Because the interpolation is applied at inference only and adds no parameters, the approach preserves Trellis-like speed, generating a textured building in about 15 seconds rather than the minutes required by CLAY or Rodin.
  • The Skylines-50K dataset (50,673 hand-authored, textured building assets with PBR materials) can serve as a training resource for other satellite-to-3D building generation models.
  • The end-to-end GPS pipeline (OpenStreetMap footprints plus heights to coarse prior, satellite image retrieval to GPT-based super-resolution to generation) means the method can be applied to arbitrary coordinates without manual voxel editing.

Reading between the lines

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

  • The cosine interpolation's Gaussian-preserving property is not specific to buildings: the same 'normalize the condition latent, then interpolate with noise' recipe could make any image-plus-shape-conditioned flow model accept much coarser geometric controls, provided the condition encoder's output distribution is approximately Gaussian.
  • A testable extension would be to vary λ per building in a real-world city reconstruction, letting the user trade geometric fidelity against hallucinated detail; for landmarks such as the Arc de Triomphe a strong prior matters, while for generic infill a lower λ may give more plausible variety.
  • The paper's real-world evidence is qualitative only; a quantitative study using airborne LiDAR or photogrammetric meshes as ground truth for a few city blocks would be the natural next step to verify whether the 0.94-level IoU transfers to noisy OSM-derived priors.
  • If the method generalizes, it implies that detailed 3D city models could be generated for any location with satellite coverage and OSM data, potentially changing the cost curve for digital twins, gaming, and urban planning.
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 / 8 minor

Summary. The paper introduces SatSkylines, a method for generating detailed 3D building models from top-down satellite imagery and coarse geometric priors such as cuboids or building footprints. The key idea is to encode a coarse geometry prior with a sparse-structure VAE, apply channel-wise latent normalization, and then cosine-interpolate the normalized latent code with Gaussian noise before feeding it to a rectified-flow transformer that generates geometry and appearance. The method is built on the Trellis architecture and is trained on a newly curated dataset, Skylines-50K, consisting of over 50,000 building assets harvested from the game Cities: Skylines. The paper also presents an end-to-end pipeline that, given a GPS bounding box, extracts OpenStreetMap footprints and heights as priors, enhances satellite imagery with a GPT-based model, and generates 3D buildings. Quantitative experiments on a 500-instance test set report large gains over Trellis and CLAY (IoU 0.9381 vs 0.4415 for Trellis), together with ablations and qualitative real-world results.

Significance. If the reported results are robust, the paper makes a useful contribution: a simple, computationally cheap conditioning mechanism (cosine interpolation with latent normalization) that enables geometric control in image-conditioned 3D generation, and a large, diverse, textured 3D building dataset that could support further research. The method also demonstrates fast inference (~15 s per building), which is valuable for city-scale applications. The paper is commendable for including a finetuned Trellis baseline in the ablations and for clearly describing the dataset curation. However, the evidence for real-world generalization is currently qualitative, and the quantitative evaluation is conducted under conditions that systematically differ from the real-world pipeline. The central claim of strong generalization therefore requires additional experimental support.

major comments (3)
  1. [§3.2, §3.3, Table 1] The quantitative evaluation uses coarse geometric priors generated from the ground-truth assets themselves: Section 3.2 states that LOD 1 is the target building's exact footprint extruded to its exact height, and LOD 2 uses two exact cross-sections. In contrast, the real-world pipeline in Section 3.3 uses OpenStreetMap footprints and height attributes, which are typically simplified, sometimes offset by meters, and often noisy or incomplete. The model is never trained or evaluated on such misaligned priors, so the reported IoU of 0.9381 in Table 1 is an upper bound measured under perfect-prior conditions and does not by itself substantiate the claimed strong generalization to real-world data. I request an additional experiment: either evaluate the method on real OSM-derived priors (with reference 3D models from LiDAR or photogrammetry), or simulate realistic prior noise on the existing test set (e.g., random footprint simplification, translation offsets, height quantization) and report the resulting metric degradation.
  2. [§4.1, Tables 1–3] No error bars, confidence intervals, or significance tests are reported for any of the quantitative results. This is particularly problematic for the CLAY comparison, which is carried out on only 20 assets (Table 1), and for the ablations where the differences are small, such as the 0.003 Chamfer Distance difference between SatSkylines and 'wo ss latent norm' in Table 3, and the LOD 0 versus LOD 1 IoU difference in Table 2. Without variance estimates, it is unclear whether the reported improvements, especially those used to justify the latent normalization and the ordering of LOD levels, are meaningful. The authors should provide multiple runs, bootstrap confidence intervals, or at least error bars over test-set splits.
  3. [§3.3, §4.1, Fig. 7] The image-condition distribution used in evaluation is also aligned with the ground truth: the top-down images for Skylines-50K are rendered from the same 3D assets, whereas the real-world pipeline uses satellite imagery from Google Maps or Mapbox, which differs in resolution, scale, lighting, and imaging artifacts, and is further post-processed by 'gpt-image-1'. The paper provides only qualitative real-world results and no quantitative assessment of this domain gap. I recommend a quantitative real-world evaluation on a small set of buildings with available reference geometry, or at minimum a clear statement that the quantitative results apply to the synthetic rendered setting and that real-world performance is likely lower.
minor comments (8)
  1. [§1] The word 'Condtional' in the first sentence should be corrected to 'Conditional'.
  2. [§4.1] The phrase 'we empoly Chamfer Distance' contains a typo; 'empoly' should be 'employ'.
  3. [Table 2 caption] The caption contains the typo 'Geormtric Priors'; it should be 'Geometric Priors'.
  4. [§4.1] The paper does not specify how the 20-instance sub-test set was selected for the CLAY comparison. For reproducibility, please describe the selection process (e.g., random or stratified).
  5. [§3.3, Supplementary Fig. 9] The prompt used for 'gpt-image-1' appears only in the supplementary material. Since the pipeline depends on this prompt, consider providing the full prompt in the main text or referencing it more prominently, as the model output is not reproducible without it.
  6. [Supplementary §3] The statement that the dataset 'will be released' appears only in the supplementary material; it should be stated in the main text, as it is an important contribution.
  7. [Tables 1 and 3] To avoid confusion, the caption of Table 3 should clarify that 'Trellis w finetune on Skylines-50K' is the image-only Trellis model finetuned on the training split of Skylines-50K, in contrast to the un-finetuned Trellis row in Table 1.
  8. [§4.1] The paper reports IoU, CD, and F-score on 'generated sparse structure voxels', but does not state the voxel resolution used for evaluation. Please specify the grid resolution to enable fair comparison with future work.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation chain; the ground-truth-derived priors are an evaluation-validity limitation, not circular reasoning.

full rationale

SatSkylines is an empirical generative model rather than a derivation chain: the cosine interpolation between normalized coarse-prior latents and Gaussian noise is a conditioning mechanism (Sec. 3.1), and the reported IoU, Chamfer Distance, and F-score are measurements on a held-out test set, not quantities obtained by substituting the priors into an equation that forces the result. The only author self-citation is reference [4] (Chen et al., which includes co-author Andrew Feng), used as one of several point-cloud dataset examples in Related Work; it is not load-bearing. The evaluation does contain a limitation relevant to the real-world generalization claim: Sec. 3.2 states that the coarse priors are generated from the ground-truth assets themselves ('we generate coarse geometric priors ourselves. Each asset is assigned three levels of priors'), while the real-world pipeline in Sec. 3.3 derives priors from OpenStreetMap instead. No quantitative test measures the effect of noisy, simplified, or misaligned OSM priors, and the supplementary material concedes that real-world outputs 'are not expected to be identical to the ground truth.' This is an evaluation-validity gap, not a circular reduction, because the model does not analytically derive its outputs from those priors. Therefore no circular step is identified; the paper's central method is self-contained as an empirical system, and the strong real-world generalization claim should be read as supported only qualitatively.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central method introduces a new interpolation scheme and a dataset, but no new physical or mathematical entities. The main unstated dependencies are the smoothness of the Trellis latent space, the representativeness of the game-derived dataset, and the comparability of OSM priors to the ground-truth-derived priors used in evaluation.

free parameters (4)
  • Inference interpolation strength lambda = 0.5
    Fixed by hand at inference to balance geometric guidance and creativity; affects the trade-off between prior fidelity and diversity (Section 4, implementation details).
  • CFG scales = SS 7.5, SLat 3.0
    Hand-tuned classifier-free guidance scales for the two flow transformers (Section 4).
  • Channel-wise latent normalization constants = Not numerically reported; empirical mean and standard deviation of ZO
    Estimated from the SS VAE latent distribution, which the paper states is roughly N(0,0.2); these constants are fixed for training and inference (Section 3.1, Section 4).
  • Training lambda distribution parameters = Logit-normal mu=1, sigma=1
    Chosen for the interpolation factor during training; affects how strongly priors are emphasized (Section 4).
assumptions (5)
  • domain assumption The SS VAE latent space of Trellis is smooth enough that interpolating between a normalized prior latent and Gaussian noise produces valid inputs for the flow transformer.
    The method relies on this property to enable geometric control; the paper does not prove it, only demonstrates empirically (Section 3.1).
  • domain assumption Rectified flow training with the interpolated latents converges to a model that maps noise-plus-prior to detailed geometry.
    The success of the training objective is assumed from prior results on Rectified Flow and Trellis; no convergence analysis is given (Section 3, references [39,55]).
  • standard math The linear combination of the normalized prior latent and Gaussian noise remains Gaussian because cos^2 + sin^2 = 1.
    This is a standard fact, but the assumption that both components exactly follow standard Gaussian distributions is empirical; the normalization is designed to make the prior latent roughly standard normal (Section 3.1).
  • domain assumption The Cities: Skylines Steam Workshop assets provide a sufficiently diverse and realistic distribution of global building styles for training a model that generalizes to real satellite imagery.
    The paper asserts global diversity from the game's community but provides no quantitative distribution analysis; the real-world generalization claim rests on this (Section 3.2, Section 5).
  • domain assumption OpenStreetMap footprints and height attributes contain enough information to construct coarse geometric priors comparable to the ground-truth-derived LOD priors used in training.
    The real-world pipeline generates priors from OSM, while the training and evaluation priors are generated from the target assets themselves (Section 3.2 versus Section 3.3).

how reviews work

0 comments
Cite this review

Pith. "Pith review of SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors." pith.science (2026). https://pith.science/paper/G3ZLCPIP

@misc{pith2026250818531,
  author       = {Pith},
  title        = {Pith review of: SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G3ZLCPIP}},
  note         = {Machine review of arXiv:2508.18531}
}
read the original abstract

We present SatSkylines, a 3D building generation approach that takes satellite imagery and coarse geometric priors. Without proper geometric guidance, existing image-based 3D generation methods struggle to recover accurate building structures from the top-down views of satellite images alone. On the other hand, 3D detailization methods tend to rely heavily on highly detailed voxel inputs and fail to produce satisfying results from simple priors such as cuboids. To address these issues, our key idea is to model the transformation from interpolated noisy coarse priors to detailed geometries, enabling flexible geometric control without additional computational cost. We have further developed Skylines-50K, a large-scale dataset of over 50,000 unique and stylized 3D building assets in order to support the generations of detailed building models. Extensive evaluations indicate the effectiveness of our model and strong generalization ability.

Figures

Figures reproduced from arXiv: 2508.18531 by the authors.

Figure 1
Figure 1. 3D building assets generated by our method using satellite imagery and coarse geometric priors. Our method not only demon [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Necessity of our method. Trellis fails to recover building heights (upper-middle). CLAY requires highly detailed voxels to work well (lower-middle). Our method takes top-down images and coarse geometric priors to generate realistic 3D buildings (right). veloped Skylines-50K, a large-scale dataset of over 50,000 unique and stylized 3D building assets in order to support the generations of detailed building models. Ex… view at source ↗
Figure 3
Figure 3. Skylines-50K is a large-scale, diverse, high quality 3D building dataset. These assets are sourced from the Steam Workshop of the famous city-building and simulation game ‘Cities: Skylines’. Examples of rendered buildings are shown here to demonstrate style diversity. itations by constructing a large-scale, globally diverse, and high-quality dataset of textured 3D building assets, enabling models to learn from both … view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: SatSkylines Architecture. The coarse geometric prior O is encoded by the SS VAE to obtain ZO. (a) A channel-wise latent normalization is applied to produce Z ′ O. (b) The cosine geometric interpolation is then performed between Z ′ O and gaussian noise ϵ, with λ contro…
Figure 5
Figure 5. Figure 5: Visual comparisons of generated 3D building assets between our method and previous approaches. The first three rows are from [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Visualization of Coarse Geometric Prior Variations. The first two rows are examples from the Skylines-50K dataset, while the last two rows are real-world cases. Methods Geormtric Priors Coarseness Level Geometry IoU ↑ CD ↓ F Score ↑ Trellis [55] LOD 2 0.6367 0.0529 0.6…
Figure 7
Figure 7. Figure 7: Visualization of GPT Satellite Image Refinement. Blue cir￾cles highlight zoomed-in details from 2D satellite images, while red circles mark the corresponding regions in SatSkylines generated 3D buildings. All examples are randomly sampled from real-world data. Methods …
Figure 8
Figure 8. Figure 8: Real-world 3D building generation pipeline. [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: An example of our GPT refinement process. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: More results generated by our SatSkylines. Left images are from Skyline-50K dataset, and right ones are sampled in real-world. [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]

Review history (2 revisions) →

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. Sat2RealCity: Geometry-Aware and Appearance-Controllable 3D Urban Generation from Satellite Imagery

    cs.CV 2025-11 conditional novelty 6.0 of 10

    A satellite-to-3D-city pipeline that generates building entities with OSM geometry priors and MLLM/T2I appearance guidance reports strong gains over existing city-generation baselines.

Reference graph

Works this paper leans on

70 extracted references · 47 canonical work pages · cited by 1 Pith paper

  1. [1]

    Cube: A roblox view of 3d intelligence

    Kiran Bhat, Nishchaie Khanna, Karun Channa, Tinghui Zhou, Yiheng Zhu, Xiaoxia Sun, Charles Shang, Anirudh Sudarshan, Maurice Chu, Daiqing Li, et al. Cube: A roblox view of 3d intelligence. arXiv preprint arXiv:2503.15475 ,

    arXiv

  2. [2]

    Blender — a 3d modeling and render- ing package, 2024

    Blender Foundation. Blender — a 3d modeling and render- ing package, 2024. Accessed: [Current Date]. 6

    work page 2024

  3. [3]

    Generalized and efficient 2d gaussian splatting for arbitrary- scale super-resolution

    Du Chen, Liyi Chen, Zhengqiang Zhang, and Lei Zhang. Generalized and efficient 2d gaussian splatting for arbitrary- scale super-resolution. arXiv preprint arXiv:2501.06838 ,

    arXiv

  4. [4]

    Stpls3d: A large-scale synthetic and real aerial photogrammetry 3d point cloud dataset.arXiv preprint arXiv:2203.09065, 2022

    Meida Chen, Qingyong Hu, Zifan Yu, Hugues Thomas, An- drew Feng, Yu Hou, Kyle McCullough, Fengbo Ren, and Lucio Soibelman. Stpls3d: A large-scale synthetic and real aerial photogrammetry 3d point cloud dataset.arXiv preprint arXiv:2203.09065, 2022. 4, 5

    arXiv 2022

  5. [5]

    Shaddr: interactive example-based geometry and texture generation via 3d shape detailization and differentiable ren- dering

    Qimin Chen, Zhiqin Chen, Hang Zhou, and Hao Zhang. Shaddr: interactive example-based geometry and texture generation via 3d shape detailization and differentiable ren- dering. In SIGGRAPH Asia 2023 Conference Papers, pages 1–11, 2023. 2, 4, 5, 6

    work page 2023

  6. [6]

    Decollage: 3d detailization by controllable, localized, and learned geom- etry enhancement

    Qimin Chen, Zhiqin Chen, Vladimir G Kim, Noam Aiger- man, Hao Zhang, and Siddhartha Chaudhuri. Decollage: 3d detailization by controllable, localized, and learned geom- etry enhancement. In European Conference on Computer Vision, pages 110–127. Springer, 2024. 4

    work page 2024

  7. [7]

    Art- deco: Arbitrary text guidance for 3d detailizer construction

    Qimin Chen, Yuezhi Yang, Yifang Wang, Vladimir G Kim, Siddhartha Chaudhuri, Hao Zhang, and Zhiqin Chen. Art- deco: Arbitrary text guidance for 3d detailizer construction. arXiv preprint arXiv:2505.20431, 2025. 2, 4, 5, 6, 7

    work page arXiv 2025

  8. [8]

    Open city information model from the WFS interface service

    City of Espoo. Open city information model from the WFS interface service. https://kartat.espoo.fi/3d/ services_en.html, 2025. 2, 4, 5

    work page 2025

Show all 70 references
  1. [9]

    3D city model

    City of Kuopio. 3D city model. https://kuopio. kunta3d.fi/Map.html?locale=en, 2025. 2, 4, 5

    2025

  2. [10]

    Detailgen3d: Generative 3d geometry enhancement via data-dependent flow

    Ken Deng, Yuan-Chen Guo, Jingxiang Sun, Zi-Xin Zou, Yangguang Li, Xin Cai, Yan-Pei Cao, Yebin Liu, and Ding Liang. Detailgen3d: Generative 3d geometry enhancement via data-dependent flow. arXiv preprint arXiv:2411.16820,

    arXiv

  3. [11]

    Coin3d: Controllable and interactive 3d assets generation with proxy- guided conditioning

    Wenqi Dong, Bangbang Yang, Lin Ma, Xiao Liu, Liyuan Cui, Hujun Bao, Yuewen Ma, and Zhaopeng Cui. Coin3d: Controllable and interactive 3d assets generation with proxy- guided conditioning. In ACM SIGGRAPH 2024 Conference Papers, pages 1–10, 2024. 2, 4, 5, 6

    2024

  4. [12]

    Worldscore: A unified evaluation benchmark for world generation

    Haoyi Duan, Hong-Xing Yu, Sirui Chen, Li Fei-Fei, and Ji- ajun Wu. Worldscore: A unified evaluation benchmark for world generation. arXiv preprint arXiv:2504.00983, 2025. 2

    2025

  5. [13]

    Unreal engine, 2024

    Epic Games. Unreal engine, 2024. Accessed: [Current Date]. 6

    2024

  6. [14]

    Geoportal Hamburg

    Freie und Hansestadt Hamburg. Geoportal Hamburg. https://geoportal-hamburg.de/, 2025. 2, 4, 5

    2025

  7. [15]

    Mars: Mesh autore- gressive model for 3d shape detailization

    Jingnan Gao, Weizhe Liu, Weixuan Sun, Senbo Wang, Xibin Song, Taizhang Shang, Shenzhou Chen, Hongdong Li, Xi- aokang Yang, Yichao Yan, et al. Mars: Mesh autore- gressive model for 3d shape detailization. arXiv preprint arXiv:2502.11390, 2025. 2, 4, 5, 6

    2025 arXiv

  8. [16]

    Sum parts: Benchmarking part-level semantic segmentation of ur- ban meshes

    Weixiao Gao, Liangliang Nan, and Hugo Ledoux. Sum parts: Benchmarking part-level semantic segmentation of ur- ban meshes. In Proceedings of the Computer Vision and Pat- tern Recognition Conference, pages 24474–24484, 2025. 2, 4, 5

    2025

  9. [17]

    rembg: Remove image backgrounds

    Daniel Gatis. rembg: Remove image backgrounds. https: //pypi.org/project/rembg/, 2022. Version 2.0.67, released 2025-07-05. 5

    2022

  10. [18]

    3d rotterdam

    Gemeente Rotterdam. 3d rotterdam. https://www. 3drotterdam.nl/, 2025. 2, 4, 5

    2025

  11. [19]

    Mambairv2: Atten- tive state space restoration

    Hang Guo, Yong Guo, Yaohua Zha, Yulun Zhang, Wenbo Li, Tao Dai, Shu-Tao Xia, and Yawei Li. Mambairv2: Atten- tive state space restoration. In Proceedings of the Computer Vision and Pattern Recognition Conference , pages 28124– 28133, 2025. 1

    2025

  12. [20]

    Sparseflex: High-resolution and arbitrary-topology 3d shape modeling

    Xianglong He, Zi-Xin Zou, Chia-Hao Chen, Yuan-Chen Guo, Ding Liang, Chun Yuan, Wanli Ouyang, Yan-Pei Cao, and Yangguang Li. Sparseflex: High-resolution and arbitrary-topology 3d shape modeling. arXiv preprint arXiv:2503.21732, 2025. 2, 4, 5

    2025 arXiv

  13. [21]

    Sensaturban: Learning semantics from urban-scale photogrammetric point clouds

    Qingyong Hu, Bo Yang, Sheikh Khalid, Wen Xiao, Niki Trigoni, and Andrew Markham. Sensaturban: Learning semantics from urban-scale photogrammetric point clouds. International Journal of Computer Vision , 130(2):316–343,

  14. [22]

    Ss3dm: Benchmarking street-view surface reconstruction with a synthetic 3d mesh dataset

    Yubin Hu, Kairui Wen, Heng Zhou, Xiaoyang Guo, and Yong-jin Liu. Ss3dm: Benchmarking street-view surface reconstruction with a synthetic 3d mesh dataset. Advances in Neural Information Processing Systems , 37:106649– 106666, 2024. 2, 4

    2024

  15. [23]

    Sat2city: 3d city generation from a single satel- lite image with cascaded latent diffusion

    Tongyan Hua, Lutao Jiang, Ying-Cong Chen, and Wufan Zhao. Sat2city: 3d city generation from a single satel- lite image with cascaded latent diffusion. arXiv preprint arXiv:2507.04403, 2025. 2, 4, 5

    2025 arXiv

  16. [24]

    City3d: Large-scale building reconstruction from airborne lidar point clouds

    Jin Huang, Jantien Stoter, Ravi Peters, and Liangliang Nan. City3d: Large-scale building reconstruction from airborne lidar point clouds. Remote Sensing, 14(9):2254, 2022. 4

    2022

  17. [25]

    Hunyuan3d 2.1: From images to high- fidelity 3d assets with production-ready pbr material

    Team Hunyuan3D, Shuhui Yang, Mingxin Yang, Yifei Feng, Xin Huang, Sheng Zhang, Zebin He, Di Luo, Haolin Liu, Yunfei Zhao, et al. Hunyuan3d 2.1: From images to high- fidelity 3d assets with production-ready pbr material. arXiv preprint arXiv:2506.15442, 2025. 2, 4, 5

    2025 arXiv

  18. [26]

    Mesh editor, 2024

    Hyper3D. Mesh editor, 2024. Accessed: [Current Date]. 6

    2024

  19. [27]

    Horizon- gs: Unified 3d gaussian splatting for large-scale aerial-to- ground scenes

    Lihan Jiang, Kerui Ren, Mulin Yu, Linning Xu, Junting Dong, Tao Lu, Feng Zhao, Dahua Lin, and Bo Dai. Horizon- gs: Unified 3d gaussian splatting for large-scale aerial-to- ground scenes. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 26789–2679...

    2025

  20. [28]

    A hierarchical 3d gaussian representation for real-time ren- 9 dering of very large datasets.ACM Transactions on Graphics (TOG), 43(4):1–15, 2024

    Bernhard Kerbl, Andreas Meuleman, Georgios Kopanas, Michael Wimmer, Alexandre Lanvin, and George Drettakis. A hierarchical 3d gaussian representation for real-time ren- 9 dering of very large datasets.ACM Transactions on Graphics (TOG), 43(4):1–15, 2024. 4, 5

    2024

  21. [29]

    Hunyuan3d 2.5: Towards high- fidelity 3d assets generation with ultimate details

    Zeqiang Lai, Yunfei Zhao, Haolin Liu, Zibo Zhao, Qingxi- ang Lin, Huiwen Shi, Xianghui Yang, Mingxin Yang, Shuhui Yang, Yifei Feng, et al. Hunyuan3d 2.5: Towards high- fidelity 3d assets generation with ultimate details. arXiv preprint arXiv:2506.16504, 2025. 2, 4, 5

    2025 arXiv

  22. [30]

    3-D-Modell der Lan- deshauptstadt Dresden

    Landeshauptstadt Dresden. 3-D-Modell der Lan- deshauptstadt Dresden. https : / / www . dresden . de / de / leben / stadtportrait / statistik / geoinformationen / 3 - d - modell . php ? shortcut=3D, 2025. 2, 4, 5

    2025

  23. [31]

    Nuiscene: Exploring efficient generation of unbounded out- door scenes

    Han-Hung Lee, Qinghong Han, and Angel X Chang. Nuiscene: Exploring efficient generation of unbounded out- door scenes. arXiv preprint arXiv:2503.16375, 2025. 4

    2025 arXiv

  24. [32]

    Crafts- man3d: High-fidelity mesh generation with 3d native gen- eration and interactive geometry refiner

    Weiyu Li, Jiarui Liu, Hongyu Yan, Rui Chen, Yixun Liang, Xuelin Chen, Ping Tan, and Xiaoxiao Long. Crafts- man3d: High-fidelity mesh generation with 3d native gen- eration and interactive geometry refiner. arXiv preprint arXiv:2405.14979, 2024. 2, 4, 5

    2024 arXiv

  25. [33]

    Step1x-3d: Towards high-fidelity and con- trollable generation of textured 3d assets

    Weiyu Li, Xuanyang Zhang, Zheng Sun, Di Qi, Hao Li, Wei Cheng, Weiwei Cai, Shihao Wu, Jiarui Liu, Zihao Wang, et al. Step1x-3d: Towards high-fidelity and con- trollable generation of textured 3d assets. arXiv preprint arXiv:2505.07747, 2025. 2, 4, 5

    2025 arXiv

  26. [34]

    Matrixcity: A large-scale city dataset for city-scale neural rendering and beyond

    Yixuan Li, Lihan Jiang, Linning Xu, Yuanbo Xiangli, Zhen- zhi Wang, Dahua Lin, and Bo Dai. Matrixcity: A large-scale city dataset for city-scale neural rendering and beyond. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 3205–3215, 2023. 4, 5

    2023

  27. [35]

    Triposg: High-fidelity 3d shape synthesis using large-scale rectified flow models

    Yangguang Li, Zi-Xin Zou, Zexiang Liu, Dehu Wang, Yuan Liang, Zhipeng Yu, Xingchao Liu, Yuan-Chen Guo, Ding Liang, Wanli Ouyang, et al. Triposg: High-fidelity 3d shape synthesis using large-scale rectified flow models. arXiv preprint arXiv:2502.06608, 2025. 2, 4, 5

    2025 arXiv

  28. [36]

    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 European Conference on Computer Vision, pages 93–109. Springer, 2022. 2, 4, 5

    2022

  29. [37]

    Google earth: a new geological resource

    Richard J Lisle. Google earth: a new geological resource. Geology today, 22(1):29–32, 2006. 4

    2006

  30. [38]

    Deep learning based multi-view stereo matching and 3d scene reconstruction from oblique aerial images

    Jin Liu, Jian Gao, Shunping Ji, Chang Zeng, Shaoyi Zhang, and JianYa Gong. Deep learning based multi-view stereo matching and 3d scene reconstruction from oblique aerial images. ISPRS Journal of Photogrammetry and Remote Sensing, 204:42–60, 2023. 4, 5

    2023

  31. [39]

    Flow straight and fast: Learning to generate and transfer data with rectified flow

    Xingchao Liu, Chengyue Gong, and Qiang Liu. Flow straight and fast: Learning to generate and transfer data with rectified flow. arXiv preprint arXiv:2209.03003, 2022. 5

    2022 arXiv

  32. [40]

    T2i-adapter: Learning adapters to dig out more controllable ability for text-to-image diffusion models

    Chong Mou, Xintao Wang, Liangbin Xie, Yanze Wu, Jian Zhang, Zhongang Qi, and Ying Shan. T2i-adapter: Learning adapters to dig out more controllable ability for text-to-image diffusion models. In Proceedings of the AAAI conference on artificial intelligence, pages 4296–4304, 2024. 4

    2024

  33. [41]

    Dinov2: Learning robust visual features without supervision

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193, 2023. 4

    2023 arXiv

  34. [42]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 4195–4205,

  35. [43]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

    2021

  36. [44]

    Build- ingnet: Learning to label 3d buildings

    Pratheba Selvaraju, Mohamed Nabail, Marios Loizou, Maria Maslioukova, Melinos Averkiou, Andreas Andreou, Sid- dhartha Chaudhuri, and Evangelos Kalogerakis. Build- ingnet: Learning to label 3d buildings. In Proceedings of the IEEE/CVF International Conference on Computer Vi- si...

    2021

  37. [45]

    Geoportal Leipzig

    Stadt Leipzig. Geoportal Leipzig. https : / / geoportal . leipzig . de / arcgis / apps / webappviewer3d / index . html ? id = 636b96152aac4769b6cf316312f3bf70, 2025. 2, 4, 5

    2025

  38. [46]

    Digital Twin Victoria

    State of Victoria. Digital Twin Victoria. https : / / digitaltwin.vic.gov.au/public/, 2025. 2, 4

    2025

  39. [47]

    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, pages 12922–12931, 2022. 4, 5

    2022

  40. [48]

    Lion: Latent point dif- fusion models for 3d shape generation

    Arash Vahdat, Francis Williams, Zan Gojcic, Or Litany, Sanja Fidler, Karsten Kreis, et al. Lion: Latent point dif- fusion models for 3d shape generation. Advances in Neural Information Processing Systems, 35:10021–10039, 2022. 4

    2022

  41. [49]

    Xscale-nvs: Cross-scale novel view synthesis with hash featurized manifold

    Guangyu Wang, Jinzhi Zhang, Fan Wang, Ruqi Huang, and Lu Fang. Xscale-nvs: Cross-scale novel view synthesis with hash featurized manifold. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 21029–21039, 2024. 4, 5

    2024

  42. [50]

    Building3d: A urban-scale dataset and benchmarks for learn- ing roof structures from point clouds

    Ruisheng Wang, Shangfeng Huang, and Hongxin Yang. Building3d: A urban-scale dataset and benchmarks for learn- ing roof structures from point clouds. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 20076–20086, 2023. 4

    2023

  43. [51]

    Build- iff: 3d building shape generation using single-image condi- tional point cloud diffusion models

    Yao Wei, George V osselman, and Michael Ying Yang. Build- iff: 3d building shape generation using single-image condi- tional point cloud diffusion models. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 2910–2919, 2023. 4, 5

    2023

  44. [52]

    Realcity3d: A large-scale georeferenced 3d shape dataset of real-world cities

    Congcong Wen, Wenyu Han, Lazarus Chok, Yan Liang Tan, Sheung Lung Chan, Hang Zhao, and Chen Feng. Realcity3d: A large-scale georeferenced 3d shape dataset of real-world cities. 4

  45. [53]

    Sampling with hammersley and halton points

    Tien-Tsin Wong, Wai-Shing Luk, and Pheng-Ann Heng. Sampling with hammersley and halton points. Journal of Graphics Tools, 2(2):9–24, 1997. 2

    1997

  46. [54]

    Direct3d-s2: Gigascale 3d generation made easy with spatial sparse attention

    Shuang Wu, Youtian Lin, Feihu Zhang, Yifei Zeng, Yikang Yang, Yajie Bao, Jiachen Qian, Siyu Zhu, Xun Cao, 10 Philip Torr, et al. Direct3d-s2: Gigascale 3d generation made easy with spatial sparse attention. arXiv preprint arXiv:2505.17412, 2025. 2, 4, 5

    2025 arXiv

  47. [55]

    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, 20...

    2025

  48. [56]

    Citydreamer: Compositional generative model of unbounded 3d cities

    Haozhe Xie, Zhaoxi Chen, Fangzhou Hong, and Ziwei Liu. Citydreamer: Compositional generative model of unbounded 3d cities. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9666–9675,

  49. [57]

    Citydreamer4d: Compositional generative model of un- bounded 4d cities

    Haozhe Xie, Zhaoxi Chen, Fangzhou Hong, and Ziwei Liu. Citydreamer4d: Compositional generative model of un- bounded 4d cities. arXiv e-prints, pages arXiv–2501, 2025. 4

    2025

  50. [58]

    Gauu-scene v2: Assessing the reliability of image-based metrics with expansive lidar image dataset using 3dgs and nerf

    Butian Xiong, Nanjun Zheng, Junhua Liu, and Zhen Li. Gauu-scene v2: Assessing the reliability of image-based metrics with expansive lidar image dataset using 3dgs and nerf. arXiv preprint arXiv:2404.04880, 2024. 4

    2024 arXiv

  51. [59]

    Twintex: Geometry-aware texture generation for abstracted 3d archi- tectural models

    Weidan Xiong, Hongqian Zhang, Botao Peng, Ziyu Hu, Yongli Wu, Jianwei Guo, and Hui Huang. Twintex: Geometry-aware texture generation for abstracted 3d archi- tectural models. ACM Transactions on Graphics (TOG), 42 (6):1–14, 2023. 2, 4

    2023

  52. [60]

    Urbanbis: a large-scale benchmark for fine- grained urban building instance segmentation

    Guoqing Yang, Fuyou Xue, Qi Zhang, Ke Xie, Chi-Wing Fu, and Hui Huang. Urbanbis: a large-scale benchmark for fine- grained urban building instance segmentation. In ACM SIG- GRAPH 2023 Conference Proceedings, pages 1–11, 2023. 2, 4

    2023

  53. [61]

    Blendedmvs: A large- scale dataset for generalized multi-view stereo networks

    Yao Yao, Zixin Luo, Shiwei Li, Jingyang Zhang, Yufan Ren, Lei Zhou, Tian Fang, and Long Quan. Blendedmvs: A large- scale dataset for generalized multi-view stereo networks. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 1790–1799,...

    2020

  54. [62]

    Hi3dgen: High-fidelity 3d geometry generation from images via normal bridging

    Chongjie Ye, Yushuang Wu, Ziteng Lu, Jiahao Chang, Xi- aoyang Guo, Jiaqing Zhou, Hao Zhao, and Xiaoguang Han. Hi3dgen: High-fidelity 3d geometry generation from images via normal bridging. arXiv preprint arXiv:2503.22236, 3:2,

    arXiv

  55. [63]

    Archidiff: Interactive design of 3d architectural forms generated from a single image

    Jun Yin, Wen Gao, Jizhizi Li, Pengjian Xu, Chenglin Wu, Borong Lin, and Shuai Lu. Archidiff: Interactive design of 3d architectural forms generated from a single image. Com- puters in Industry, 168:104275, 2025. 4, 5

    2025

  56. [64]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In Proceedings of the IEEE/CVF international conference on computer vision, pages 3836–3847, 2023. 4

    2023

  57. [65]

    Clay: A controllable large-scale generative model for creat- ing high-quality 3d assets

    Longwen Zhang, Ziyu Wang, Qixuan Zhang, Qiwei Qiu, Anqi Pang, Haoran Jiang, Wei Yang, Lan Xu, and Jingyi Yu. Clay: A controllable large-scale generative model for creat- ing high-quality 3d assets. ACM Transactions on Graphics (TOG), 43(4):1–20, 2024. 2, 4, 5, 6, 7

    2024

  58. [66]

    Drone-assisted road gaussian splatting with cross- view uncertainty

    Saining Zhang, Baijun Ye, Xiaoxue Chen, Yuantao Chen, Zongzheng Zhang, Cheng Peng, Yongliang Shi, and Hao Zhao. Drone-assisted road gaussian splatting with cross- view uncertainty. arXiv preprint arXiv:2408.15242 , 2024. 4, 5

    2024 arXiv

  59. [67]

    Hunyuan3d 2.0: Scaling diffu- sion models for high resolution textured 3d assets generation

    Zibo Zhao, Zeqiang Lai, Qingxiang Lin, Yunfei Zhao, Haolin Liu, Shuhui Yang, Yifei Feng, Mingxin Yang, Sheng Zhang, Xianghui Yang, et al. Hunyuan3d 2.0: Scaling diffu- sion models for high resolution textured 3d assets generation. arXiv preprint arXiv:2501.12202, 2025. 2, 4, 5...

    2025 arXiv

  60. [68]

    The overview of this pipeline is in Fig

    Real-World End-to-End Pipeline In§ 3.3, we have developed an end-to-end pipeline that re- quires only a geo-spatial bounding box of (min-lat, max- lat, min-lon, max-lon) to simplify the data preparation and building model generations. The overview of this pipeline is in Fig. 8...

  61. [69]

    Given only coarse geometric pri- ors, SatSkylines generates plausible buildings from satellite Figure 9

    More Visualizations Figure 10 shows additional 3D assets produced by our method: the left panel displays samples from the Skylines- 50K test set, while the right panel presents randomly se- lected real-world cases. Given only coarse geometric pri- ors, SatSkylines generates pl...

  62. [70]

    More results generated by our SatSkylines

    Dataset As mentioned in §3.2, our Skylines-50K dataset contains 50, 673 3D building assets, and we will release the dataset 1 Figure 10. More results generated by our SatSkylines. Left images are from Skyline-50K dataset, and right ones are sampled in real-world. for future re...

Pith tools

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