Pith. sign in

REVIEW 3 major objections 6 minor 75 references

A single 360° image is enough to generate a complete indoor 3D scene with floors, walls, and separately textured furniture.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 22:28 UTC pith:TDQFNHCI

load-bearing objection Solid engineering paper that correctly targets the missing layout problem in indoor scene generation and ships a useful dataset; comparative claims rest on thin real-world and baseline evidence. the 3 major comments →

arxiv 2607.03990 v1 pith:TDQFNHCI submitted 2026-07-04 cs.CV

InSpace: Structure-Aware 3D Indoor Scene Generation from a Single 360{deg} Image

Gwanhyeong Koo , Hyunsu Kim , Youngji Kim , Taejae Lee , Siwoo Lim , Sunjae Yoon , Suyong Yeon , Chang D. Yoo This is my paper
classification cs.CV
keywords 3D scene generationequirectangular projectionindoor layoutflow matchingview-selective attentionsparse voxelsERP-FRONT
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Single-image 3D generators already make good individual objects, but indoor scenes need the room itself—floors, walls, and the geometry that anchors every piece of furniture. A narrow camera view does not supply enough spatial coverage for that layout, so assets float, collide, or sit in the wrong place. InSpace takes one equirectangular 360° image instead, estimates a partial point cloud and camera center, then runs a three-stage flow-matching pipeline that first builds a coarse voxel room structure with view-selective attention and then fills in detailed layout and per-asset geometry and texture with a hybrid global-local attention. The authors also release ERP-FRONT, a large paired ERP-to-mesh dataset built from 3D-FRONT. The result is a complete, textured indoor scene whose layout and assets are spatially coherent, outperforming prior single-image scene generators on both geometric and rendering metrics and generalizing to real panoramic scans.

Core claim

From a single equirectangular 360° image, InSpace recovers a complete 3D indoor scene that includes both the structural layout (floors and walls) and separately textured assets, with placements that respect the recovered geometry, by cascading partial-geometry priors, view-selective coarse structure generation, and hybrid global-local refinement under flow matching.

What carries the argument

View-selective cross-attention (and its asset-selective counterpart): each voxel or asset token is allowed to attend only to the cubemap faces or image regions that are geometrically visible from the calibrated camera center, so the model composes the full room without receiving conflicting cues from opposite walls.

Load-bearing premise

The monocular depth map and the camera center derived from it must be accurate enough that the resulting visibility masks and partial geometry really do point each voxel at the correct faces of the panorama.

What would settle it

Replace the estimated depth with systematically biased or noisy depth on the same ERP inputs; if view-selective attention and layout-guided inversion still produce coherent rooms whose IoU and asset placement match the reported numbers, the claim that the geometric prior is essential fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. InSpace generates complete 3D indoor scenes—structural layout (floors/walls) plus separately textured assets—from a single equirectangular (ERP) 360° image. The pipeline has three stages: (1) monocular depth lifting to a Partial Scene Geometry (PSG) and calibrated camera center; (2) coarse dense-voxel structure generation with view-selective cross-attention over cubemap features, plus optional Layout-Guided Structure Inversion from the PSG and a 3D OBB detector; (3) layout- and asset-aware fine geometry/texture via global self-attention and asset-selective cross-attention under flow matching, built on TRELLIS.2 O-Voxel latents. The authors introduce ERP-FRONT (~29K synthetic ERP–mesh pairs from 3D-FRONT) and report 3D (IoU, CD, F1) and 2D (PSNR, LPIPS) metrics, ablations of view-selective attention and inversion, and limited comparisons to SceneGen, MIDI, and SAM3D plus qualitative/self-reconstruction results on ReplicaPano.

Significance. The problem framing is well motivated: limited-FOV single-image scene generators omit structural layout and often produce floating or interpenetrating assets. Using full 360° ERP context as the spatial anchor, together with view-selective and asset-selective attention that couple voxel latents to geometrically visible image regions, is a clear technical step beyond asset-only multi-instance pipelines. ERP-FRONT is a useful community resource for paired ERP-to-scene training. Ablations cleanly isolate view-selective attention (Stage-2 IoU ~44→57, CD roughly halved) and inversion (t0 sweeps), which strengthens confidence in the core mechanisms. If the comparative and real-world evidence were tightened, this would be a solid contribution to structure-aware indoor scene generation.

major comments (3)
  1. Sec. 6.2 and Supp. D: Head-to-head comparison with SceneGen, MIDI, and SAM3D uses only 20 scenes, requires manual alignment then ICP, and evaluates asset-level metrics only inside a perspective crop (red dashed boxes), not full-scene layout. This protocol is too narrow and operator-dependent to support a general outperformance claim for complete layout+asset scenes. Either expand to a larger automatic-alignment protocol (or report full-scene metrics where baselines can be fairly scored), or clearly restrict claims to the ERP-FRONT self-benchmark and treat baselines as qualitative context.
  2. Supp. C.2–C.3 (ReplicaPano): Generalization is shown on 7 scenes with self-reconstruction metrics and no competing methods. Domain-gap failures (frames, curtains, texture style) are acknowledged, but the abstract/conclusion still state that InSpace “generalizes well” to realistic panoramas. Strengthen this with more scenes, failure-rate reporting, and/or a limited baseline comparison under the same ERP input; otherwise tone the generalization claim to match the evidence.
  3. Sec. 4.1–4.2, Eq. (5), Fig. 3: View-selective masks and Layout-Guided Structure Inversion both depend on monocular depth and the calibrated camera center c. There is no sensitivity study to depth error, camera-height error, or PSG incompleteness (e.g., wrong depth → wrong visibility cones and inversion prior). A controlled noise/ablation on depth or c would show whether the Stage-2 gains remain when the spatial prior is imperfect—especially important for real ERP inputs where depth is noisier than on ERP-FRONT.
minor comments (6)
  1. Table 1: Optimal t0 differs between Stage 2 (0.5) and Stage 3 (0.7); the text explains this, but a single recommended default and a short note on how users should choose t0 would help reproducibility.
  2. Fig. 2 and Sec. 1: “Existing methods” are illustrated with floating/misplacement; cite which specific outputs (method + figure) are shown so the failure modes are attributable.
  3. Sec. 4.2: Cubemap FoV=120° and α=50 are fixed; a brief sensitivity note (or appendix) would clarify robustness of M_vs.
  4. Supp. B / Table B.1: 3D OBB F1@0.75 is low (~29%); discuss how OBB size/yaw error propagates into Stage-3 asset quality, since Stage 3 is conditioned on these boxes.
  5. Related Work: PanoContext-Former is noted as closest prior with unavailable code; a short qualitative discussion of its reported limitations versus InSpace’s outputs would better position novelty.
  6. Notation: N0/s0 vs 16^3 latent resolution appears in several places; keep a single consistent symbol for the coarse latent grid size.

Circularity Check

0 steps flagged

No circularity: standard cascaded generative pipeline trained and evaluated on a held-out synthetic split with external geometric/perceptual metrics; no prediction reduces to a fitted free parameter or self-citation by construction.

full rationale

InSpace is an engineering paper proposing a three-stage flow-matching pipeline (PSG initialization from monocular depth + equirectangular lift, view-selective DiT for coarse voxels, then OBB-guided global-local hybrid attention for layout+assets). The only free parameters are ordinary network weights trained by rectified flow matching on the authors' new ERP-FRONT training split (26.5k pairs derived from 3D-FRONT); evaluation uses a disjoint test split plus standard external metrics (Voxel IoU, Chamfer, F1, PSNR, LPIPS) and a small external real-world set (ReplicaPano). Layout-Guided Structure Inversion (t0) is an inference-time SDEdit-style schedule, not a fit that is then re-reported as a prediction. View-selective and asset-selective masks (Eqs. 5-9) are geometric constructions from camera center and OBB projection; they do not encode the target geometry by definition. Citations to TRELLIS.2, DINOv3, flow matching, CenterPoint-style detection, etc., are ordinary background tooling, not load-bearing uniqueness theorems by overlapping authors. No equation or claim reduces a reported result to its own input by construction. Circularity burden is therefore zero.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 4 invented entities

The central claim rests on standard generative-modeling machinery (flow matching, DiT, sparse VAEs from TRELLIS.2) plus several domain assumptions about panoramic geometry and a handful of hand-chosen scalars that control visibility and inversion strength. No new physical entities are postulated; the invented constructs are algorithmic (masks, priors, dataset).

free parameters (4)
  • inversion noise level t0 = 0.7 (default)
    Chosen at inference (default 0.7, ablated 0.3/0.5/1.0); controls how strongly PSG anchors Stage 2. Not learned from first principles.
  • visibility sigmoid scale α = 50
    Fixed to 50 to produce near-binary view-selective masks (Eq. 5); hand-tuned for soft but decisive transitions.
  • cubemap FoV = 120°
    Set to 120° to create overlap between faces; design choice that directly shapes the visibility mask.
  • 3D OBB detector loss weights (λ_hm, λ_off, λ_sz, …) = (1,1,5,2,2,1)
    Hand-set composite weights (1,1,5,2,2,1) emphasizing size accuracy; affect downstream asset grounding.
axioms (5)
  • domain assumption Monocular depth estimator (DA2) produces sufficiently accurate ERP depth for back-projection into a usable Partial Scene Geometry.
    Stage 1 and all subsequent spatial conditioning rest on this; invoked in Sec. 4.1 and Fig. 3.
  • domain assumption Sparse voxel latents and SC-VAEs pretrained on TRELLIS.2 transfer without finetuning to indoor scene geometry and materials.
    Explicitly stated in Sec. 6.1; if the latent space cannot represent room-scale structure, later stages fail.
  • standard math Rectified flow matching with DiT blocks can denoise dense/sparse voxel latents conditioned on DINOv3 cubemap tokens.
    Standard generative-modeling assumption (Sec. 3–4.4); not re-derived.
  • domain assumption A 64³ coarse occupancy grid plus a lightweight 3D U-Net detector yields 3D OBBs accurate enough for asset-selective generation.
    Stage 2→3 interface (Sec. 4.2–4.3); quantitative OBB F1@0.5 ≈ 78 % is accepted as sufficient.
  • ad hoc to paper Synthetic rooms rendered from 3D-FRONT are distributionally close enough to real panoramas for the learned model to generalize (or at least produce plausible outputs).
    ERP-FRONT construction (Sec. 5) and ReplicaPano transfer experiments; domain gap is acknowledged but not eliminated.
invented entities (4)
  • View-selective cross-attention mask M_vs no independent evidence
    purpose: Restricts each voxel latent to attend only to geometrically visible cubemap faces given camera center c.
    Core algorithmic novelty of Stage 2 (Eqs. 5–7); no independent physical existence outside the model.
  • Asset-selective cross-attention mask M_as no independent evidence
    purpose: Lets layout tokens use view-selective visibility while each asset attends only to its OBB-projected ROI.
    Stage 3 hybrid attention (Eqs. 8–9); purely architectural.
  • Partial Scene Geometry (PSG) + Layout-Guided Structure Inversion no independent evidence
    purpose: Provides a spatial prior that initializes the flow-matching trajectory at intermediate noise level t0 < 1.
    Inference-time technique (Sec. 4.2); depends on depth quality.
  • ERP-FRONT dataset independent evidence
    purpose: Supplies the paired ERP–mesh supervision required to train and evaluate the pipeline.
    New synthetic resource built from 3D-FRONT (Sec. 5); independent of the model once released.

pith-pipeline@v1.1.0-grok45 · 30433 in / 3645 out tokens · 31974 ms · 2026-07-11T22:28:24.501398+00:00 · methodology

0 comments
read the original abstract

Recent advances in single image-to-3D generation have enabled high-quality asset synthesis, yet extending these capabilities to indoor scene generation remains challenging. Existing methods focus on asset-level generation while neglecting the structural layout, which is essential for downstream applications and serves as the spatial anchor for grounding assets. However, a single image with a limited field of view lacks the spatial coverage to recover a coherent global layout. To this end, we use a 360{\deg} image represented in equirectangular projection (ERP) and propose InSpace, a structure-aware framework for 3D indoor scene generation. InSpace comprises three stages: (1) estimating partial scene geometry as spatial priors, (2) generating coarse scene structure with view-selective cross-attention, and (3) producing detailed layout and asset geometry with textures through a global-local hybrid attention, using flow matching. We also propose ERP-FRONT, a paired ERP-Image-to-3D indoor scene dataset based on 3D-FRONT. Experiments show that InSpace generates complete 3D indoor scenes with structural layout, along with separate textured assets from a single ERP image, achieving strong performance across 3D and 2D metrics. Project Page: https://kookie12.github.io/InSpace-Project-Page/

Figures

Figures reproduced from arXiv: 2607.03990 by Chang D. Yoo, Gwanhyeong Koo, Hyunsu Kim, Siwoo Lim, Sunjae Yoon, Suyong Yeon, Taejae Lee, Youngji Kim.

Figure 1
Figure 1. Figure 1: InSpace generates complete 3D indoor scenes with structural layout and indi￾vidual assets from a single 360° image. Leveraging the full 360° context of panoramic images, it produces spatially coherent scenes across diverse room layouts. Abstract. Recent advances in single image-to-3D generation have en￾abled high-quality asset synthesis, yet extending these capabilities to indoor scene generation remains c… view at source ↗
Figure 2
Figure 2. Figure 2: (a) Existing single-image methods generate individual assets without structural layout, causing floating, misplacement, and artifacts. (b) InSpace uses an ERP image to generate complete indoor scenes with structural layout and well-grounded assets. 1 Introduction Recent advances in generative modeling have dramatically accelerated 3D con￾tent creation from visual inputs. In particular, single-image-to-3D g… view at source ↗
Figure 3
Figure 3. Figure 3: Indoor Scene Geometry Initialization. From a single ERP image and its estimated depth, we lift the scene to 3D via equirectangular back-projection, producing an initial point cloud with the camera center at the origin (0, 0, 0). The point cloud is then normalized into the canonical space [−0.5, 0.5]3 , yielding the Partial Scene Geometry (PSG) and shifting the camera center to a calibrated position c. Seco… view at source ↗
Figure 4
Figure 4. Figure 4: Coarse Scene Geometry Generation. Six cubemap faces (FoV 120) are extracted from the ERP image and encoded with view position embeddings. The model denoises voxel latents via DiT blocks with view-selective cross-attention, where each voxel attends only to its visible faces based on camera center c. At inference, Layout￾Guided Structure Inversion optionally initializes the latent from the PSG. The decoded c… view at source ↗
Figure 5
Figure 5. Figure 5: View-Selective Cross-Attention. (a) For each voxel at position pi, we com￾pute the cosine similarity between the direction di from camera center c and each cubemap face normal nˆf to determine visibility. (b) Voxels are partitioned by their visible faces, shown in corresponding colors. (c) The resulting mask Mvs assigns each voxel high attention weights (dark) only to its visible cubemap tokens. at face bo… view at source ↗
Figure 6
Figure 6. Figure 6: Layout and Asset-Aware Scene Generation. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: ERP-FRONT Dataset. Each room from 3D-FRONT is decomposed into (a) the complete indoor scene, (b) individual assets, (c) structural layout, and (d) 3D oriented bounding boxes. (e) ERP images with depth maps are rendered from valid camera positions in (f) the floorplan. (g) Room type distribution across the dataset. where Q ∈ R Ntotal×dh are voxel queries from the unified token sequence, and K, V ∈ R T ×dh a… view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative results on ERP-FRONT. Given an ERP image and its cube￾map decomposition (top), InSpace generates the overall scene, structural layout, and individual assets. Four examples are shown with ground truth for comparison. ground truth. To assess generalization beyond the synthetic ERP-FRONT, we further test on ReplicaPano [12], which provides ERP images extracted from the realistic Replica dataset [5… view at source ↗
Figure 9
Figure 9. Figure 9: Rendered evaluation views on ERP-FRONT. (a) Coarse structure (Stage 2) and (b) full scene generation (Stage 3), rendered from a top-down view and six inte￾rior views from the camera center along cubemap directions for quantitative evaluation. Cyan dots indicate the calibrated camera center [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Ablation on Stage 2 across camera positions. [PITH_FULL_IMAGE:figures/full_fig_p015_10.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

75 extracted references · 19 linked inside Pith

  1. [1]

    ai et al

    Ai, H., Cao, Z., Wang, L.: A survey of representation learning, optimization strate- gies, and applications for omnidirectional vision: H. ai et al. International Journal of Computer Vision133(8), 4973–5012 (2025) 2

    2025

  2. [2]

    arXiv preprint arXiv:2205.10468 (2022) 2

    Ai, H., Cao, Z., Zhu, J., Bai, H., Chen, Y., Wang, L.: Deep learning for omnidi- rectional vision: A survey and new perspectives. arXiv preprint arXiv:2205.10468 (2022) 2

    Pith/arXiv arXiv 2022

  3. [3]

    In: Sensor fusion IV: control paradigms and data structures

    Besl, P.J., McKay, N.D.: Method for registration of 3-d shapes. In: Sensor fusion IV: control paradigms and data structures. vol. 1611, pp. 586–606. Spie (1992) 12

    1992

  4. [4]

    In: European conference on computer vision

    Carion, N., Massa, F., Synnaeve, G., Usunier, N., Kirillov, A., Zagoruyko, S.: End- to-end object detection with transformers. In: European conference on computer vision. pp. 213–229. Springer (2020) 3

    2020

  5. [5]

    In: 2017 International Conference on 3D Vision (3DV)

    Chang, A., Dai, A., Funkhouser, T., Halber, M., Niebner, M., Savva, M., Song, S., Zeng, A., Zhang, Y.: Matterport3d: Learning from rgb-d data in indoor environ- ments. In: 2017 International Conference on 3D Vision (3DV). pp. 667–676. IEEE Computer Society (2017) 3

    2017

  6. [6]

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

    Chen, X., Chu, F.J., Gleize, P., Liang, K.J., Sax, A., Tang, H., Wang, W., Guo, M., Hardin, T., Li, X., et al.: Sam 3d: 3dfy anything in images. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 7220–7232 (2026) 2, 4, 12, 35, 37

    2026

  7. [7]

    arXiv preprint arXiv:2507.17745 (2025) 4 InSpace 17

    Chen, Y., Li, Z., Wang, Y., Zhang, H., Li, Q., Zhang, C., Lin, G.: Ultra3d: Efficient and high-fidelity 3d generation with part attention. arXiv preprint arXiv:2507.17745 (2025) 4 InSpace 17

    Pith/arXiv arXiv 2025

  8. [8]

    In: ICLR 2023 (Eleventh International Conference on Learning Representations) (2023) 8

    Couairon, G., Verbeek, J., Schwenk, H., Cord, M.: Diffedit: Diffusion-based se- mantic image editing with mask guidance. In: ICLR 2023 (Eleventh International Conference on Learning Representations) (2023) 8

    2023

  9. [9]

    arXiv preprint arXiv:1911.01911 (2019) 11, 24

    Denninger, M., Sundermeyer, M., Winkelbauer, D., Zidan, Y., Olefir, D., El- badrawy, M., Lodhi, A., Katam, H.: Blenderproc. arXiv preprint arXiv:1911.01911 (2019) 11, 24

    Pith/arXiv arXiv 1911

  10. [10]

    Ding, L., Dong, S., Li, Y., Gao, C., Chen, X., Han, R., Kuang, Y., Zhang, H., Huang, B., Huang, Z., Wang, Z., Xu, D., Xue, T.: Fullpart: Generating each 3d part at full resolution (2025),https://arxiv.org/abs/2510.261405

    arXiv 2025

  11. [11]

    In: Proceedings of the 33rd ACM International Conference on Multimedia

    Dong, W., Yang, B., Yang, Z., Li, Y., Hu, T., Bao, H., Ma, Y., Cui, Z.: Hiscene: creating hierarchical 3d scenes with isometric view generation. In: Proceedings of the 33rd ACM International Conference on Multimedia. pp. 9783–9792 (2025) 4

    2025

  12. [12]

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

    Dong, Y., Fang, C., Bo, L., Dong, Z., Tan, P.: Panocontext-former: Panoramic total scene understanding with a transformer. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 28087–28097 (2024) 3, 4, 12, 13, 32

    2024

  13. [13]

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

    Fu, H., Cai, B., Gao, L., Zhang, L.X., Wang, J., Li, C., Zeng, Q., Sun, C., Jia, R., Zhao, B., et al.: 3d-front: 3d furnished rooms with layouts and semantics. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 10933–10942 (2021) 3, 11, 23

    2021

  14. [14]

    International Journal of Computer Vision129(12), 3313–3337 (2021) 23

    Fu, H., Jia, R., Gao, L., Gong, M., Zhao, B., Maybank, S., Tao, D.: 3d-future: 3d furniture shape with texture. International Journal of Computer Vision129(12), 3313–3337 (2021) 23

    2021

  15. [15]

    ACM Trans- actions on Graphics (TOG)43(4), 1–15 (2024) 4

    Gao, D., Rozenberszki, D., Leutenegger, S., Dai, A.: Diffcad: Weakly-supervised probabilistic cad model retrieval and alignment from an rgb image. ACM Trans- actions on Graphics (TOG)43(4), 1–15 (2024) 4

    2024

  16. [16]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Gu, Z., Cui, Y., Li, Z., Wei, F., Ge, Y., Gu, J., Liu, M.Y., Davis, A., Ding, Y.: Ar- tiscene: Language-driven artistic 3d scene generation through image intermediary. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 2891–2901 (2025) 2, 4

    2025

  17. [17]

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

    Gümeli, C., Dai, A., Nießner, M.: Roca: Robust cad model retrieval and alignment from a single image. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 4022–4031 (2022) 4

    2022

  18. [18]

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

    He, X., Zou, Z.X., Chen, C.H., Guo, Y.C., Liang, D., Yuan, C., Ouyang, W., Cao, Y.P., Li, Y.: Sparseflex: High-resolution and arbitrary-topology 3d shape modeling. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 14822–14833 (2025) 4

    2025

  19. [19]

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

    Hong, J.W., Ton, T., Pham, T.X., Koo, G., Yoon, S., Yoo, C.D.: Ita-mdt: Image- timestep-adaptive masked diffusion transformer framework for image-based virtual try-on. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 28284–28294 (2025) 37

    2025

  20. [20]

    In: International Conference on Learning Representations

    Hong, Y., Zhang, K., Gu, J., Bi, S., Zhou, Y., Liu, D., Liu, F., Sunkavalli, K., Bui, T., Tan, H.: Lrm: Large reconstruction model for single image to 3d. In: International Conference on Learning Representations. vol. 2024, pp. 50678–50702 (2024) 2

    2024

  21. [21]

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

    Huang, Z., Guo, Y.C., An, X., Yang, Y., Li, Y., Zou, Z.X., Liang, D., Liu, X., Cao, Y.P., Sheng, L.: Midi: Multi-instance diffusion for single image to 3d scene generation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 23646–23657 (2025) 2, 4, 12, 35, 37 18 G. Koo et al

    2025

  22. [22]

    arXiv preprint arXiv:2506.15442 (2025) 2, 4, 7

    Hunyuan3D, T., Yang, S., Yang, M., Feng, Y., Huang, X., Zhang, S., He, Z., Luo, D., Liu, H., Zhao, Y., 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, 7

    Pith/arXiv arXiv 2025

  23. [23]

    In: International conference on machine learning

    Jaegle, A., Gimeno, F., Brock, A., Vinyals, O., Zisserman, A., Carreira, J.: Per- ceiver: General perception with iterative attention. In: International conference on machine learning. pp. 4651–4664. PMLR (2021) 3

    2021

  24. [24]

    Jia, T., Yan, D., Hao, D., Li, Y., Zhang, K., He, X., Li, L., Wang, Y., Chen, J., Jiang, L., Yin, Q., Quan, L., Chen, Y.C., Yuan, L.: Ultrashape 1.0: High-fidelity 3d shape generation via scalable geometric refinement (2025),https://arxiv.org/ abs/2512.211854

    arXiv 2025

  25. [25]

    arXiv preprint arXiv:1312.6114 (2013) 3

    Kingma, D.P., Welling, M.: Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114 (2013) 3

    Pith/arXiv arXiv 2013

  26. [26]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.Y., et al.: Segment anything. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 4015–4026 (2023) 12

    2023

  27. [27]

    In: European Conference on Computer Vision

    Koo, G., Yoon, S., Hong, J.W., Yoo, C.D.: Flexiedit: Frequency-aware latent re- finement for enhanced non-rigid editing. In: European Conference on Computer Vision. pp. 363–379. Springer (2024) 8, 37

    2024

  28. [28]

    In: International Conference on Machine Learning

    Koo, G., Yoon, S., Lee, Y., Hong, J.W., Yoo, C.D.: Flowdrag: 3d-aware drag-based image editing with mesh-guided deformation vector flow fields. In: International Conference on Machine Learning. pp. 31467–31485. PMLR (2025) 8, 37

    2025

  29. [29]

    In: ICASSP 2024-2024 IEEE International Confer- ence on Acoustics, Speech and Signal Processing (ICASSP)

    Koo, G., Yoon, S., Yoo, C.D.: Wavelet-guided acceleration of text inversion in diffusion-based image editing. In: ICASSP 2024-2024 IEEE International Confer- ence on Acoustics, Speech and Signal Processing (ICASSP). pp. 4380–4384. IEEE (2024) 37

    2024

  30. [30]

    In: European Conference on Computer Vision

    Kuo, W., Angelova, A., Lin, T.Y., Dai, A.: Mask2cad: 3d shape prediction by learning to segment and retrieve. In: European Conference on Computer Vision. pp. 260–277. Springer (2020) 4

    2020

  31. [31]

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

    Kuo, W., Angelova, A., Lin, T.Y., Dai, A.: Patch2cad: Patchwise embedding learn- ing for in-the-wild shape retrieval from a single image. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 12589–12599 (2021) 4

    2021

  32. [32]

    1 kontext: Flow match- ing for in-context image generation and editing in latent space

    Labs, B.F., Batifol, S., Blattmann, A., Boesel, F., Consul, S., Diagne, C., Dock- horn, T., English, J., English, Z., Esser, P., et al.: Flux. 1 kontext: Flow match- ing for in-context image generation and editing in latent space. arXiv preprint arXiv:2506.15742 (2025) 3

    Pith/arXiv arXiv 2025

  33. [33]

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

    Lai, Z., Zhao, Y., Zhao, Z., Liu, H., Lin, Q., Huang, J., Guo, C., Yue, X.: Lattice: Democratize high-fidelity 3d generation at scale. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 19982–19992 (2026) 4

    2026

  34. [34]

    arXiv preprint arXiv:2503.16302 (2025) 2, 4

    Lai, Z., Zhao, Y., Zhao, Z., Liu, H., Wang, F., Shi, H., Yang, X., Lin, Q., Huang, J., Liu, Y., et al.: Unleashing vecset diffusion model for fast shape generation. arXiv preprint arXiv:2503.16302 (2025) 2, 4

    Pith/arXiv arXiv 2025

  35. [35]

    In: Proceed- ings of the European conference on computer vision (ECCV)

    Law, H., Deng, J.: Cornernet: Detecting objects as paired keypoints. In: Proceed- ings of the European conference on computer vision (ECCV). pp. 734–750 (2018) 27

    2018

  36. [36]

    arXiv preprint arXiv:2605.10922 (2026) 4 InSpace 19

    Li, D.Y., Zhao, W., Chen, Y., Hu, W., Guo, M.H., Zhang, F.L., Shan, Y., Hu, S.M.: Pixal3d: Pixel-aligned 3d generation from images. arXiv preprint arXiv:2605.10922 (2026) 4 InSpace 19

    Pith/arXiv arXiv 2026

  37. [37]

    Li, H., Zheng, W., He, J., Liu, Y., Lin, X., Yang, X., Chen, Y.C., Guo, C.: Da2: Depth anything in any direction (2025),https://arxiv.org/abs/2509.266182, 6

    arXiv 2025

  38. [38]

    arXiv preprint arXiv:2405.14979 (2024) 2, 4

    Li, W., Liu, J., Chen, R., Liang, Y., Chen, X., Tan, P., Long, X.: Craftsman: High-fidelity mesh generation with 3d native generation and interactive geometry refiner. arXiv preprint arXiv:2405.14979 (2024) 2, 4

    Pith/arXiv arXiv 2024

  39. [39]

    arXiv preprint arXiv:2505.07747 (2025) 2, 4

    Li, W., Zhang, X., Sun, Z., Qi, D., Li, H., Cheng, W., Cai, W., Wu, S., Liu, J., Wang, Z., et al.: Step1x-3d: Towards high-fidelity and controllable generation of textured 3d assets. arXiv preprint arXiv:2505.07747 (2025) 2, 4

    Pith/arXiv arXiv 2025

  40. [40]

    IEEE Transactions on Pattern Analysis and Machine Intel- ligence (2025) 2, 4

    Li, Y., Zou, Z.X., Liu, Z., Wang, D., Liang, Y., Yu, Z., Liu, X., Guo, Y.C., Liang, D., Ouyang, W., et al.: Triposg: High-fidelity 3d shape synthesis using large-scale rectified flow models. IEEE Transactions on Pattern Analysis and Machine Intel- ligence (2025) 2, 4

    2025

  41. [41]

    Advances in Neural Informa- tion Processing Systems38, 118582–118600 (2026) 4

    Li, Z., Wang, Y., Zheng, H., Luo, Y., Wen, B.: Sparc3d: Sparse representation and construction for high-resolution 3d shapes modeling. Advances in Neural Informa- tion Processing Systems38, 118582–118600 (2026) 4

    2026

  42. [42]

    Lin, Y., Lin, C., Pan, P., Yan, H., Feng, Y., Mu, Y., Fragkiadaki, K.: Partcrafter: Structured 3d mesh generation via compositional latent diffusion transformers (2025),https://arxiv.org/abs/2506.055734

    Pith/arXiv arXiv 2025

  43. [43]

    Ling, L., Lin, C.H., Lin, T.Y., Ding, Y., Zeng, Y., Sheng, Y., Ge, Y., Liu, M.Y., Bera, A., Li, Z.: Scenethesis: A language and vision agentic framework for 3d scene generation (2025),https://arxiv.org/abs/2505.028362, 4

    Pith/arXiv arXiv 2025

  44. [44]

    arXiv preprint arXiv:2210.02747 (2022) 3, 5

    Lipman, Y., Chen, R.T., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. arXiv preprint arXiv:2210.02747 (2022) 3, 5

    Pith/arXiv arXiv 2022

  45. [45]

    In: Proceedings of the IEEE/CVF inter- national conference on computer vision

    Liu, R., Wu, R., Van Hoorick, B., Tokmakov, P., Zakharov, S., Vondrick, C.: Zero- 1-to-3: Zero-shot one image to 3d object. In: Proceedings of the IEEE/CVF inter- national conference on computer vision. pp. 9298–9309 (2023) 2

    2023

  46. [46]

    In: The Eleventh International Conference on Learning Representations (ICLR) (2023) 11

    Liu, X., Gong, C., Liu, Q.: Flow straight and fast: Learning to generate and transfer data with rectified flow. In: The Eleventh International Conference on Learning Representations (ICLR) (2023) 11

    2023

  47. [47]

    In: European conference on computer vision

    Liu, Y., Wang, T., Zhang, X., Sun, J.: Petr: Position embedding transformation for multi-view 3d object detection. In: European conference on computer vision. pp. 531–548. Springer (2022) 3, 9

    2022

  48. [48]

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

    Lugmayr, A., Danelljan, M., Romero, A., Yu, F., Timofte, R., Van Gool, L.: Re- paint: Inpainting using denoising diffusion probabilistic models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 11461– 11471 (2022) 8

    2022

  49. [49]

    arXiv preprint arXiv:2108.01073 (2021) 8

    Meng, C., He, Y., Song, Y., Song, J., Wu, J., Zhu, J.Y., Ermon, S.: Sdedit: Guided image synthesis and editing with stochastic differential equations. arXiv preprint arXiv:2108.01073 (2021) 8

    Pith/arXiv arXiv 2021

  50. [50]

    In: 2026 International Conference on 3D Vision (3DV)

    Meng, Y., Wu, H., Zhang, Y., Xie, W.: Scenegen: Single-image 3d scene generation in one feedforward pass. In: 2026 International Conference on 3D Vision (3DV). pp. 543–553. IEEE (2026) 2, 4, 12, 35, 37

    2026

  51. [51]

    Peebles,W.,Xie,S.:Scalablediffusionmodelswithtransformers.In:Proceedingsof the IEEE/CVF international conference on computer vision. pp. 4195–4205 (2023) 3, 5

    2023

  52. [52]

    Advances in neural information processing systems32(2019) 3

    Razavi, A., Van den Oord, A., Vinyals, O.: Generating diverse high-fidelity images with vq-vae-2. Advances in neural information processing systems32(2019) 3

    2019

  53. [53]

    In: Proceedings of the 20 G

    Ren, X., Huang, J., Zeng, X., Museth, K., Fidler, S., Williams, F.: Xcube: Large- scale 3d generative modeling using sparse voxel hierarchies. In: Proceedings of the 20 G. Koo et al. IEEE/CVF conference on computer vision and pattern recognition. pp. 4209–4219 (2024) 4

    2024

  54. [54]

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

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10684–10695 (2022) 3

    2022

  55. [55]

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

    Shi, Y., Xue, C., Liew, J.H., Pan, J., Yan, H., Zhang, W., Tan, V.Y., Bai, S.: Dragdiffusion: Harnessing diffusion models for interactive point-based image edit- ing. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 8839–8849 (2024) 8

    2024

  56. [56]

    arXiv preprint arXiv:2508.10104 (2025) 7, 12

    Siméoni, O., Vo, H.V., Seitzer, M., Baldassarre, F., Oquab, M., Jose, C., Khali- dov, V., Szafraniec, M., Yi, S., Ramamonjisoa, M., et al.: Dinov3. arXiv preprint arXiv:2508.10104 (2025) 7, 12

    Pith/arXiv arXiv 2025

  57. [57]

    arXiv preprint arXiv:1906.05797 (2019) 3, 13, 32

    Straub, J., Whelan, T., Ma, L., Chen, Y., Wijmans, E., Green, S., Engel, J.J., Mur-Artal, R., Ren, C., Verma, S., et al.: The replica dataset: A digital replica of indoor spaces. arXiv preprint arXiv:1906.05797 (2019) 3, 13, 32

    Pith/arXiv arXiv 1906

  58. [58]

    Advances in Neural Information Processing Systems38, 170778–170804 (2026) 2, 4

    Wu, S., Lin, Y., Zhang, F., Zeng, Y., Yang, Y., Qian, J., Zhu, S., Cao, X., Torr, P., Yao, Y., et al.: Direct3d-s2: Gigascale 3d generation made easy with spatial sparse attention. Advances in Neural Information Processing Systems38, 170778–170804 (2026) 2, 4

    2026

  59. [59]

    arXiv preprint arXiv:2512.14692 (2025) 2, 3, 4, 5, 6, 7, 12, 32

    Xiang, J., Chen, X., Xu, S., Wang, R., Lv, Z., Deng, Y., Zhu, H., Dong, Y., Zhao, H., Yuan, N.J., et al.: Native and compact structured latents for 3d generation. arXiv preprint arXiv:2512.14692 (2025) 2, 3, 4, 5, 6, 7, 12, 32

    Pith/arXiv arXiv 2025

  60. [60]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Xiang, J., Lv, Z., Xu, S., Deng, Y., Wang, R., Zhang, B., Chen, D., Tong, X., Yang, J.: Structured 3d latents for scalable and versatile 3d generation. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 21469–21480 (2025) 2, 4

    2025

  61. [61]

    arXiv preprint arXiv:2404.07191 (2024) 2

    Xu, J., Cheng, W., Gao, Y., Wang, X., Gao, S., Shan, Y.: Instantmesh: Efficient 3d mesh generation from a single image with sparse-view large reconstruction models. arXiv preprint arXiv:2404.07191 (2024) 2

    Pith/arXiv arXiv 2024

  62. [62]

    Advances in neural information processing systems38, 140319–140351 (2026) 2, 4

    Yang, Y., Jia, B., Zhang, S., Huang, S.: Sceneweaver: All-in-one 3d scene syn- thesis with an extensible and self-reflective agent. Advances in neural information processing systems38, 140319–140351 (2026) 2, 4

    2026

  63. [63]

    In: Proceedings of the SIGGRAPH Asia 2025 Conference Papers

    Yang, Y., Zhou, Y., Guo, Y.C., Zou, Z.X., Huang, Y., Liu, Y.T., Xu, H., Liang, D., Cao, Y.P., Liu, X.: Omnipart: Part-aware 3d generation with semantic decoupling and structural cohesion. In: Proceedings of the SIGGRAPH Asia 2025 Conference Papers. pp. 1–12 (2025) 5, 9, 10

    2025

  64. [64]

    ACM Transactions on Graphics (TOG)44(4), 1–19 (2025) 2, 4

    Yao, K., Zhang, L., Yan, X., Zeng, Y., Zhang, Q., Xu, L., Yang, W., Gu, J., Yu, J.: Cast: Component-aligned 3d scene reconstruction from an rgb image. ACM Transactions on Graphics (TOG)44(4), 1–19 (2025) 2, 4

    2025

  65. [65]

    In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision

    Ye, C., Wu, Y., Lu, Z., Chang, J., Guo, X., Zhou, J., Zhao, H., Han, X.: Hi3dgen: High-fidelity 3d geometry generation from images via normal bridging. In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision. pp. 25050– 25061 (2025) 2, 4

    2025

  66. [66]

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

    Yin, T., Zhou, X., Krahenbuhl, P.: Center-based 3d object detection and track- ing. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 11784–11793 (2021) 3, 9, 12, 26

    2021

  67. [67]

    In: European Conference on Computer Vision

    Yoon, S., Koo, G., Hong, J.W., Yoo, C.D.: Dni: Dilutional noise initialization for diffusion video editing. In: European Conference on Computer Vision. pp. 180–195. Springer (2024) 8, 37 InSpace 21

    2024

  68. [68]

    In: International Conference on Machine Learning

    Yoon,S.,Koo,G.,Kim,G.,Yoo,C.D.:Frag:Frequencyadaptinggroupfordiffusion video editing. In: International Conference on Machine Learning. pp. 57315–57330. PMLR (2024) 37

    2024

  69. [69]

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

    Yoon, S., Koo, G., Lee, Y., Hong, J.W., Yoo, C.D.: Occlusion-robust stylization for drawing-based 3d animation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 12263–12273 (2025) 37

    2025

  70. [70]

    Advances in Neural Information Process- ing Systems37, 118654–118677 (2024) 37

    Yoon, S., Koo, G., Lee, Y., Yoo, C.: Tpc: Test-time procrustes calibration for diffusion-based human image animation. Advances in Neural Information Process- ing Systems37, 118654–118677 (2024) 37

    2024

  71. [71]

    ACM Transactions On Graphics (TOG)42(4), 1–16 (2023) 3

    Zhang, B., Tang, J., Niessner, M., Wonka, P.: 3dshape2vecset: A 3d shape repre- sentation for neural fields and generative diffusion models. ACM Transactions On Graphics (TOG)42(4), 1–16 (2023) 3

    2023

  72. [72]

    ACM Transactions on Graphics (TOG)43(4), 1–20 (2024) 2, 4

    Zhang, L., Wang, Z., Zhang, Q., Qiu, Q., Pang, A., Jiang, H., Yang, W., Xu, L., Yu, J.: Clay: A controllable large-scale generative model for creating high-quality 3d assets. ACM Transactions on Graphics (TOG)43(4), 1–20 (2024) 2, 4

    2024

  73. [73]

    arXiv preprint arXiv:2501.12202 (2025) 2, 4

    Zhao, Z., Lai, Z., Lin, Q., Zhao, Y., Liu, H., Yang, S., Feng, Y., Yang, M., Zhang, S., Yang, X., et al.: Hunyuan3d 2.0: Scaling diffusion models for high resolution textured 3d assets generation. arXiv preprint arXiv:2501.12202 (2025) 2, 4

    Pith/arXiv arXiv 2025

  74. [74]

    Advances in neural information processing sys- tems36, 73969–73982 (2023) 2, 4

    Zhao, Z., Liu, W., Chen, X., Zeng, X., Wang, R., Cheng, P., Fu, B., Chen, T., Yu, G., Gao, S.: Michelangelo: Conditional 3d shape generation based on shape-image- text aligned latent representation. Advances in neural information processing sys- tems36, 73969–73982 (2023) 2, 4

    2023

  75. [75]

    In: European Conference on Computer Vision

    Zheng, J., Zhang, J., Li, J., Tang, R., Gao, S., Zhou, Z.: Structured3d: A large photo-realistic dataset for structured 3d modeling. In: European Conference on Computer Vision. pp. 519–535. Springer (2020) 3 22 G. Koo et al. Supplementary Material for InSpace: Structure-Aware 3D Indoor Scene Generation from a Single 360°Image Fig. A.1:InSpace generates co...

    2020