pith. sign in

arxiv: 2604.01972 · v3 · submitted 2026-04-02 · 💻 cs.CV

SDesc3D: Towards Layout-Aware 3D Indoor Scene Generation from Short Descriptions

Jie Feng , Jiawei Shen , Junjia Huang , Junpeng Zhang , Mingtao Feng , Weisheng Dong , Guanbin Li This is my paper

Pith reviewed 2026-05-13 22:07 UTC · model grok-4.3

classification 💻 cs.CV
keywords 3D scene generationtext-conditioned generationindoor sceneslayout reasoningshort descriptionsmulti-view priorsfunctionality groundingscene plausibility
0
0 comments X

The pith

Short text descriptions suffice to generate physically plausible 3D indoor scenes once multi-view structural priors and regional functionality cues are added to supply missing layout relations.

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

The paper presents SDesc3D, a framework that generates 3D indoor scenes from brief textual prompts by enriching those prompts with aggregated multi-view structural knowledge and regional functionality implications. Existing approaches fail when descriptions omit explicit objects and spatial relations, producing implausible layouts. SDesc3D counters this with multi-view prior augmentation, functionality-aware layout grounding that creates implicit spatial anchors, and an iterative reflection-rectification loop that progressively refines structural plausibility. A reader would care because the method removes the need for labor-intensive layout specifications, opening easier routes to interactive 3D environment creation.

Core claim

We propose SDesc3D, a short-text conditioned 3D indoor scene generation framework that leverages multi-view structural priors and regional functionality implications to enable 3D layout reasoning under sparse textual guidance. Multi-view scene prior augmentation enriches underspecified inputs by shifting from inaccessible semantic relation cues to aggregated multi-view relational priors. Functionality-aware layout grounding then employs regional functionality for implicit spatial anchors and conducts hierarchical layout reasoning to improve scene organization. An iterative reflection-rectification scheme progressively refines structural plausibility via self-rectification. Experiments show S

What carries the argument

Multi-view scene prior augmentation that aggregates relational knowledge across views to replace missing semantic cues, combined with functionality-aware layout grounding that supplies implicit spatial anchors through regional functionality analysis.

If this is right

  • Generated scenes exhibit higher physical plausibility and richer semantic detail than those from prior short-text methods.
  • Hierarchical layout reasoning produces better-organized room structures without explicit user-supplied relations.
  • Iterative self-rectification progressively reduces implausible arrangements during generation.
  • The approach enables scene creation from prompts that lack object counts, positions, or connectivity information.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same prior-augmentation strategy could support scene generation from single-word prompts by pulling in even broader structural databases.
  • Functionality grounding might transfer to outdoor or mixed indoor-outdoor environments if regional activity maps are available.
  • Integration into design software could let users iterate scenes by editing short text rather than direct 3D manipulation.

Load-bearing premise

Aggregated multi-view structural priors and regional functionality implications can reliably supply the spatial and relational details absent from sparse short-text descriptions.

What would settle it

Generate scenes from a fixed set of short descriptions, measure physical violations such as object intersections or unsupported placements against ground-truth layouts, and check whether the method shows no measurable reduction in violations compared with prior text-to-3D baselines.

Figures

Figures reproduced from arXiv: 2604.01972 by Guanbin Li, Jiawei Shen, Jie Feng, Junjia Huang, Junpeng Zhang, Mingtao Feng, Weisheng Dong.

Figure 1
Figure 1. Figure 1: Short descriptions condense semantics, making 3D indoor scene generation challenging in terms of physical plausibility [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of our SDesc3D Framework. Given a short user descriptions, SDesc3D first performs Multi-view Scene [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Qualitive comparison on the scenes generated on five different short descriptions. Our method achieves better overall [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison of HSM, Reason3D, and our [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Examples of scene editing results of object addition, [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
read the original abstract

3D indoor scene generation conditioned on short textual descriptions provides a promising avenue for interactive 3D environment construction without the need for labor-intensive layout specification. Despite recent progress in text-conditioned 3D scene generation, existing works suffer from poor physical plausibility and insufficient detail richness in such semantic condensation cases, largely due to their reliance on explicit semantic cues about compositional objects and their spatial relationships. This limitation highlights the need for enhanced 3D reasoning capabilities, particularly in terms of prior integration and spatial anchoring. Motivated by this, we propose SDesc3D, a short-text conditioned 3D indoor scene generation framework, that leverages multi-view structural priors and regional functionality implications to enable 3D layout reasoning under sparse textual guidance. Specifically, we introduce a Multi-view scene prior augmentation that enriches underspecified textual inputs with aggregated multi-view structural knowledge, shifting from inaccessible semantic relation cues to multi-view relational prior aggregation. Building on this, we design a Functionality-aware layout grounding, employing regional functionality grounding for implicit spatial anchors and conducting hierarchical layout reasoning to enhance scene organization and semantic plausibility. Furthermore, an Iterative reflection-rectification scheme is employed for progressive structural plausibility refinement via self-rectification. Extensive experiments show that our method outperforms existing approaches on short-text conditioned 3D indoor scene generation. Code will be publicly available.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes SDesc3D, a framework for 3D indoor scene generation conditioned on short textual descriptions. It introduces a Multi-view scene prior augmentation module to enrich underspecified inputs via aggregated structural knowledge, a Functionality-aware layout grounding component that uses regional functionality implications and hierarchical reasoning for implicit spatial anchors, and an Iterative reflection-rectification scheme for progressive plausibility refinement. The central claim is that these components enable superior performance over existing approaches on short-text conditioned 3D indoor scene generation.

Significance. If the quantitative claims hold, the work would advance text-conditioned 3D scene generation by addressing the practical challenge of sparse inputs, which is relevant for interactive applications. The explicit separation of prior aggregation from semantic cues and the grounding mechanism represent a targeted response to a known limitation in the field.

major comments (2)
  1. [§3.2] §3.2 (Functionality-aware layout grounding): The claim that regional functionality implications supply reliable implicit spatial anchors for truly sparse short-text inputs (lacking any object or layout cues) is load-bearing for the outperformance assertion, yet the section provides no concrete validation, failure-case analysis, or coverage statistics on how these priors resolve ambiguities when textual guidance is minimal.
  2. [§4] §4 (Experiments): The abstract and method sections assert that extensive experiments demonstrate outperformance, but no quantitative metrics (e.g., FID, layout accuracy), baselines specific to short-text cases, ablation results isolating multi-view augmentation or functionality grounding, or dataset details for sparse inputs are referenced, preventing evaluation of whether the upstream grounding succeeds as assumed.
minor comments (2)
  1. [Abstract] Abstract: The statement 'Code will be publicly available' should include a specific repository link or release timeline for reproducibility.
  2. [§3] Notation throughout §3: The distinction between 'multi-view relational prior aggregation' and 'regional functionality grounding' would benefit from a single diagram or pseudocode block to clarify the data flow.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We sincerely thank the referee for the constructive and detailed feedback on our manuscript. We address each major comment point by point below, acknowledging areas where additional substantiation is warranted, and commit to revisions that will strengthen the presentation without altering the core claims.

read point-by-point responses

  1. Referee: [§3.2] §3.2 (Functionality-aware layout grounding): The claim that regional functionality implications supply reliable implicit spatial anchors for truly sparse short-text inputs (lacking any object or layout cues) is load-bearing for the outperformance assertion, yet the section provides no concrete validation, failure-case analysis, or coverage statistics on how these priors resolve ambiguities when textual guidance is minimal.

    Authors: We appreciate the referee's emphasis on this critical aspect of the Functionality-aware layout grounding module. The design relies on regional functionality implications combined with hierarchical reasoning to derive implicit spatial anchors from multi-view structural priors when textual cues are minimal. We agree that §3.2 currently lacks explicit quantitative validation, failure-case analysis, and coverage statistics demonstrating ambiguity resolution. In the revised manuscript, we will expand this section with coverage statistics on how the priors address sparse inputs, a set of representative failure cases, and qualitative/quantitative examples illustrating the hierarchical reasoning process. These additions will provide direct evidence supporting the load-bearing claim. revision: yes

  2. Referee: [§4] §4 (Experiments): The abstract and method sections assert that extensive experiments demonstrate outperformance, but no quantitative metrics (e.g., FID, layout accuracy), baselines specific to short-text cases, ablation results isolating multi-view augmentation or functionality grounding, or dataset details for sparse inputs are referenced, preventing evaluation of whether the upstream grounding succeeds as assumed.

    Authors: We thank the referee for identifying this gap in experimental transparency. The experiments section reports quantitative results on short-text conditioned generation, but we acknowledge that the abstract and method sections do not sufficiently reference the specific metrics (such as FID and layout accuracy), short-text-specific baselines, ablations isolating the multi-view augmentation and functionality grounding modules, or dataset characteristics for sparse inputs. In the revision, we will update the abstract to highlight key quantitative metrics, add explicit ablation tables isolating each proposed component, include baselines adapted for short-text scenarios, and provide dataset details focused on sparse input cases. This will allow readers to directly evaluate the grounding mechanism's contribution. revision: yes

Circularity Check

0 steps flagged

No circularity: method relies on external priors and experimental validation without self-referential fitting or definitional loops

full rationale

The paper's core claims rest on introducing Multi-view scene prior augmentation and Functionality-aware layout grounding to handle sparse text inputs, followed by iterative refinement and experimental outperformance. No equations, parameter-fitting steps, or self-citations are presented that reduce predictions or uniqueness claims back to the same fitted inputs or prior author results by construction. The derivation chain is self-contained against external benchmarks and priors, with no evidence of renaming known results or smuggling ansatzes via self-citation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available, so concrete free parameters, axioms, and invented entities cannot be extracted. The approach implicitly relies on standard assumptions of pre-trained vision-language models and learned priors from large 3D datasets.

pith-pipeline@v0.9.0 · 5561 in / 1055 out tokens · 36467 ms · 2026-05-13T22:07:10.982144+00:00 · methodology

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

43 extracted references · 43 canonical work pages · 1 internal anchor

  1. [1]

    Kenny Jones, Qiuhong Anna Wei, Kailiang Fu, and Daniel Ritchie

    Rio Aguina-Kang, Maxim Gumin, Do Heon Han, Stewart Morris, Seung Jean Yoo, Aditya Ganeshan, R. Kenny Jones, Qiuhong Anna Wei, Kailiang Fu, and Daniel Ritchie. 2024. Open-Universe Indoor Scene Generation using LLM Program Synthesis and Uncurated Object Databases.arXiv preprint arXiv:2403.09675 (2024)

    work page arXiv 2024

  2. [2]

    Lanzendörfer, Nick Tuninga, and Roger Wattenhofer

    Frédéric Berdoz, Luca A. Lanzendörfer, Nick Tuninga, and Roger Wattenhofer

    work page

  3. [3]

    InProceedings of the AAAI Conference on Artificial Intelligence, Vol

    Text-to-Scene with Large Reasoning Models. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 40. 2435–2443

    work page

  4. [4]

    Ata Celen, Guo Han, Konrad Schindler, Luc Van Gool, Iro Armeni, Anton Obukhov, and Xi Wang. 2024. I-Design: Personalized LLM Interior Designer. In Computer Vision – ECCV 2024 Workshops

    work page 2024

  5. [5]

    Learning to Place Objects with Programs and Iterative Self Training

    Adrian Chang, Kai Wang, Yuanbo Li, Manolis Savva, Angel X. Chang, and Daniel Ritchie. 2025. Learning Object Placement Programs for Indoor Scene Synthesis with Iterative Self Training.arXiv preprint arXiv:2503.04496(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  6. [6]

    Chang, Manolis Savva, and Christopher D

    Angel X. Chang, Manolis Savva, and Christopher D. Manning. 2014. Learning Spatial Knowledge for Text to 3D Scene Generation. InProceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. 2028–2038

    work page 2014

  7. [7]

    Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, and Matthias Nießner

    Angela Dai, Angel X. Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, and Matthias Nießner. 2017. ScanNet: Richly-Annotated 3D Reconstructions of Indoor Scenes. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 5828–5839

    work page 2017

  8. [8]

    Wei Deng, Mengshi Qi, and Huadong Ma. 2025. Global-Local Tree Search in VLMs for 3D Indoor Scene Generation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 8975–8984

    work page 2025

  9. [9]

    Wenqi Dong, Bangbang Yang, Zesong Yang, Yuan Li, Tao Hu, Hujun Bao, Yuewen Ma, and Zhaopeng Cui. 2025. HiScene: Creating Hierarchical 3D Scenes with Iso- metric View Generation. InProceedings of the 33rd ACM International Conference on Multimedia. 9783–9792

    work page 2025

  10. [10]

    Chuan Fang, Heng Li, Yixun Liang, Jia Zheng, Yongsen Mao, Yuan Liu, Rui Tang, Zihan Zhou, and Ping Tan. 2026. SPATIALGEN: Layout-guided 3D Indoor Scene Generation. InProceedings of the International Conference on 3D Vision (3DV)

    work page 2026

  11. [11]

    Weixi Feng, Wanrong Zhu, Tsu-Jui Fu, Varun Jampani, Arjun Akula, Xuehai He, Sugato Basu, Xin Eric Wang, and William Yang Wang. 2023. LayoutGPT: Compositional Visual Planning and Generation with Large Language Models. In Advances in Neural Information Processing Systems

    work page 2023

  12. [12]

    Rao Fu, Zehao Wen, Zichen Liu, and Srinath Sridhar. 2024. AnyHome: Open- Vocabulary Generation of Structured and Textured 3D Homes. InComputer Vision – ECCV 2024. 52–70

    work page 2024

  13. [13]

    Jialin Gao, Donghao Zhou, Mingjian Liang, Lihao Liu, Chi-Wing Fu, Xiaowei Hu, and Pheng-Ann Heng. 2025. DisCo-Layout: Disentangling and Coordinating Semantic and Physical Refinement in a Multi-Agent Framework for 3D Indoor Layout Synthesis.arXiv preprint arXiv:2510.02178(2025)

    work page arXiv 2025

  14. [14]

    Ross, Cordelia Schmid, and Alireza Fathi

    Ziniu Hu, Ahmet Iscen, Aashi Jain, Thomas Kipf, Yisong Yue, David A. Ross, Cordelia Schmid, and Alireza Fathi. 2024. SceneCraft: An LLM Agent for Syn- thesizing 3D Scenes as Blender Code. InProceedings of the 41st International Conference on Machine Learning. 19252–19282

    work page 2024

  15. [15]

    Ian Huang, Yanan Bao, Karen Truong, Howard Zhou, Cordelia Schmid, Leonidas Guibas, and Alireza Fathi. 2025. FirePlace: Geometric Refinements of LLM Com- mon Sense Reasoning for 3D Object Placement. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 13466–13476

    work page 2025

  16. [16]

    Shi-Sheng Huang, Hongbo Fu, and Shi-Min Hu. 2016. Structure Guided Interior Scene Synthesis via Graph Matching.Graphical Models85 (2016), 46–55

    work page 2016

  17. [17]

    Chang, and Manolis Savva

    Mukul Khanna, Yongsen Mao, Hanxiao Jiang, Sanjay Haresh, Brennan Shacklett, Dhruv Batra, Alexander Clegg, Eric Undersander, Angel X. Chang, and Manolis Savva. 2024. Habitat Synthetic Scenes Dataset (HSSD-200): An Analysis of 3D Scene Scale and Realism Tradeoffs for ObjectGoal Navigation. InProceedings of the IEEE/CVF Conference on Computer Vision and Patt...

    work page 2024

  18. [18]

    Manyi Li, Akshay Gadi Patil, Kai Xu, Siddhartha Chaudhuri, Owais Khan, Ariel Shamir, Changhe Tu, Baoquan Chen, Daniel Cohen-Or, and Hao (Richard) Zhang

    work page

  19. [19]

    GRAINS: Generative Recursive Autoencoders for INdoor Scenes.ACM Transactions on Graphics38, 2 (2019), 12:1–12:16

    work page 2019

  20. [20]

    Chenguo Lin and Yadong Mu. 2024. InstructScene: Instruction-Driven 3D In- door Scene Synthesis with Semantic Graph Prior. InThe Twelfth International Conference on Learning Representations

    work page 2024

  21. [21]

    Yangkai Lin, Jiabao Lei, and Kui Jia. 2025. SceneLCM: End-to-End Layout-Guided Interactive Indoor Scene Generation with Latent Consistency Model.arXiv preprint arXiv:2506.07091(2025)

    work page arXiv 2025

  22. [22]

    Lu Ling, Chen-Hsuan Lin, Tsung-Yi Lin, Yifan Ding, Yu Zeng, Yichen Sheng, Yunhao Ge, Ming-Yu Liu, Aniket Bera, and Max Li. 2026. Scenethesis: A Language and Vision Agentic Framework for 3D Scene Generation. InThe Fourteenth International Conference on Learning Representations (ICLR 2026)

    work page 2026

  23. [23]

    Libin Liu, Shen Chen, Sen Jia, Jingzhe Shi, Can Jin, Zongkai Wu, Jenq-Neng Hwang, and Lei Li. 2025. Graph Canvas for Controllable 3D Scene Generation. In Proceedings of the 33rd ACM International Conference on Multimedia. 2536–2545

    work page 2025

  24. [24]

    Tenenbaum

    Andrew Luo, Zhoutong Zhang, Jiajun Wu, and Joshua B. Tenenbaum. 2020. End- to-End Optimization of Scene Layout. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 3754–3763

    work page 2020

  25. [25]

    Wang, Xiaoliang Huo, Angel X

    Hou In Derek Pun, Hou In Ivan Tam, Austin T. Wang, Xiaoliang Huo, Angel X. Chang, and Manolis Savva. 2026. HSM: Hierarchical Scene Motifs for Multi-Scale Indoor Scene Generation. InProceedings of the International Conference on 3D Vision (3DV)

    work page 2026

  26. [26]

    Siyuan Qi, Yixin Zhu, Siyuan Huang, Chenfanfu Jiang, and Song-Chun Zhu. 2018. Human-Centric Indoor Scene Synthesis Using Stochastic Grammar. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 5899– 5908

    work page 2018

  27. [27]

    Xingjian Ran, Yixuan Li, Linning Xu, Mulin Yu, and Bo Dai. 2025. Direct Numer- ical Layout Generation for 3D Indoor Scene Synthesis via Spatial Reasoning. In Advances in Neural Information Processing Systems

    work page 2025

  28. [28]

    Robertson, Steve Walker, Susan Jones, Micheline Hancock-Beaulieu, and Mike Gatford

    Stephen E. Robertson, Steve Walker, Susan Jones, Micheline Hancock-Beaulieu, and Mike Gatford. 1995. Okapi at TREC-3. InProceedings of the Third Text REtrieval Conference (TREC-3). 109–126

    work page 1995

  29. [29]

    Fan-Yun Sun, Weiyu Liu, Siyi Gu, Dylan Lim, Goutam Bhat, Federico Tombari, Manling Li, Nick Haber, and Jiajun Wu. 2025. LayoutVLM: Differentiable Opti- mization of 3D Layout via Vision-Language Models. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 29469–29478

    work page 2025

  30. [30]

    Weilin Sun, Xinran Li, Manyi Li, Kai Xu, Xiangxu Meng, and Lei Meng. 2025. Hierarchically-Structured Open-Vocabulary Indoor Scene Synthesis with Pre- trained Large Language Model. InProceedings of the AAAI Conference on Artificial Intelligence

    work page 2025

  31. [31]

    Wenzhuo Sun, Mingjian Liang, Wenxuan Song, Xuelian Cheng, and Zongyuan Ge. 2025. RoomPlanner: Explicit Layout Planner for Easier LLM-Driven 3D Room Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Trovato et al. Generation.arXiv preprint arXiv:2511.17048(2025)

    work page arXiv 2025

  32. [32]

    Jiapeng Tang, Yinyu Nie, Lev Markhasin, Angela Dai, Justus Thies, and Matthias Nießner. 2024. DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 20507–20518

    work page 2024

  33. [33]

    Chang, and Daniel Ritchie

    Kai Wang, Yu-An Lin, Ben Weissmann, Manolis Savva, Angel X. Chang, and Daniel Ritchie. 2019. PlanIT: Planning and Instantiating Indoor Scenes with Relation Graph and Spatial Prior Networks.ACM Transactions on Graphics38, 4 (2019), 132:1–132:15

    work page 2019

  34. [34]

    Yian Wang, Xiaowen Qiu, Jiageng Liu, Zhehuan Chen, Jiting Cai, Yufei Wang, Tsun-Hsuan Wang, Zhou Xian, and Chuang Gan. 2024. ARCHITECT: Generating Vivid and Interactive 3D Scenes with Hierarchical 2D Inpainting. InAdvances in Neural Information Processing Systems 37 (NeurIPS 2024)

    work page 2024

  35. [35]

    Yandan Yang, Baoxiong Jia, Shujie Zhang, and Siyuan Huang. 2025. SceneWeaver: All-in-One 3D Scene Synthesis with an Extensible and Self-Reflective Agent. In Advances in Neural Information Processing Systems

    work page 2025

  36. [36]

    Yandan Yang, Baoxiong Jia, Peiyuan Zhi, and Siyuan Huang. 2024. PhyScene: Physically Interactable 3D Scene Synthesis for Embodied AI. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 16262–16272

    work page 2024

  37. [37]

    Yue Yang, Fan-Yun Sun, Luca Weihs, Eli VanderBilt, Alvaro Herrasti, Winson Han, Jiajun Wu, Nick Haber, Ranjay Krishna, Lingjie Liu, Chris Callison-Burch, Mark Yatskar, Aniruddha Kembhavi, and Christopher Clark. 2024. Holodeck: Language Guided Generation of 3D Embodied AI Environments. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern...

    work page 2024

  38. [38]

    Zhaoda Ye, Yang Liu, and Yuxin Peng. 2024. MAAN: Memory-Augmented Auto- regressive Network for Text-driven 3D Indoor Scene Generation.IEEE Transac- tions on Multimedia(2024), 1–14

    work page 2024

  39. [39]

    Guangyao Zhai, Evin Pinar Ornek, Shun-Cheng Wu, Yan Di, Federico Tombari, Nassir Navab, and Benjamin Busam. 2023. CommonScenes: Generating Com- monsense 3D Indoor Scenes with Scene Graph Diffusion. InAdvances in Neural Information Processing Systems

    work page 2023

  40. [40]

    Song-Hai Zhang, Shao-Kui Zhang, Wei-Yu Xie, Cheng-Yang Luo, Yong-Liang Yang, and Hongbo Fu. 2022. Fast 3D Indoor Scene Synthesis by Learning Spatial Relation Priors of Objects.IEEE Transactions on Visualization and Computer Graphics28, 9 (2022), 3082–3092

    work page 2022

  41. [41]

    Yunzhi Zhang, Zizhang Li, Matt Zhou, Shangzhe Wu, and Jiajun Wu. 2025. The Scene Language: Representing Scenes with Programs, Words, and Embeddings. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 24625–24634

    work page 2025

  42. [42]

    Zaiwei Zhang, Zhenpei Yang, Chongyang Ma, Linjie Luo, Alexander Huth, Eti- enne Vouga, and Qixing Huang. 2020. Deep Generative Modeling for Scene Synthesis via Hybrid Representations.ACM Transactions on Graphics39, 3 (2020), 17:1–17:21

    work page 2020

  43. [43]

    Xiaoming Zhu, Xu Huang, Qinghongbing Xie, Zhi Deng, Junsheng Yu, Yirui Guan, Zhongyuan Liu, Lin Zhu, Qijun Zhao, Ligang Liu, and Long Zeng. 2025. Imaginarium: Vision-guided High-Quality 3D Scene Layout Generation.ACM Transactions on Graphics44, 6 (2025), 1–24

    work page 2025