mlsanityData to 14 Oct 2026

MegaScenes: Scene-Level View Synthesis at Scale

arXiv 2406.11819 ↗First posted 17 Jun 2024cs.CVcited by 58

Not on the map. Why?

Abstract (arXiv)

Scene-level novel view synthesis (NVS) is fundamental to many vision and graphics applications. Recently, pose-conditioned diffusion models have led to significant progress by extracting 3D information from 2D foundation models, but these methods are limited by the lack of scene-level training data. Common dataset choices either consist of isolated objects (Objaverse), or of object-centric scenes with limited pose distributions (DTU, CO3D). In this paper, we create a large-scale scene-level dataset from Internet photo collections, called MegaScenes, which contains over 100K structure from motion (SfM) reconstructions from around the world. Internet photos represent a scalable data source but come with challenges such as lighting and transient objects. We address these issues to further create a subset suitable for the task of NVS. Additionally, we analyze failure cases of state-of-the-art NVS methods and significantly improve generation consistency. Through extensive experiments, we validate the effectiveness of both our dataset and method on generating in-the-wild scenes. For details on the dataset and code, see our project page at https://megascenes.github.io.

Key facts. Cited by 58 papers in the map (data through 14 Oct 2026). Builds on 62 of its 72 references that are in the map. First posted on arXiv 17 Jun 2024.

Builds on

62 of 72 references.

References · 62

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  3. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  4. 2020Denoising Diffusion Implicit Modelscited by 9,781
  5. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,355
  6. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,226
  7. 2022Classifier-Free Diffusion Guidancecited by 5,847
  8. 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,117
  9. 2022DreamFusion: Text-to-3D using 2D Diffusioncited by 3,025
  10. 2015ShapeNet: An Information-Rich 3D Model Repositorycited by 2,897
  11. 2021Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fieldscited by 2,032
  12. 2020pixelNeRF: Neural Radiance Fields from One or Few Imagescited by 1,584
38 more
  1. 2023DUSt3R: Geometric 3D Vision Made Easycited by 1,567
  2. 2023Zero-1-to-3: Zero-shot One Image to 3D Objectcited by 1,489
  3. 2017Matterport3D: Learning from RGB-D Data in Indoor Environmentscited by 1,376
  4. 2024Depth Anything: Unleashing the Power of Large-Scale Unlabeled Datacited by 1,360
  5. 2018Demystifying MMD GANscited by 1,293
  6. 2020NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collectionscited by 1,232
  7. 2023ProlificDreamer: High-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillationcited by 1,219
  8. 2023MVDream: Multi-view Diffusion for 3D Generationcited by 938
  9. 2023LRM: Large Reconstruction Model for Single Image to 3Dcited by 849
  10. 2023Objaverse-XL: A Universe of 10M+ 3D Objectscited by 774
  11. 2023ScanNet++: A High-Fidelity Dataset of 3D Indoor Scenescited by 764
  12. 2018MegaDepth: Learning Single-View Depth Prediction from Internet Photoscited by 737
  13. 2023ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depthcited by 657
  14. 2021Common Objects in 3D: Large-Scale Learning and Evaluation of Real-life 3D Category Reconstructioncited by 652
  15. 2023SyncDreamer: Generating Multiview-consistent Images from a Single-view Imagecited by 645
  16. 2021RegNeRF: Regularizing Neural Radiance Fields for View Synthesis from Sparse Inputscited by 615
  17. 2020Hypersim: A Photorealistic Synthetic Dataset for Holistic Indoor Scene Understandingcited by 604
  18. 2022Score Jacobian Chaining: Lifting Pretrained 2D Diffusion Models for 3D Generationcited by 597
  19. 2023LightGlue: Local Feature Matching at Light Speedcited by 590
  20. 2021Ref-NeRF: Structured View-Dependent Appearance for Neural Radiance Fieldscited by 575
  21. 2023One-2-3-45: Any Single Image to 3D Mesh in 45 Seconds without Per-Shape Optimizationcited by 560
  22. 2023DL3DV-10K: A Large-Scale Scene Dataset for Deep Learning-based 3D Visioncited by 537
  23. 2019BlendedMVS: A Large-scale Dataset for Generalized Multi-view Stereo Networkscited by 490
  24. 2023Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priorscited by 397
  25. 2019SynSin: End-to-end View Synthesis from a Single Imagecited by 364
  26. 2023ReconFusion: 3D Reconstruction with Diffusion Priorscited by 318
  27. 2020DISK: Learning local features with policy gradientcited by 312
  28. 2020Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrievalcited by 309
  29. 2022Novel View Synthesis with Diffusion Modelscited by 300
  30. 2023Generative Novel View Synthesis with 3D-Aware Diffusion Modelscited by 274
  31. 2020Neural Radiance Flow for 4D View Synthesis and Video Processingcited by 270
  32. 2020Infinite Nature: Perpetual View Generation of Natural Scenes from a Single Imagecited by 243
  33. 2023LucidDreamer: Domain-free Generation of 3D Gaussian Splatting Scenescited by 231
  34. 2023MVImgNet: A Large-scale Dataset of Multi-view Imagescited by 220
  35. 2022SparseFusion: Distilling View-conditioned Diffusion for 3D Reconstructioncited by 216
  36. 2023MatrixCity: A Large-scale City Dataset for City-scale Neural Rendering and Beyondcited by 185
  37. 2016View Synthesis by Appearance Flowcited by 172
  38. 2023PF-LRM: Pose-Free Large Reconstruction Model for Joint Pose and Shape Predictioncited by 165

50 most cited of 62 All 62 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 11; 2025: 23; 2026: 24.

58 papers.

Sort:

By area:

  1. 2024Navigation World Modelscited by 272
  2. 2024LVSM: A Large View Synthesis Model with Minimal 3D Inductive Biascited by 189
  3. 2025RayZer: A Self-supervised Large View Synthesis Modelcited by 77
  4. 2024You See it, You Got it: Learning 3D Creation on Pose-Free Videos at Scalecited by 65
  5. 2025MegaLoc: One Retrieval to Place Them Allcited by 60
  6. 2025AerialMegaDepth: Learning Aerial-Ground Reconstruction and View Synthesiscited by 39
  7. 2024MVGenMaster: Scaling Multi-View Generation from Any Image via 3D Priors Enhanced Diffusion Modelcited by 32
  8. 2024Feat2GS: Probing Visual Foundation Models with Gaussian Splattingcited by 27
  9. 2024MegaSynth: Scaling Up 3D Scene Reconstruction with Synthesized Datacited by 21
  10. 2024Doppelgangers++: Improved Visual Disambiguation with Geometric 3D Featurescited by 20
  11. 2025Sharp Monocular View Synthesis in Less Than a Secondcited by 20
  12. 2025SpatialCrafter: Unleashing the Imagination of Video Diffusion Models for Scene Reconstruction from Limited Observationscited by 16
38 more
  1. 2025FlashDepth: Real-time Streaming Video Depth Estimation at 2K Resolutioncited by 14
  2. 2025Reconstruct, Inpaint, Test-Time Finetune: Dynamic Novel-view Synthesis from Monocular Videoscited by 14
  3. 2026LoMa: Local Feature Matching Revisitedcited by 11
  4. 2024MOVIS: Enhancing Multi-Object Novel View Synthesis for Indoor Scenescited by 10
  5. 2025Seeing in the Dark: Benchmarking Egocentric 3D Vision with the Oxford Day-and-Night Datasetcited by 8
  6. 2024KFC-W: Generating 3D-Consistent Videos from Unposed Internet Photoscited by 7
  7. 2025HumanGif: Single-View Human Diffusion with Generative Priorcited by 7
  8. 2025Vid-CamEdit: Video Camera Trajectory Editing with Generative Rendering from Estimated Geometrycited by 7
  9. 2024Extreme Rotation Estimation in the Wildcited by 6
  10. 2025GarmentCrafter: Progressive Novel View Synthesis for Single-View 3D Garment Reconstruction and Editingcited by 6
  11. 2025Emergent Extreme-View Geometry in 3D Foundation Modelscited by 6
  12. 2025MuM: Multi-View Masked Image Modeling for 3D Visioncited by 5
  13. 2025C3Po: Cross-View Cross-Modality Correspondence by Pointmap Predictioncited by 4
  14. 2025UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesiscited by 4
  15. 2026Long-tail Internet photo reconstructioncited by 4
  16. 2025To Match or Not to Match: Revisiting Image Matching for Reliable Visual Place Recognitioncited by 3
  17. 2025WAVE: Warp-Based View Guidance for Consistent Novel View Synthesis Using a Single Imagecited by 3
  18. 2025Wanderland: Geometrically Grounded Simulation for Open-World Embodied AIcited by 3
  19. 2026AirZoo: A Unified Large-Scale Dataset for Grounding Aerial Geometric 3D Visioncited by 3
  20. 2026UniWorld-View: Large-Baseline View Synthesis via Video Diffusion Modelscited by 3
  21. 2025Accenture-NVS1: A Novel View Synthesis Datasetcited by 2
  22. 2025Geometric Consistency Refinement for Single Image Novel View Synthesis via Test-Time Adaptation of Diffusion Modelscited by 2
  23. 2025The Less You Depend, The More You Learn: Synthesizing Novel Views from Sparse, Unposed Images with Minimal 3D Knowledgecited by 2
  24. 2025WildCAT3D: Appearance-Aware Multi-View Diffusion in the Wildcited by 2
  25. 2026TerraSky3D: Multi-View Reconstructions of European Landmarks in 4Kcited by 2
  26. 2026G3T Up! Gravity Aligned Coordinate Frames Simplify Pointmap Processingcited by 2
  27. 2026Understanding and Optimizing Attention-Based Sparse Matching for Diverse Local Featurescited by 1
  28. 2026Scene Grounding In the Wildcited by 1
  29. 2026Who Handles Orientation? Investigating Invariance in Feature Matchingcited by 1
  30. 2026CityRAG: Stepping Into a City via Spatially-Grounded Video Generationcited by 1
  31. 2026Generalizable Sparse-View 3D Reconstruction from Unconstrained Imagescited by 1
  32. 20243D-free meets 3D priors: Novel View Synthesis from a Single Image with Pretrained Diffusion Guidancecited by 0
  33. 2025ArchGPT: Understanding the World's Architectures with Large Multimodal Modelscited by 0
  34. 2026DF3DV-1K: A Large-Scale Dataset and Benchmark for Distractor-Free Novel View Synthesiscited by 0
  35. 2026WildSplatter: Feed-forward 3D Gaussian Splatting with Appearance Control from Unconstrained Imagescited by 0
  36. 2026ArchSym: Detecting 3D-Grounded Architectural Symmetries in the Wildcited by 0
  37. 2026Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depthcited by 0
  38. 2026Honey, I Shrunk the Arc de Triomphe!cited by 0

50 most cited of 58 All 58 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2023ZeroNVS: Zero-Shot 360-Degree View Synthesis from a Single Image
  2. 2026OrbitNVS: Harnessing Video Diffusion Priors for Novel View Synthesis
  3. 2024Novel View Synthesis with Pixel-Space Diffusion Models
  4. 2023ViVid-1-to-3: Novel View Synthesis with Video Diffusion Models
  5. 2025Next-Scale Autoregressive Models are Zero-Shot Single-Image Object View Synthesizers
  6. 2026BetterScene: 3D Scene Synthesis with Representation-Aligned Generative Model
  7. 2025DT-NVS: Diffusion Transformers for Novel View Synthesis
  8. 2024FSViewFusion: Few-Shots View Generation of Novel Objects