mlsanityData to 7 Oct 2026

Long-LRM: Long-sequence Large Reconstruction Model for Wide-coverage Gaussian Splats

arXiv 2410.12781 ↗First posted 16 Oct 2024cs.CVcited by 111

Not on the map. Why?

Abstract (arXiv)

We propose Long-LRM, a feed-forward 3D Gaussian reconstruction model for instant, high-resolution, 360deg wide-coverage, scene-level reconstruction. Specifically, it takes in 32 input images at a resolution of 960x540 and produces the Gaussian reconstruction in just 1 second on a single A100 GPU. To handle the long sequence of 250K tokens brought by the large input size, Long-LRM features a mixture of the recent Mamba2 blocks and the classical transformer blocks, enhanced by a light-weight token merging module and Gaussian pruning steps that balance between quality and efficiency. We evaluate Long-LRM on the large-scale DL3DV benchmark and Tanks&Temples, demonstrating reconstruction quality comparable to the optimization-based methods while achieving an 800x speedup w.r.t. the optimization-based approaches and an input size at least 60x larger than the previous feed-forward approaches. We conduct extensive ablation studies on our model design choices for both rendering quality and computation efficiency. We also explore Long-LRM's compatibility with other Gaussian variants such as 2D GS, which enhances Long-LRM's ability in geometry reconstruction. Project page: https://arthurhero.github.io/projects/llrm

Key facts. Cited by 111 papers in the map (data through 7 Oct 2026). Builds on 59 of its 66 references that are in the map. First posted on arXiv 16 Oct 2024.

Builds on

59 of 66 references.

References · 59

  1. 2017Attention Is All You Needcited by 59,932
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,141
  4. 2023Mamba: Linear-Time Sequence Modeling with Selective State Spacescited by 4,616
  5. 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,698
  6. 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,088
  7. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,551
  8. 2021Efficiently Modeling Long Sequences with Structured State Spacescited by 2,424
  9. 2021Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fieldscited by 2,013
  10. 2020Transformers are RNNs: Fast Autoregressive Transformers with Linear Attentioncited by 1,979
  11. 2020pixelNeRF: Neural Radiance Fields from One or Few Imagescited by 1,574
  12. 2024Depth Anything: Unleashing the Power of Large-Scale Unlabeled Datacited by 1,346
38 more
  1. 2024Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Dualitycited by 1,283
  2. 2022TensoRF: Tensorial Radiance Fieldscited by 1,275
  3. 20242D Gaussian Splatting for Geometrically Accurate Radiance Fieldscited by 1,061
  4. 2024Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Modelcited by 1,051
  5. 2023LRM: Large Reconstruction Model for Single Image to 3Dcited by 834
  6. 2018MVSNet: Depth Inference for Unstructured Multi-view Stereocited by 780
  7. 2021MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereocited by 735
  8. 2024LGM: Large Multi-View Gaussian Model for High-Resolution 3D Content Creationcited by 730
  9. 2021IBRNet: Learning Multi-View Image-Based Renderingcited by 698
  10. 2023Scaffold-GS: Structured 3D Gaussians for View-Adaptive Renderingcited by 684
  11. 2023pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstructioncited by 668
  12. 2023Zip-NeRF: Anti-Aliased Grid-Based Neural Radiance Fieldscited by 667
  13. 2023Mip-Splatting: Alias-free 3D Gaussian Splattingcited by 661
  14. 2023Nerfstudio: A Modular Framework for Neural Radiance Field Developmentcited by 596
  15. 2024MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Imagescited by 553
  16. 2023DL3DV-10K: A Large-Scale Scene Dataset for Deep Learning-based 3D Visioncited by 514
  17. 2022Point-NeRF: Point-based Neural Radiance Fieldscited by 503
  18. 2024Jamba: A Hybrid Transformer-Mamba Language Modelcited by 390
  19. 2023Instant3D: Fast Text-to-3D with Sparse-View Generation and Large Reconstruction Modelcited by 376
  20. 2023Splatter Image: Ultra-Fast Single-View 3D Reconstructioncited by 359
  21. 2024GS-LRM: Large Reconstruction Model for 3D Gaussian Splattingcited by 344
  22. 2021NeuralRecon: Real-Time Coherent 3D Reconstruction from Monocular Videocited by 241
  23. 2024LocalMamba: Visual State Space Model with Windowed Selective Scancited by 235
  24. 2024Octree-GS: Towards Consistent Real-time Rendering with LOD-Structured 3D Gaussianscited by 225
  25. 2023DMV3D: Denoising Multi-View Diffusion using 3D Large Reconstruction Modelcited by 213
  26. 2024VastGaussian: Vast 3D Gaussians for Large Scene Reconstructioncited by 209
  27. 2020Atlas: End-to-End 3D Scene Reconstruction from Posed Imagescited by 206
  28. 2021GeoNeRF: Generalizing NeRF with Geometry Priorscited by 195
  29. 2017Learning a Multi-View Stereo Machinecited by 194
  30. 2021Neural Rays for Occlusion-aware Image-based Renderingcited by 192
  31. 2024CityGaussian: Real-time High-quality Large-Scale Scene Rendering with Gaussianscited by 190
  32. 2024An Empirical Study of Mamba-based Language Modelscited by 177
  33. 2023PF-LRM: Pose-Free Large Reconstruction Model for Joint Pose and Shape Predictioncited by 165
  34. 2015Stereo Matching by Training a Convolutional Neural Network to Compare Image Patchescited by 151
  35. 2019Deep Stereo using Adaptive Thin Volume Representation with Uncertainty Awarenesscited by 151
  36. 2021Efficient Neural Radiance Fields for Interactive Free-viewpoint Videocited by 150
  37. 2024MeshLRM: Large Reconstruction Model for High-Quality Meshescited by 138
  38. 2022Generalizable Patch-Based Neural Renderingcited by 121

50 most cited of 59 All 59 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 10; 2025: 46; 2026: 55.

111 papers.

Sort:

By area:

  1. 2025AnySplat: Feed-forward 3D Gaussian Splatting from Unconstrained Viewscited by 234
  2. 2025Stable Virtual Camera: Generative View Synthesis with Diffusion Modelscited by 179
  3. 2025Test-Time Training Done Rightcited by 121
  4. 2024MVSplat360: Feed-Forward 360 Scene Synthesis from Sparse Viewscited by 111
  5. 2024Wonderland: Navigating 3D Scenes from a Single Imagecited by 98
  6. 2025RayZer: A Self-supervised Large View Synthesis Modelcited by 74
  7. 2025SpatialVID: A Large-Scale Video Dataset with Spatial Annotationscited by 72
  8. 2024You See it, You Got it: Learning 3D Creation on Pose-Free Videos at Scalecited by 65
  9. 2025DeepMesh: Auto-Regressive Artist-mesh Creation with Reinforcement Learningcited by 62
  10. 2025Bolt3D: Generating 3D Scenes in Secondscited by 57
  11. 20254DGT: Learning a 4D Gaussian Transformer Using Real-World Monocular Videoscited by 46
  12. 2025GenFusion: Closing the Loop between Reconstruction and Generation via Videoscited by 41
38 more
  1. 2025VolSplat: Rethinking Feed-Forward 3D Gaussian Splatting with Voxel-Aligned Predictioncited by 41
  2. 2025Lyra: Generative 3D Scene Reconstruction via Video Diffusion Model Self-Distillationcited by 37
  3. 2025ShapeLLM-Omni: A Native Multimodal LLM for 3D Generation and Understandingcited by 34
  4. 2025YoNoSplat: You Only Need One Model for Feedforward 3D Gaussian Splattingcited by 34
  5. 2025Matrix-3D: Omnidirectional Explorable 3D World Generationcited by 33
  6. 2025MoVieS: Motion-Aware 4D Dynamic View Synthesis in One Secondcited by 31
  7. 2024Feat2GS: Probing Visual Foundation Models with Gaussian Splattingcited by 26
  8. 2025Advances in Feed-Forward 3D Reconstruction and View Synthesis: A Surveycited by 23
  9. 2024MegaSynth: Scaling Up 3D Scene Reconstruction with Synthesized Datacited by 21
  10. 2025A Survey on 3D Gaussian Splatting Applications: Segmentation, Editing, and Generationcited by 19
  11. 2024Generative Densification: Learning to Densify Gaussians for High-Fidelity Generalizable 3D Reconstructioncited by 18
  12. 2025iLRM: An Iterative Large 3D Reconstruction Modelcited by 18
  13. 2025C3G: Learning Compact 3D Representations with 2K Gaussianscited by 18
  14. 2025SceneSplat++: A Large Dataset and Comprehensive Benchmark for Language Gaussian Splattingcited by 14
  15. 2025DriveGen3D: Boosting Feed-Forward Driving Scene Generation with Efficient Video Diffusioncited by 14
  16. 20254D-LRM: Large Space-Time Reconstruction Model From and To Any View at Any Timecited by 13
  17. 2025RegGS: Unposed Sparse Views Gaussian Splatting with 3DGS Registrationcited by 12
  18. 2025LongSplat: Online Generalizable 3D Gaussian Splatting from Long Sequence Imagescited by 12
  19. 2025Quantifying and Alleviating Co-Adaptation in Sparse-View 3D Gaussian Splattingcited by 12
  20. 2025ReSplat: Learning Recurrent Gaussian Splattingcited by 12
  21. 2025QuickSplat: Fast 3D Surface Reconstruction via Learned Gaussian Initializationcited by 11
  22. 2026TokenGS: Decoupling 3D Gaussian Prediction from Pixels with Learnable Tokenscited by 11
  23. 2025Revisiting Depth Representations for Feed-Forward 3D Gaussian Splattingcited by 10
  24. 2026tttLRM: Test-Time Training for Long Context and Autoregressive 3D Reconstructioncited by 10
  25. 2025HumanRAM: Feed-forward Human Reconstruction and Animation Model using Transformerscited by 9
  26. 2024Turbo-GS: Accelerating 3D Gaussian Fitting for High-Quality Radiance Fieldscited by 8
  27. 2025PhysGM: Large Physical Gaussian Model for Feed-Forward 4D Synthesiscited by 8
  28. 2025MeshRipple: Structured Autoregressive Generation of Artist-Meshescited by 8
  29. 2024SizeGS: Size-aware Compression of 3D Gaussian Splatting via Mixed Integer Programmingcited by 7
  30. 2025Seeing World Dynamics in a Nutshellcited by 7
  31. 2025Flux4D: Flow-based Unsupervised 4D Reconstructioncited by 7
  32. 2025Multi-view Pyramid Transformer: Look Coarser to See Broadercited by 7
  33. 2025EcoSplat: Efficiency-controllable Feed-forward 3D Gaussian Splatting from Multi-view Imagescited by 7
  34. 2025LVT: Large-Scale Scene Reconstruction via Local View Transformerscited by 6
  35. 2026SceneTok: A Compressed, Diffusable Token Space for 3D Scenescited by 6
  36. 2026GaussFusion: Improving 3D Reconstruction in the Wild with A Geometry-Informed Video Generatorcited by 6
  37. 2026SparseSplat: Towards Applicable Feed-Forward 3D Gaussian Splatting with Pixel-Unaligned Predictioncited by 6
  38. 2026Stream3D: Sequential Multi-View 3D Generation via Evidential Memorycited by 6

50 most cited of 111 All 111 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024LM-Gaussian: Boost Sparse-view 3D Gaussian Splatting with Large Model Priors
  2. 2025LL-Gaussian: Low-Light Scene Reconstruction and Enhancement via Gaussian Splatting for Novel View Synthesis
  3. 2024GRM: Large Gaussian Reconstruction Model for Efficient 3D Reconstruction and Generation
  4. 2026GaussVid: Sparse-View Gaussian Splatting with 3D-Aware Video Diffusion Priors
  5. 2025Diffusion-Guided Gaussian Splatting for Large-Scale Unconstrained 3D Reconstruction and Novel View Synthesis
  6. 2025FixingGS: Enhancing 3D Gaussian Splatting via Training-Free Score Distillation
  7. 2024GSD: View-Guided Gaussian Splatting Diffusion for 3D Reconstruction
  8. 2025ProSplat: Improved Feed-Forward 3D Gaussian Splatting for Wide-Baseline Sparse Views