Extreme View Synthesis
Not on the map. Why?
Abstract (arXiv)
We present Extreme View Synthesis, a solution for novel view extrapolation that works even when the number of input images is small--as few as two. In this context, occlusions and depth uncertainty are two of the most pressing issues, and worsen as the degree of extrapolation increases. We follow the traditional paradigm of performing depth-based warping and refinement, with a few key improvements. First, we estimate a depth probability volume, rather than just a single depth value for each pixel of the novel view. This allows us to leverage depth uncertainty in challenging regions, such as depth discontinuities. After using it to get an initial estimate of the novel view, we explicitly combine learned image priors and the depth uncertainty to synthesize a refined image with less artifacts. Our method is the first to show visually pleasing results for baseline magnifications of up to 30X.
Key facts. Cited by 94 papers in the map (data through 7 Oct 2026). First posted on arXiv 12 Dec 2018.
Builds on
No reference list yet.
Built on by
Citing papers per year: 2021: 36; 2022: 19; 2023: 23; 2024: 10; 2025: 5; 2026: 1.
94 papers.
Sort:
By area:
- 2021MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereocited by 735
- 2022DynIBaR: Neural Dynamic Image-Based Renderingcited by 224
- 2021NeX: Real-time View Synthesis with Neural Basis Expansioncited by 215
- 2023Robust Dynamic Radiance Fieldscited by 193
- 2021Neural Rays for Occlusion-aware Image-based Renderingcited by 192
- 2021Stereo Radiance Fields (SRF): Learning View Synthesis for Sparse Views of Novel Scenescited by 190
- 2021Point-Based Neural Rendering with Per-View Optimizationcited by 188
- 2024LVSM: A Large View Synthesis Model with Minimal 3D Inductive Biascited by 178
- 2021NeRF--: Neural Radiance Fields Without Known Camera Parameterscited by 155
- 2021Editable Free-viewpoint Video Using a Layered Neural Representationcited by 148
- 2021Unconstrained Scene Generation with Locally Conditioned Radiance Fieldscited by 130
- 2021Hallucinated Neural Radiance Fields in the Wildcited by 130
38 more
- 2021HDR-NeRF: High Dynamic Range Neural Radiance Fieldscited by 113
- 2021Light Field Neural Renderingcited by 108
- 2021Geometry-Free View Synthesis: Transformers and no 3D Priorscited by 99
- 2021NeRF-SR: High-Quality Neural Radiance Fields using Supersamplingcited by 95
- 2021Neural Trajectory Fields for Dynamic Novel View Synthesiscited by 85
- 2023MonoHuman: Animatable Human Neural Field from Monocular Videocited by 85
- 2022InfiniteNature-Zero: Learning Perpetual View Generation of Natural Scenes from Single Imagescited by 82
- 2021FoV-NeRF: Foveated Neural Radiance Fields for Virtual Realitycited by 79
- 2024NVS-Solver: Video Diffusion Model as Zero-Shot Novel View Synthesizercited by 79
- 2023Neural Residual Radiance Fields for Streamably Free-Viewpoint Videoscited by 77
- 2024Fully Explicit Dynamic Gaussian Splattingcited by 75
- 2022ViewFormer: NeRF-free Neural Rendering from Few Images Using Transformerscited by 70
- 2023Forward Flow for Novel View Synthesis of Dynamic Scenescited by 67
- 2024Advances in 3D Generation: A Surveycited by 62
- 2021Watch It Move: Unsupervised Discovery of 3D Joints for Re-Posing of Articulated Objectscited by 50
- 2023Persistent Nature: A Generative Model of Unbounded 3D Worldscited by 44
- 2022Neural Deformable Voxel Grid for Fast Optimization of Dynamic View Synthesiscited by 43
- 2022FWD: Real-time Novel View Synthesis with Forward Warping and Depthcited by 39
- 2023DynPoint: Dynamic Neural Point For View Synthesiscited by 37
- 2023Real-Time Neural Rasterization for Large Scenescited by 37
- 2022RTMV: A Ray-Traced Multi-View Synthetic Dataset for Novel View Synthesiscited by 33
- 2021Fast Training of Neural Lumigraph Representations using Meta Learningcited by 29
- 2021Fast and Explicit Neural View Synthesiscited by 28
- 2023Pointersect: Neural Rendering with Cloud-Ray Intersectioncited by 23
- 2021Hybrid Neural Fusion for Full-frame Video Stabilizationcited by 21
- 2022Grounded Affordance from Exocentric Viewcited by 21
- 2023iNVS: Repurposing Diffusion Inpainters for Novel View Synthesiscited by 20
- 2021Reassessing the Limitations of CNN Methods for Camera Pose Regressioncited by 19
- 2021Layout-Guided Novel View Synthesis from a Single Indoor Panoramacited by 16
- 2021iButter: Neural Interactive Bullet Time Generator for Human Free-viewpoint Renderingcited by 16
- 2021Self-Supervised Visibility Learning for Novel View Synthesiscited by 14
- 2022Multi-sensor large-scale dataset for multi-view 3D reconstructioncited by 14
- 2023Tiled Multiplane Images for Practical 3D Photographycited by 14
- 2022Neural Mesh-Based Graphicscited by 12
- 2021VirtualCube: An Immersive 3D Video Communication Systemcited by 11
- 2024LiftImage3D: Lifting Any Single Image to 3D Gaussians with Video Generation Priorscited by 11
- 2021Deep View Synthesis via Self-Consistent Generative Networkcited by 10
- 2022Learning Generalizable Light Field Networks from Few Imagescited by 10
50 most cited of 94 All 94 (JSON)
- 2026Efficient-LVSM: Faster, Cheaper, and Better Large View Synthesis Model via Decoupled Co-Refinement Attention6 Feb 2026
- 2025UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesis23 Dec 2025
- 2025CPSL: Representing Volumetric Video via Content-Promoted Scene Layers18 Nov 2025
- 2025Inverse Image-Based Rendering for Light Field Generation from Single Images23 Oct 2025
- 2025Perceive-Sample-Compress: Towards Real-Time 3D Gaussian Splatting7 Aug 2025
- 2025MultimodalStudio: A Heterogeneous Sensor Dataset and Framework for Neural Rendering across Multiple Imaging Modalities25 Mar 2025
- 2024LiftImage3D: Lifting Any Single Image to 3D Gaussians with Video Generation Priors12 Dec 2024
- 2024Human Multi-View Synthesis from a Single-View Model:Transferred Body and Face Representations4 Dec 2024
- 2024Planar Reflection-Aware Neural Radiance Fields7 Nov 2024
- 2024LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias22 Oct 2024
- 2024Fully Explicit Dynamic Gaussian Splatting21 Oct 2024
- 2024GMT: Enhancing Generalizable Neural Rendering via Geometry-Driven Multi-Reference Texture Transfer1 Oct 2024
38 more
- 2024UC-NeRF: Uncertainty-aware Conditional Neural Radiance Fields from Endoscopic Sparse Views4 Sep 2024
- 2024Lightweight Predictive 3D Gaussian Splats27 Jun 2024
- 2024NVS-Solver: Video Diffusion Model as Zero-Shot Novel View Synthesizer24 May 2024
- 2024Advances in 3D Generation: A Survey31 Jan 2024
- 2023Re-Nerfing: Improving Novel View Synthesis through Novel View Synthesis4 Dec 2023
- 2023Anisotropic Neural Representation Learning for High-Quality Neural Rendering30 Nov 2023
- 2023Rethinking Directional Integration in Neural Radiance Fields28 Nov 2023
- 2023Hyb-NeRF: A Multiresolution Hybrid Encoding for Neural Radiance Fields21 Nov 2023
- 2023Real-Time Neural Rasterization for Large Scenes9 Nov 2023
- 2023DynPoint: Dynamic Neural Point For View Synthesis29 Oct 2023
- 2023LightSpeed: Light and Fast Neural Light Fields on Mobile Devices25 Oct 2023
- 2023iNVS: Repurposing Diffusion Inpainters for Novel View Synthesis24 Oct 2023
- 2023Perceptual Assessment and Optimization of HDR Image Rendering19 Oct 2023
- 2023Shielding the Unseen: Privacy Protection through Poisoning NeRF with Spatial Deformation4 Oct 2023
- 2023Forward Flow for Novel View Synthesis of Dynamic Scenes29 Sep 2023
- 2023Tiled Multiplane Images for Practical 3D Photography25 Sep 2023
- 2023Depth self-supervision for single image novel view synthesis27 Aug 2023
- 2023LoLep: Single-View View Synthesis with Locally-Learned Planes and Self-Attention Occlusion Inference23 Jul 2023
- 2023Pointersect: Neural Rendering with Cloud-Ray Intersection24 Apr 2023
- 2023Single-View View Synthesis with Self-Rectified Pseudo-Stereo19 Apr 2023
- 2023CAMM: Building Category-Agnostic and Animatable 3D Models from Monocular Videos14 Apr 2023
- 2023Neural Residual Radiance Fields for Streamably Free-Viewpoint Videos10 Apr 2023
- 2023MonoHuman: Animatable Human Neural Field from Monocular Video4 Apr 2023
- 2023Persistent Nature: A Generative Model of Unbounded 3D Worlds23 Mar 2023
- 2023Harnessing Low-Frequency Neural Fields for Few-Shot View Synthesis15 Mar 2023
- 2023Robust Dynamic Radiance Fields5 Jan 2023
- 2023Detachable Novel Views Synthesis of Dynamic Scenes Using Distribution-Driven Neural Radiance Fields1 Jan 2023
- 2022DynIBaR: Neural Dynamic Image-Based Rendering20 Nov 2022
- 2022Learning A Locally Unified 3D Point Cloud for View Synthesis12 Sep 2022
- 2022Grounded Affordance from Exocentric View28 Aug 2022
- 2022Neural Mesh-Based Graphics10 Aug 2022
- 2022Cascaded and Generalizable Neural Radiance Fields for Fast View Synthesis9 Aug 2022
- 2022Globally Consistent Video Depth and Pose Estimation with Efficient Test-Time Training4 Aug 2022
- 2022Learning Generalizable Light Field Networks from Few Images24 Jul 2022
- 2022InfiniteNature-Zero: Learning Perpetual View Generation of Natural Scenes from Single Images22 Jul 2022
- 2022Progressively-connected Light Field Network for Efficient View Synthesis10 Jul 2022
- 2022FWD: Real-time Novel View Synthesis with Forward Warping and Depth16 Jun 2022
- 2022Neural Deformable Voxel Grid for Fast Optimization of Dynamic View Synthesis15 Jun 2022
50 newest of 94 All 94 (JSON)
- 2023Persistent Nature: A Generative Model of Unbounded 3D Worldscited by 44
- 2023Persistent Nature: A Generative Model of Unbounded 3D Worlds23 Mar 2023
- 2021MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereocited by 735
- 2022DynIBaR: Neural Dynamic Image-Based Renderingcited by 224
- 2021NeX: Real-time View Synthesis with Neural Basis Expansioncited by 215
- 2023Robust Dynamic Radiance Fieldscited by 193
- 2021Neural Rays for Occlusion-aware Image-based Renderingcited by 192
- 2021Stereo Radiance Fields (SRF): Learning View Synthesis for Sparse Views of Novel Scenescited by 190
- 2021Point-Based Neural Rendering with Per-View Optimizationcited by 188
- 2024LVSM: A Large View Synthesis Model with Minimal 3D Inductive Biascited by 178
- 2021NeRF--: Neural Radiance Fields Without Known Camera Parameterscited by 155
- 2021Editable Free-viewpoint Video Using a Layered Neural Representationcited by 148
- 2021Unconstrained Scene Generation with Locally Conditioned Radiance Fieldscited by 130
- 2021Hallucinated Neural Radiance Fields in the Wildcited by 130
12 most cited of 91 All 91 (JSON)
- 2026Efficient-LVSM: Faster, Cheaper, and Better Large View Synthesis Model via Decoupled Co-Refinement Attention6 Feb 2026
- 2025UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesis23 Dec 2025
- 2025CPSL: Representing Volumetric Video via Content-Promoted Scene Layers18 Nov 2025
- 2025Inverse Image-Based Rendering for Light Field Generation from Single Images23 Oct 2025
- 2025Perceive-Sample-Compress: Towards Real-Time 3D Gaussian Splatting7 Aug 2025
- 2025MultimodalStudio: A Heterogeneous Sensor Dataset and Framework for Neural Rendering across Multiple Imaging Modalities25 Mar 2025
- 2024LiftImage3D: Lifting Any Single Image to 3D Gaussians with Video Generation Priors12 Dec 2024
- 2024Planar Reflection-Aware Neural Radiance Fields7 Nov 2024
- 2024LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias22 Oct 2024
- 2024Fully Explicit Dynamic Gaussian Splatting21 Oct 2024
- 2024GMT: Enhancing Generalizable Neural Rendering via Geometry-Driven Multi-Reference Texture Transfer1 Oct 2024
- 2024UC-NeRF: Uncertainty-aware Conditional Neural Radiance Fields from Endoscopic Sparse Views4 Sep 2024
12 newest of 91 All 91 (JSON)