Multiview Neural Surface Reconstruction by Disentangling Geometry and Appearance

By Lior Yariv, Yoni Kasten, Dror Moran, Meirav Galun, Matan Atzmon, Ronen Basri, Yaron Lipman. Neural Information Processing Systems, 2020

997 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/785eb47a72b579d7da4a99414288349afc025116

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Make-It-3D: High-Fidelity 3D Creation from A Single Image with Diffusion Prior (2023, methodology): “…with the input image, we penalize the pixel-wise difference between the rendering and the input image at the reference view β ref : ( Here we apply the foreground matting mask m to segment out the foreground as we empirically find that this eases the geometry reconstruction, which conforms to [58].”
DreamFusion: Text-to-3D using 2D Diffusion (2022, unknown): “This surface normal vector can be computed by normalizing the negative gradient of density τ with respect to the 3D coordinate µ: n = −∇µτ/ ∥∇µτ∥ (Yariv et al., 2020; Srinivasan et al., 2021).”
NeRF: Neural Radiance Field in 3D Vision, A Comprehensive Review (2022, unknown): “UNISURF outperformed benchmark methods including using density threshold in baseline NeRF models, as well as IDR [148].”
Pixel-aligned Volumetric Avatars (2021, background): “Implicit Methods Works such as [31] use explicit 3D information during training.”
Extracting Triangular 3D Models, Materials, and Lighting From Images (2021, methodology): “Chamfer distances are measured in the same way as NeuS [19], IDR [21], and DVR [14].”
Block-NeRF: Scalable Large Scene Neural View Synthesis (2022, methodology): “Our model also incorporates a learned camera pose reﬁnement which has been explored in previous works [34,59,66,69,70].”
Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors (2023, background): “NeRF does not store any 3D geometry explicitly (only the density field), and several works propose to use a signed distance function to recover a scene’s surface [98, 89, 97, 90, 13], including in the few-shot setting as well (e.”
3D Neural Field Generation Using Triplane Diffusion (2022, unknown): “…[1,4,5,10,11,15,21,23,45, 47,48,57,65,73] or use posed images to jointly optimize geometry and appearance [6,7,18,25,29,33,36–39, 44, 49, 50, 54, 55 Triplanesareefficient, scalingwiththesurfacearearatherthan volume, and naturally integrate with expressive, fine-tuned 2D generator architectures.”
Geo-Neus: Geometry-Consistent Neural Implicit Surfaces Learning for Multi-view Reconstruction (2022, methodology): “Following previous practices [33, 39, 40], we reconstruct surfaces from 15 scans of DTU dataset [1] to evaluate our method.”
Shape As Points: A Differentiable Poisson Solver (2021, methodology): “Shape representations are central to many of the recent advancements in 3D computer vision and computer graphics, ranging from neural rendering [43,47,50,58,61] to shape reconstruction [10,28,42, 49,52,54,73].”
Unsupervised Learning of 3D Object Categories from Videos in the Wild (2021, methodology): “NSVF [31] combined NeRF and voxel grids to improve the scalability and expressivity of the model whereas Yariv et al [65] uses sphere tracing to render signed distance fields.”
OmniObject3D: Large-Vocabulary 3D Object Dataset for Realistic Perception, Reconstruction and Generation (2023, background): “Some approaches [35, 38, 53, 88] regard the color of an intersection point between the ray and the surface as the rendered color, namely surface rendering, and they typically rely on accurate object masks.”
Volume Rendering of Neural Implicit Surfaces (2021, methodology): “A closely related line of research, often referred to as neural implicit surfaces [22, 38, 14], have been focusing on representing the scene’s geometry implicitly using a neural network, making the surface rendering process differentiable.”
NeuralRecon: Real-Time Coherent 3D Reconstruction from Monocular Video (2021, background): “Recently, neural implicit representations [29, 33, 36, 17, 54, 25] have gained significant advances.”
2D Gaussian Splatting for Geometrically Accurate Radiance Fields (2024, unknown): “NeRF employs a multi-layer perceptron (MLP) to represent geometry and view-dependent appearance, optimized via volume rendering to deliver exceptional rendering quality.”
AutoInt: Automatic Integration for Fast Neural Volume Rendering (2020, background): “Such representations have been explored for modeling shape parts [16, 17], objects [3, 6, 10, 19, 25, 27, 30, 34, 35, 40, 42, 43, 49, 53, 60], or scenes [14, 21, 29, 37, 45, 51].”