54,138 Hits in 5.1 sec

Shape Reconstruction by Learning Differentiable Surface Representations [article]

Jan Bednarik, Shaifali Parashar, Erhan Gundogdu, Mathieu Salzmann, Pascal Fua
2019 arXiv   pre-print
Generative models that produce point clouds have emerged as a powerful tool to represent 3D surfaces, and the best current ones rely on learning an ensemble of parametric representations.  ...  As a consequence, computing shape properties such as surface normals and curvatures becomes difficult and unreliable.  ...  surface representation whose differentiable properties can be estimated easily.  ... 
arXiv:1911.11227v1 fatcat:2ubd2s3a2bhn7nay7b5f2z7pfa

Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision [article]

Michael Niemeyer, Lars Mescheder, Michael Oechsle, Andreas Geiger
2020 arXiv   pre-print
This allows us to learn implicit shape and texture representations directly from RGB images. We experimentally show that our single-view reconstructions rival those learned with full 3D supervision.  ...  In this work, we propose a differentiable rendering formulation for implicit shape and texture representations.  ...  Acknowledgments This work was supported by an NVIDIA research gift. The authors thank the International Max Planck Research School for Intelligent Systems (IMPRS-IS) for supporting Michael Niemeyer.  ... 
arXiv:1912.07372v2 fatcat:vy36p6k2jzh47f6ih6crqlt334

Learning to Infer Implicit Surfaces without 3D Supervision [article]

Shichen Liu, Shunsuke Saito, Weikai Chen, Hao Li
2019 arXiv   pre-print
We address the fundamental problem of learning implicit surfaces for shape inference without the need of 3D supervision.  ...  Mesh-based representations are more efficient but are limited by their ability to handle varying topologies.  ...  learning of implicit surface representations by differentiating the implicit field rendering.  ... 
arXiv:1911.00767v1 fatcat:vq33mtgfvvc7llzkigbvpljo4i

SDFDiff: Differentiable Rendering of Signed Distance Fields for 3D Shape Optimization [article]

Yue Jiang, Dantong Ji, Zhizhong Han, Matthias Zwicker
2022 arXiv   pre-print
We propose SDFDiff, a novel approach for image-based shape optimization using differentiable rendering of 3D shapes represented by signed distance functions (SDFs).  ...  Compared to other representations, SDFs have the advantage that they can represent shapes with arbitrary topology, and that they guarantee watertight surfaces.  ...  DeepSDF [39] was proposed as a learned continuous signed distance function representation of a class of shapes.  ... 
arXiv:1912.07109v2 fatcat:7gjbfiqpnfeb3bt2btskwu6p6a

OctField: Hierarchical Implicit Functions for 3D Modeling [article]

Jia-Heng Tang, Weikai Chen, Jie Yang, Bo Wang, Songrun Liu, Bo Yang, Lin Gao
2021 arXiv   pre-print
We demonstrate the value of OctField for a range of shape modeling and reconstruction tasks, showing superiority over alternative approaches.  ...  both octree structure and surface geometry in a differentiable manner.  ...  This work was supported by CCF-Tencent Open Fund, the National Natural Science Foundation of China (No. 61872440 and No. 62061136007), the Beijing Municipal Natural Science Foundation (No.  ... 
arXiv:2111.01067v1 fatcat:l2d4t423srdwnbkkgd3rxtm4uy

DeepCurrents: Learning Implicit Representations of Shapes with Boundaries [article]

David Palmer and Dmitriy Smirnov and Stephanie Wang and Albert Chern and Justin Solomon
2022 arXiv   pre-print
Many of these methods, however, learn only closed surfaces and are unable to reconstruct shapes with boundary curves.  ...  Recent techniques have been successful in reconstructing surfaces as level sets of learned functions (such as signed distance fields) parameterized by deep neural networks.  ...  [27] reconstruct shapes by learning multiple implicit representations arranged according to a learned template configuration.  ... 
arXiv:2111.09383v2 fatcat:l7jv62o5jrevddktahtmph5n5e

Neural Ray-Tracing: Learning Surfaces and Reflectance for Relighting and View Synthesis [article]

Julian Knodt, Joe Bartusek, Seung-Hwan Baek, Felix Heide
2021 arXiv   pre-print
By learning decomposed transport with surface representations established in conventional rendering methods, the method naturally facilitates editing shape, reflectance, lighting and scene composition.  ...  In this work, we explicitly model the light transport between scene surfaces and we rely on traditional integration schemes and the rendering equation to reconstruct a scene.  ...  Instead of using a volumetric representation like NeRF, we use a surface-based representation with learned BSDFs in order to reconstruct scenes.  ... 
arXiv:2104.13562v2 fatcat:bgy25vm77jaudfcx4l4ftvwvfq

Shape As Points: A Differentiable Poisson Solver [article]

Songyou Peng, Chiyu "Max" Jiang, Yiyi Liao, Michael Niemeyer, Marc Pollefeys, Andreas Geiger
2021 arXiv   pre-print
We demonstrate the effectiveness of SAP on the task of surface reconstruction from unoriented point clouds and learning-based reconstruction.  ...  In this paper, we revisit the classic yet ubiquitous point cloud representation and introduce a differentiable point-to-mesh layer using a differentiable formulation of Poisson Surface Reconstruction (  ...  The authors thank the Max Planck ETH Center for Learning Systems (CLS) for supporting Songyou Peng and the International Max Planck Research School for Intelligent Systems (IMPRS-IS) for supporting Michael  ... 
arXiv:2106.03452v2 fatcat:guf3qerjkzgtnh5evwadetcncm

Learning Deformable Tetrahedral Meshes for 3D Reconstruction [article]

Jun Gao, Wenzheng Chen, Tommy Xiang, Clement Fuji Tsang, Alec Jacobson, Morgan McGuire, Sanja Fidler
2020 arXiv   pre-print
3D shape representations that accommodate learning-based 3D reconstruction are an open problem in machine learning and computer graphics.  ...  Previous work on neural 3D reconstruction demonstrated benefits, but also limitations, of point cloud, voxel, surface mesh, and implicit function representations.  ...  This work was fully funded by NVIDIA and no third-party funding was used.  ... 
arXiv:2011.01437v2 fatcat:cvrenxyzpzdg5nieg6segqlbdq

Coupling Explicit and Implicit Surface Representations for Generative 3D Modeling [article]

Omid Poursaeed and Matthew Fisher and Noam Aigerman and Vladimir G. Kim
2020 arXiv   pre-print
Additionally, our surface reconstruction step can directly leverage the explicit atlas-based representation.  ...  This process is computationally efficient, and can be directly used by differentiable rasterizers, enabling training our hybrid representation with image-based losses.  ...  We demonstrate the advantage of our joint representation by using it to train 3D-shape autoencoders and reconstruct a surface from a single image.  ... 
arXiv:2007.10294v2 fatcat:pl7yk2ucpnfonifedy5v7nediq

Neural Star Domain as Primitive Representation [article]

Yuki Kawana, Yusuke Mukuta, Tatsuya Harada
2020 arXiv   pre-print
Accurate structured reconstruction by parsimonious and semantic primitive representation further broadens its application.  ...  When reconstructing a target shape with multiple primitives, it is preferable that one can instantly access the union of basic properties of the shape such as collective volume and surface, treating the  ...  This work was partially supported by JST AIP Acceleration Research Grant Number JPMJCR20U3, and partially supported by JSPS KAKENHI Grant Number JP19H01115.  ... 
arXiv:2010.11248v2 fatcat:aaiisdqa25gfpdi32ynhaupxw4

MeshSDF: Differentiable Iso-Surface Extraction [article]

Edoardo Remelli, Artem Lukoianov, Stephan R. Richter, Benoît Guillard, Timur Bagautdinov, Pierre Baque, Pascal Fua
2020 arXiv   pre-print
We use two different applications to validate our theoretical insight: Single-View Reconstruction via Differentiable Rendering and Physically-Driven Shape Optimization.  ...  Our key insight is that by reasoning on how implicit field perturbations impact local surface geometry, one can ultimately differentiate the 3D location of surface samples with respect to the underlying  ...  The non-differentiability of Marching Cubes has been addressed by learning differentiable approximations of it [26, 56] .  ... 
arXiv:2006.03997v2 fatcat:msmzxi66xrbjtjym3lylz7zlmm

Towards Learning Neural Representations from Shadows [article]

Kushagra Tiwary, Tzofi Klinghoffer, Ramesh Raskar
2022 arXiv   pre-print
We observe that shadows are a powerful cue that can constrain neural scene representations to learn SfS, and even outperform NeRF to reconstruct otherwise hidden geometry.  ...  We present a method that learns neural scene representations from only shadows present in the scene.  ...  To reconstruct the meshes, we run marching cubes on the learned implicit representations.  ... 
arXiv:2203.15946v1 fatcat:yxzpcld4zvffrpm2ry6rrp3dqq

SDF-SRN: Learning Signed Distance 3D Object Reconstruction from Static Images [article]

Chen-Hsuan Lin, Chaoyang Wang, Simon Lucey
2020 arXiv   pre-print
SDF-SRN learns implicit 3D shape representations to handle arbitrary shape topologies that may exist in the datasets.  ...  To this end, we derive a novel differentiable rendering formulation for learning signed distance functions (SDF) from 2D silhouettes.  ...  CHL is supported by the NVIDIA Graduate Fellowship. This work was supported by the CMU Argo AI Center for Autonomous Vehicle Research.  ... 
arXiv:2010.10505v1 fatcat:ucowbzjzt5anbjqwkpk7xw3ufu

DeepMesh: Differentiable Iso-Surface Extraction [article]

Benoit Guillard, Edoardo Remelli, Artem Lukoianov, Stephan R. Richter, Timur Bagautdinov, Pierre Baque, Pascal Fua
2022 arXiv   pre-print
We validate our theoretical insight through several applications: Single view 3D Reconstruction via Differentiable Rendering, Physically-Driven Shape Optimization, Full Scene 3D Reconstruction from Scans  ...  Our key insight is that by reasoning on how implicit field perturbations impact local surface geometry, one can ultimately differentiate the 3D location of surface samples with respect to the underlying  ...  ACKNOWLEDGMENTS This work was supported in part by the Swiss National Science Foundation.  ... 
arXiv:2106.11795v2 fatcat:lwuetyfe5nfo3fqtrwcy3qll5a
« Previous Showing results 1 — 15 out of 54,138 results