← 3D Vision

NeRF & 3D Reconstruction

Neural Radiance Fields and classical 3D reconstruction — from SfM and MVS to text/image-to-3D generation pipelines.

⌘K

NeRF Foundation Models (2020–2025) 22+ papers

Model / MethodFull TitleVenueYear
NeRFRepresenting Scenes as Neural Radiance Fields for View Synthesis ECCV2020
Mip-NeRFA Multiscale Representation for Anti-Aliasing Neural Radiance Fields ICCV2021
Mip-NeRF 360Unbounded Anti-Aliased Neural Radiance Fields CVPR2022
Instant NGPInstant Neural Graphics Primitives with a Multiresolution Hash Encoding SIGGRAPH2022
TensoRFTensorial Radiance Fields ECCV2022
Point-NeRFPoint-based Neural Radiance Fields CVPR2022
Zip-NeRFAnti-Aliased Grid-Based Neural Radiance Fields ICCV2023
F2-NeRFFast Neural Radiance Field Training with Free Camera Trajectories CVPR2023
K-PlanesExplicit Radiance Fields in Space, Time, and Appearance CVPR2023
NeuSLearning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction NeurIPS2021
NeuS2Fast Learning of Neural Implicit Surfaces for Multi-view Reconstruction ICCV2023
VolSDFVolume Rendering of Neural Implicit Surfaces NeurIPS2021
UNISURFUnifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction ICCV2021
NerfactoNerfstudio: A Modular Framework for Neural Radiance Field Development SIGGRAPH2023
NeRFPlayerA Streamable Dynamic Scene Representation with Decomposed Neural Radiance Fields CVPR2023
NeRF-XLNeRF-XL: Scaling NeRFs with Multiple GPUs ECCV2024
LIRMLIRM: Large Inverse Rendering Model for Progressive Reconstruction of Shape, Materials and Radiance Fields arXiv2025
LaRaLaRa: Efficient Large-Baseline Radiance Fields ECCV2024
3DGS3D Gaussian Splatting for Real-Time Radiance Field Rendering SIGGRAPH2023

Text-to-3D & Image-to-3D Generation (2022–2026) 35+ papers

Model / MethodFull TitleVenueYear
DreamCraft3D++DreamCraft3D++: Efficient Hierarchical 3D Generation with Multi-Plane Reconstruction Model arXiv2025
Wonder3DWonder3D: Single Image to 3D using Cross-Domain Diffusion CVPR2024
Era3DEra3D: High-Resolution Multiview Diffusion using Efficient Row-wise Attention NeurIPS2024
CRMCRM: Single Image to 3D Textured Mesh with Convolutional Reconstruction Model arXiv2024
Unique3DUnique3D: High-Quality and Efficient 3D Mesh Generation from a Single Image arXiv2024
DreamFusionText-to-3D using 2D Diffusion ICLR2023
Magic3DHigh-Resolution Text-to-3D Content Creation CVPR2023
ProlificDreamerHigh-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillation NeurIPS2023
Zero-1-to-3Zero-shot One Image to 3D Object ICCV2023
One-2-3-45Any Single Image to 3D Mesh in 45 Seconds without Per-Shape Optimization NeurIPS2023
Instant3DFast Text-to-3D with Sparse-View Generation and Large Reconstruction Model ICLR2024
TripoSRFast 3D Object Reconstruction from a Single Image arXiv2024
Step1X-3DTowards High-Fidelity and Controllable Generation of Textured 3D Assets arXiv2025
TriplaneTurboProgressive Rendering Distillation for Instant Text-to-Mesh (1.2s) CVPR2025
Meta 3D AssetGenText-to-Mesh with High-Fidelity Texture and PBR Materials arXiv2024
SAM 3D3Dfy Anything in Images — Geometry, Texture, and Layout from Single Images arXiv2025
TextMesh4DText-to-4D Mesh via Jacobian Deformation Fields arXiv2025
LRMLarge Reconstruction Model for Single Image to 3D ICLR2024
LGMLarge Multi-View Gaussian Model for High-Resolution 3D Content Creation ECCV2024
SV3DNovel Multi-view Synthesis and 3D Generation from a Single Image using Latent Video Diffusion ECCV2024
DMV3DDenoising Multi-View Diffusion using 3D Large Reconstruction Model ICLR2024
CLAYCLAY: A Controllable Large-scale Generative Model for Creating High-quality 3D Assets arXiv2024
Hunyuan3D 1.0Hunyuan3D 1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation Tencent2024
Hunyuan3D 2.0Hunyuan3D 2.0: Scaling Diffusion Models for High-Resolution Textured 3D Asset Generation Tencent2025
Hunyuan3D-OmniHunyuan3D-Omni: A Unified Framework for Controllable 3D Asset Generation Tencent2025
TRELLISTRELLIS: Structured 3D Latents for Scalable and Versatile 3D Generation Microsoft2024
TRELLIS.2TRELLIS.2: Native and Compact Structured Latents for 3D Generation Microsoft2025
InstantMeshInstantMesh: Efficient 3D Mesh Generation from a Single Image with Sparse-view Large Reconstruction Model NeurIPS2024
MeshAnythingMeshAnything: Artist-Created Mesh Generation with Autoregressive Transformers arXiv2024
MeshAnything V2MeshAnything V2: Artist-Created Mesh Generation with Adjacent Mesh Tokenization arXiv2025
GRMGRM: Large Gaussian Reconstruction Model for Efficient 3D Reconstruction and Generation ICML2024

Structure from Motion & Multi-View Stereo 15+ papers

Model / MethodFull TitleVenueYear
COLMAPStructure-from-Motion Revisited CVPR2016
OpenMVGOpen Multiple View Geometry library Open-source2015
OpenMVSOpen Multi-View Stereo reconstruction library Open-source2020
PMVSAccurate, Dense, and Robust Multiview Stereopsis PAMI2010
MVSNetDepth Inference for Unstructured Multi-view Stereo ECCV2018
CasMVSNetCascade Cost Volume for High-Resolution Multi-View Stereo CVPR2020
TransMVSNetGlobal Context-aware Multi-view Stereo Network with Transformers CVPR2022
DeepMVSLearning Multi-View Stereopsis CVPR2018
RayNetLearning Volumetric 3D Reconstruction with Ray Potentials CVPR2018
Vis-MVSNetVisibility-Aware Multi-View Stereo Network IJCV2022
SurfaceNetAn End-to-end 3D Neural Network for Multiview Stereopsis ICCV2017
PolyFitPolygonal Surface Reconstruction from Point Clouds ICCV2017

3D Datasets & Benchmarks 20+ datasets

Dataset / BenchmarkDescriptionScaleYear
ShapeNetLarge-scale 3D shape repository with rich annotations 51k models, 55 categories2015
ModelNetPrinceton 3D object classification benchmark 127k CAD models, 662 categories2015
ScanNetRichly-annotated 3D reconstructions of indoor scenes 1513 scenes2017
Tanks & TemplesLarge-scale scene reconstruction benchmark 21 scenes2017
DTURobot Image Data Sets for MVS evaluation 80 scenes2014
Matterport3DRGB-D dataset with panoramic views from buildings 10.8k panoramas, 194k images2017
KITTIVision benchmark for autonomous driving Multiple sequences2012
ObjaverseA Universe of Annotated 3D Objects (800k+ models) 800k+ models2023
Objaverse-XL10M+ 3D objects from diverse sources 10M+ objects2023
PartNetFine-grained, hierarchical part-level 3D object understanding 26k models2019
S3DISStanford Large-Scale 3D Indoor Spaces 6 areas2016
nuScenesLarge-scale autonomous driving dataset with 3D annotations 1000 scenes2019
Waymo OpenHigh-resolution sensor data from autonomous vehicles 1150 scenes2020

Open-Source SfM & MVS Software 12+ tools

ProjectLanguageLicenseNotes
COLMAPC++BSD 3-ClauseSfM + MVS, most widely used
OpenMVGC++MPL2Multiple View Geometry library
OpenMVSC++ (CUDA opt.) AGPL3Dense reconstruction from SfM output
OpenSfMPythonBSDMapillary's SfM pipeline
TheiaSfMC++BSDEfficient global SfM
MVEC++BSD 3-ClauseMulti-View Environment
NerfstudioPythonApache-2.0Modular NeRF development framework
threestudioPythonApache-2.0Unified framework for 3D generation
PCLC++BSD 3-ClausePoint Cloud Library for 3D processing
Open3DC++/PythonMIT3D data processing library