Methods¶

Zip-NeRF¶

CamP: Camera Preconditioning for Neural Radiance Fields

Authors:: Keunhong Park, Philipp Henzler, Ben Mildenhall, Jonathan T. Barron, Ricardo Martin-Brualla
Paper:: https://arxiv.org/pdf/2308.10902.pdf
Web:: https://camp-nerf.github.io/
ID:: camp

CamP is an extension of Zip-NeRF which adds pose refinement to the training process.

Gaussian Opacity Fields¶

Gaussian Opacity Fields: Efficient and Compact Surface Reconstruction in Unbounded Scenes

Authors:: Zehao Yu, Torsten Sattler, Andreas Geiger
Paper:: https://arxiv.org/pdf/2404.10772.pdf
Web:: https://niujinshuchong.github.io/gaussian-opacity-fields/
ID:: gaussian-opacity-fields

Improved Mip-Splatting with better geometry.

Gaussian Splatting¶

3D Gaussian Splatting for Real-Time Radiance Field Rendering

Authors:: Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler, George Drettakis
Paper:: https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/3d_gaussian_splatting_low.pdf
Web:: https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/
ID:: gaussian-splatting

Official Gaussian Splatting implementation extended to support distorted camera models. It is fast to train (1 hous) and render (200 FPS).

Instant NGP¶

Instant Neural Graphics Primitives with a Multiresolution Hash Encoding

Authors:: Thomas Müller, Alex Evans, Christoph Schied, Alexander Keller
Paper:: https://nvlabs.github.io/instant-ngp/assets/mueller2022instant.pdf
Web:: https://nvlabs.github.io/instant-ngp/
ID:: instant-ngp

Instant-NGP is a method that uses hash-grid and a shallow MLP to accelerate training and rendering. This method trains very fast (~6 min) and renders also fast ~3 FPS.

K-Planes¶

K-Planes: Explicit Radiance Fields in Space, Time, and Appearance

Authors:: Sara Fridovich-Keil, Giacomo Meanti, Frederik Warburg, Benjamin Recht, Angjoo Kanazawa
Paper:: https://arxiv.org/pdf/2301.10241
Web:: https://sarafridov.github.io/K-Planes/
ID:: kplanes

K-Planes is a NeRF-based method representing d-dimensional space using 2 planes allowing for a seamless way to go from static (d=3) to dynamic (d=4) scenes.

Mip-Splatting¶

Mip-Splatting: Alias-free 3D Gaussian Splatting

Authors:: Zehao Yu, Anpei Chen, Binbin Huang, Torsten Sattler, Andreas Geiger
Paper:: https://arxiv.org/pdf/2311.16493.pdf
Web:: https://niujinshuchong.github.io/mip-splatting/
ID:: mip-splatting

A modification of Gaussian Splatting designed to better handle aliasing artifacts.

Mip-NeRF 360¶

Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fields

Authors:: Jonathan T. Barron, Ben Mildenhall, Dor Verbin, Pratul P. Srinivasan, Peter Hedman
Paper:: https://arxiv.org/pdf/2111.12077.pdf
Web:: https://jonbarron.info/mipnerf360/
ID:: mipnerf360

Official Mip-NeRF 360 implementation addapted to handle different camera distortion/intrinsic parameters. It was designed for unbounded object-centric 360-degree capture and handles anti-aliasing well. It is, however slower to train and render compared to other approaches.

NeRF¶

NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis

Authors:: Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, Ren Ng
Paper:: https://arxiv.org/pdf/2003.08934.pdf
Web:: https://www.matthewtancik.com/nerf
ID:: nerf

Original NeRF method representing radiance field using a large MLP.

NerfStudio¶

Nerfstudio: A Modular Framework for Neural Radiance Field Development

Authors:: Matthew Tancik, Ethan Weber, Evonne Ng, Ruilong Li, Brent Yi, Justin Kerr, Terrance Wang, Alexander Kristoffersen, Jake Austin, Kamyar Salahi, Abhik Ahuja, David McAllister, Angjoo Kanazawa
Paper:: https://arxiv.org/pdf/2302.04264.pdf
Web:: https://docs.nerf.studio/
ID:: nerfacto

NerfStudio (Nerfacto) is a method based on Instant-NGP which combines several improvements from different papers to achieve good quality on real-world scenes captured under normal conditions. It is fast to train (12 min) and render speed is ~1 FPS.

NeRF On-the-go¶

NeRF On-the-go: Exploiting Uncertainty for Distractor-free NeRFs in the Wild

Authors:: Weining Ren, Zihan Zhu, Boyang Sun, Julia Chen, Marc Pollefeys, Songyou Peng
Paper:: https://arxiv.org/pdf/2405.18715.pdf
Web:: https://rwn17.github.io/nerf-on-the-go/
ID:: nerfonthego

NeRF On-the-go enables novel view synthesis in in-the-wild scenes from casually captured images.

NeRF-W (reimplementation)¶

ID:: nerfw-reimpl

Unofficial reimplementation of NeRF-W. Does not reach the performance reported in the original paper, but is widely used for benchmarking.

TensoRF¶

TensoRF: Tensorial Radiance Fields

Authors:: Anpei Chen, Zexiang Xu, Andreas Geiger, Jingyi Yu, Hao Su
Paper:: https://arxiv.org/pdf/2203.09517.pdf
Web:: https://apchenstu.github.io/TensoRF/
ID:: tensorf

TensoRF factorizes the radiance field into a multiple compact low-rank tensor components. It was designed and tester primarily on Blender, LLFF, and NSVF datasets.

Tetra-NeRF¶

Tetra-NeRF: Representing Neural Radiance Fields Using Tetrahedra

Authors:: Jonas Kulhanek, Torsten Sattler
Paper:: https://arxiv.org/pdf/2304.09987.pdf
Web:: https://jkulhanek.com/tetra-nerf
ID:: tetra-nerf

Tetra-NeRF is a method that represents the scene as tetrahedral mesh obtained using Delaunay tetrahedralization. The input point cloud has to be provided (for COLMAP datasets the point cloud is automatically extracted). This is the official implementation from the paper.

TRIPS¶

TRIPS: Trilinear Point Splatting for Real-Time Radiance Field Rendering

Authors:: Linus Franke, Darius Rückert, Laura Fink, Marc Stamminger
Paper:: https://arxiv.org/pdf/2401.06003
Web:: https://lfranke.github.io/trips/
ID:: trips

TRIPS performs point splatting into feature pyramid processed using CNNs. Speed comparable to Gaussian Splatting

Zip-NeRF¶

Zip-NeRF: Anti-Aliased Grid-Based Neural Radiance Fields

Authors:: Jonathan T. Barron, Ben Mildenhall, Dor Verbin, Pratul P. Srinivasan, Peter Hedman
Paper:: https://arxiv.org/pdf/2304.06706.pdf
Web:: https://jonbarron.info/zipnerf/
ID:: zipnerf

Zip-NeRF is a radiance field method which addresses the aliasing problem in the case of hash-grid based methods (iNGP-based). Instead of sampling along the ray it samples along a spiral path - approximating integration along the frustum.