NeX360: Real-Time All-Around View Synthesis With Neural Basis Expansion

Article Computer Science, Software Engineering

Instant Neural Graphics Primitives with a Multiresolution Hash Encoding

Thomas Mueller et al.

Summary: The research introduces a versatile new input encoding that allows for a reduction in the cost of training and evaluation by using a multiresolution hash table in neural networks.

ACM TRANSACTIONS ON GRAPHICS (2022)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

Light Field Neural Rendering

Mohammed Suhail et al.

Summary: This paper proposes a model that combines light field rendering and geometric reconstruction methods, which can accurately represent view-dependent effects and learn the geometry of the scene from sparse views.

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR) (2022)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

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

Jonathan T. Barron et al.

Summary: Researchers propose a model called mip-NeRF 360 that uses non-linear scene parameterization, online distillation, and distortion-based regularization to address the challenges posed by unbounded scenes. Compared to existing models, it performs better in view synthesis and depth map generation, and is able to handle highly intricate unbounded real-world scenes.

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022) (2022)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

Plenoxels: Radiance Fields without Neural Networks

Sara Fridovich-Keil et al.

Summary: We introduce Plenoxels, a system for photorealistic view synthesis using a sparse 3D grid representation with spherical harmonics. This representation can be optimized without neural components and achieves a two orders of magnitude faster speed compared to Neural Radiance Fields on benchmark tasks while maintaining visual quality.

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022) (2022)

Add to Collection

Article Computer Science, Software Engineering

Neural Light Transport for Relighting and View Synthesis

Xiuming Zhang et al.

Summary: This article introduces an image-based light transport acquisition method, focusing on lighting for human bodies. It proposes a semi-parametric approach to learning a neural representation of LT embedded in a texture atlas, allowing the synthesis of novel views under different lighting conditions. The approach can handle complex material effects and global illumination while guaranteeing physical correctness of the diffuse LT.

ACM TRANSACTIONS ON GRAPHICS (2021)

Add to Collection

Article Computer Science, Software Engineering

Point-Based Neural Rendering with Per-View Optimization

T. Neff et al.

Summary: Recently, there has been a growing interest in neural rendering methods. This study introduces a general approach that utilizes MVS for initialization but allows further optimization of scene properties in the input view space, resulting in improved quality and speed in novel-view synthesis compared to previous methods.

COMPUTER GRAPHICS FORUM (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

GRF: Learning a General Radiance Field for 3D Representation and Rendering

Alex Trevithick et al.

Summary: The translation introduces a neural network that represents and renders 3D objects and scenes from 2D observations. Key to the approach is learning local features for each pixel and projecting them to 3D points, resulting in high-quality and realistic novel views.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

FastNeRF: High-Fidelity Neural Rendering at 200FPS

Stephan J. Garbin et al.

Summary: This study introduces FastNeRF, the first NeRF-based system capable of rendering high-fidelity photorealistic images at 200Hz on a high-end consumer GPU. Through a graphics-inspired factorization, the method achieves compact caching of deep radiance maps and efficient querying, making it 3000 times faster than the original NeRF algorithm.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fields

Jonathan T. Barron et al.

Summary: Mip-NeRF is an extension of NeRF that reduces aliasing artifacts, improves fine detail representation, and is more efficient in speed and size compared to NeRF.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence