A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Filters
Neural-PIL: Neural Pre-Integrated Lighting for Reflectance Decomposition
[article]
2021
arXiv
pre-print
Preserved Fulltext
Our key technique is a novel illumination integration network called Neural-PIL that replaces a costly illumination integral operation in the rendering with a simple network query. ...
Project page: https://markboss.me/publication/2021-neural-pil/ ...
In contrast, our Neural-PIL is an MLP that directly produces pre-integrated light required for the rendering. ...
arXiv:2110.14373v1
fatcat:vg5s62ydwnffzivbyrqdu2g5tu
Extracting Triangular 3D Models, Materials, and Lighting From Images
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We present an efficient method for joint optimization of topology, materials and lighting from multi-view image observations. ...
Finally, we introduce a differentiable formulation of the split sum approximation of environment lighting to efficiently recover all-frequency lighting. ...
In contrast, we propose a differentiable split sum lighting model, also adopted by the concurrent work Neural-PIL [6] . ...
arXiv:2111.12503v3
fatcat:pnd3yiexwvcqlkgxbebcwadlki
NeROIC: Neural Rendering of Objects from Online Image Collections
[article]
2022
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Extensive evaluations and comparisons demonstrate the advantages of our approach in capturing high-quality geometry and appearance properties useful for rendering applications. ...
Using a multi-stage approach extending neural radiance fields, we first infer the surface geometry and refine the coarsely estimated initial camera parameters, while leveraging coarse foreground object ...
We will release our code, pre-trained models, and training datasets upon publication to facilitate further research effort in this area.
Related Work Neural Rendering for Novel View Synthesis. ...
doi:10.48550/arxiv.2201.02533
fatcat:ymyxxmpk2vfppfbzcuvhaozmda