Radiance Fields from Photons

Capturing neural radiance fields and Gaussian splats at high speeds and without dataset careful curation

Neural radiance fields, or NeRFs, have become the de facto approach for high-quality view synthesis from a collection of images captured from multiple viewpoints. However, many issues remain when capturing images in-the-wild under challenging conditions, such as low light, high dynamic range, or rapid motion leading to smeared reconstructions with noticeable artifacts. In this work, we introduce quanta radiance fields, a novel class of neural radiance fields that are trained at the granularity of individual photons using single-photon cameras (SPCs). We develop theory and practical computational techniques for building radiance fields and estimating dense camera poses from unconventional, stochastic, and high-speed binary frame sequences captured by SPCs. We demonstrate, both via simulations and a SPC hardware prototype, high-fidelity reconstructions under high-speed motion, in low light, and for extreme dynamic range settings.

Publications

Radiance Fields from Photons

Sacha Jungerman, Aryan Garg, Mohit Gupta

ACM Trans. on Graphics, presented at SIGGRAPH 2026

pdf
@article{10.1145/3770578,
author = {Jungerman, Sacha and Garg, Aryan and Gupta, Mohit},
title = {Radiance Fields from Photons},
year = {2025},
issue_date = {February 2026},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
volume = {45},
number = {1},
issn = {0730-0301},
url = {https://doi.org/10.1145/3770578},
doi = {10.1145/3770578},
abstract = {Neural radiance fields, or NeRFs, have become the de facto approach for high-quality view synthesis from a collection of images captured from multiple viewpoints. However, many issues remain when capturing images in-the-wild under challenging conditions, such as in low light, high dynamic range, or with rapid motion, leading to smeared reconstructions with noticeable artifacts. In this work, we introduce quanta radiance fields, a novel class of neural radiance fields that are trained at the granularity of individual photons using single-photon cameras (SPCs). We develop theory and practical computational techniques for building radiance fields and estimating dense camera poses from unconventional, stochastic, and high-speed binary frame sequences captured by SPCs. We demonstrate, both via simulations and a SPC hardware prototype, high-fidelity reconstructions under high-speed motion, in low light, and for extreme dynamic range settings.},
journal = {ACM Trans. Graph.},
month = oct,
articleno = {11},
numpages = {16},
keywords = {Neural radiance fields, pose estimation, single photon cameras, SPADs, high-speed cameras, high dynamic range \& low-light imaging, computational imaging}
}