Efficient Differentiation of Pixel Reconstruction Filters
for Path-Space Differentiable Rendering
Zihan Yu1, Cheng Zhang1,2, Derek Nowrouzezahrai3, Zhao Dong2, and Shuang Zhao1
1University of California, Irvine          2Meta Reality Labs          3McGill University
ACM Transactions on Graphics (SIGGRAPH Asia 2022), 41(6), 2022
teaser
Abstract

Pixel reconstruction filters play an important role in physics-based rendering and have been thoroughly studied. In physics-based differentiable rendering, however, the proper treatment of pixel filters remains largely under-explored. We present a new technique to efficiently differentiate pixel reconstruction filters based on the path-space formulation. Specifically, we formulate the pixel boundary integral that models discontinuities in pixel filters and introduce new antithetic sampling methods that support differentiable path sampling methods, such as adjoint particle tracing and bidirectional path tracing. We demonstrate both the need and efficacy of antithetic sampling when estimating this integral, and we evaluate its effectiveness across several differentiable- and inverse-rendering settings.

Downloads
  • Paper: pdf (22 MB)
  • Supplemental material: html, zip (310 MB)
  • Code: Github (coming soon)
Bibtex citation
@article{Yu:2022:Diff-Pixel,
  title={Efficient Differentiation of Pixel Reconstruction Filters for Path-Space Differentiable Rendering},
  author={Yu, Zihan and Zhang, Cheng and Nowrouzezahrai, Derek and Dong, Zhao and Zhao, Shuang},
  journal={ACM Trans. Graph.},
  volume={41},
  number={6},
  year={2022},
  pages={??:1--??:16}
}
Acknowledgments

We thank the anonymous reviewers for their constructive comments. This work was partially supported by NSF grant 1900927 and Meta Reality Labs.