Adaptively weighted discrete Laplacian for inverse rendering

Hyeonjang An, Wonjun Lee, Bochang Moon

Overview

This code is the official implementation of The Visual Computer paper, Adaptively weighted discrete Laplacian for inverse rendering. For the further informations, please refer to the project page.

Requirements

We recommend running the code under conda environment.

conda create -f environment.yml
conda activate awl

Usage

We have tested our discrete Laplacian in existing differentiable rendering frameworks, Large Steps in Inverse Rendering of Geometry, and Continuous Remeshing For Inverse Rendering. Please replace the existing Laplacian in the frameworks with ours for the test.

Example code

from lap import laplacian_cotangent, laplacian_adaptive
L_c = laplacian_cotangent(mesh.verts, mesh.faces)                   # cotangent Laplacian
L_a = laplacian_adaptive(mesh.verts, mesh.faces, LAMBDA, SCALE)     # adaptively weighted Laplacian

Here, LAMBDA, $\lambda$ is the smoothing factor of Laplacian smoothing and SCALE is global scaling parameter for different domains. We add the a framework-specific parameter SCALE. For example, SCALE is set as 0.1 in LSIG.

Issue

If there is an issue, please send an email to this address, [email protected]

License

All source codes are released under a BSD License

Citation

@article{An2023,
  title={Adaptively weighted discrete Laplacian for inverse rendering},
  author={An, Hyeonjang and Lee, Wonjun and Moon, Bochang},
  journal={The Visual Computer},
  year={2023},
  issn={1432-2315},
  doi={10.1007/s00371-023-02955-2},
  url={https://doi.org/10.1007/s00371-023-02955-2}
}