GAN-based NeRF Noise Simulation in Mesh Denoising Task

Vladimir Mashurov, Vasilii Latonov, Anastasia Martynova, Natalia Semenova

Keywords: Generative networks, Noise generation, 3D Denoising, Pointclouds

Abstract: In the present paper, we propose a new approach and a dataset for generating NeRF-like noise on the mesh surface. Our approach is based on GAN and was trained on a dataset that we collect using real NeRF noise. The core idea of our method lies in the use of graph convolutions in the generator. Our pipeline demonstrates generated NeRF-like noise more accurate than other methods by mesh denoising benchmarking. We also present a new NeRF noise analysis approach HTPH based on a conditional probability model to measure the similarity of mesh noise.

Primary Area: generative models

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 11821