Saved in:
Bibliographic Details
Main Authors: Sun, Cheng, Cai, Guangyan, Li, Zhengqin, Yan, Kai, Zhang, Cheng, Marshall, Carl, Huang, Jia-Bin, Zhao, Shuang, Dong, Zhao
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2304.13445
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913217876328448
author Sun, Cheng
Cai, Guangyan
Li, Zhengqin
Yan, Kai
Zhang, Cheng
Marshall, Carl
Huang, Jia-Bin
Zhao, Shuang
Dong, Zhao
author_facet Sun, Cheng
Cai, Guangyan
Li, Zhengqin
Yan, Kai
Zhang, Cheng
Marshall, Carl
Huang, Jia-Bin
Zhao, Shuang
Dong, Zhao
contents Reconstructing the shape and spatially varying surface appearances of a physical-world object as well as its surrounding illumination based on 2D images (e.g., photographs) of the object has been a long-standing problem in computer vision and graphics. In this paper, we introduce an accurate and highly efficient object reconstruction pipeline combining neural based object reconstruction and physics-based inverse rendering (PBIR). Our pipeline firstly leverages a neural SDF based shape reconstruction to produce high-quality but potentially imperfect object shape. Then, we introduce a neural material and lighting distillation stage to achieve high-quality predictions for material and illumination. In the last stage, initialized by the neural predictions, we perform PBIR to refine the initial results and obtain the final high-quality reconstruction of object shape, material, and illumination. Experimental results demonstrate our pipeline significantly outperforms existing methods quality-wise and performance-wise.
format Preprint
id arxiv_https___arxiv_org_abs_2304_13445
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Neural-PBIR Reconstruction of Shape, Material, and Illumination
Sun, Cheng
Cai, Guangyan
Li, Zhengqin
Yan, Kai
Zhang, Cheng
Marshall, Carl
Huang, Jia-Bin
Zhao, Shuang
Dong, Zhao
Computer Vision and Pattern Recognition
Reconstructing the shape and spatially varying surface appearances of a physical-world object as well as its surrounding illumination based on 2D images (e.g., photographs) of the object has been a long-standing problem in computer vision and graphics. In this paper, we introduce an accurate and highly efficient object reconstruction pipeline combining neural based object reconstruction and physics-based inverse rendering (PBIR). Our pipeline firstly leverages a neural SDF based shape reconstruction to produce high-quality but potentially imperfect object shape. Then, we introduce a neural material and lighting distillation stage to achieve high-quality predictions for material and illumination. In the last stage, initialized by the neural predictions, we perform PBIR to refine the initial results and obtain the final high-quality reconstruction of object shape, material, and illumination. Experimental results demonstrate our pipeline significantly outperforms existing methods quality-wise and performance-wise.
title Neural-PBIR Reconstruction of Shape, Material, and Illumination
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2304.13445