Pre-rendered training and testing dataset for the paper Deep Convolutional Reconstruction for Gradient-Domain Rendering

Description

Pre-rendered data for training and evaluating the Neural Reconstruction for Gradient-Domain Path Tracing ("NGPT") algorithm described in the paper Deep Convolutional Reconstruction for Gradient-Domain Rendering. See https://github.com/mkettune/ngpt/. Download 'all_datasets.zip' for easy access to all of the data. Alternatively, you may download the datasets one by one. The displayed license (CC BY 4.0) covers the rendering and collection work. Copyrights of the individual scenes may need to be respected as well and are displayed in file 'LICENSE'.

Year of publication

2019

Type of data

Authors

Aalto University

Department of Computer Science

Erik Härkönen - Creator

Jaakko Lehtinen - Creator

Markus Kettunen - Creator

Zenodo - Publisher

Project

Other information

Fields of science

Computer and information sciences

Language

Open access

Open

License

Creative Commons Attribution 4.0 International (CC BY 4.0)

Keywords

Subject headings

Temporal coverage

undefined