Frequency-Aware Reconstruction of Fluid Simulations with Generative Networks
View/ Open
Date
2020Author
Biland, Simon
Azevedo, Vinicius C.
Kim, Byungsoo
Solenthaler, Barbara
Metadata
Show full item recordAbstract
Convolutional neural networks were recently employed to fully reconstruct fluid simulation data from a set of reduced parameters. However, since (de-)convolutions traditionally trained with supervised l1-loss functions do not discriminate between low and high frequencies in the data, the error is not minimized efficiently for higher bands. This directly correlates with the quality of the perceived results, since missing high frequency details are easily noticeable. In this paper, we analyze the reconstruction quality of generative networks and present a frequency-aware loss function that is able to focus on specific bands of the dataset during training time. We show that our approach improves reconstruction quality of fluid simulation data in mid-frequency bands, yielding perceptually better results while requiring comparable training time.
BibTeX
@inproceedings {s.20201019,
booktitle = {Eurographics 2020 - Short Papers},
editor = {Wilkie, Alexander and Banterle, Francesco},
title = {{Frequency-Aware Reconstruction of Fluid Simulations with Generative Networks}},
author = {Biland, Simon and Azevedo, Vinicius C. and Kim, Byungsoo and Solenthaler, Barbara},
year = {2020},
publisher = {The Eurographics Association},
ISSN = {1017-4656},
ISBN = {978-3-03868-101-4},
DOI = {10.2312/egs.20201019}
}
booktitle = {Eurographics 2020 - Short Papers},
editor = {Wilkie, Alexander and Banterle, Francesco},
title = {{Frequency-Aware Reconstruction of Fluid Simulations with Generative Networks}},
author = {Biland, Simon and Azevedo, Vinicius C. and Kim, Byungsoo and Solenthaler, Barbara},
year = {2020},
publisher = {The Eurographics Association},
ISSN = {1017-4656},
ISBN = {978-3-03868-101-4},
DOI = {10.2312/egs.20201019}
}