Frequency-Aware Reconstruction of Fluid Simulations with Generative Networks
This work addresses the issue of missing high-frequency details in fluid simulations for computer graphics applications, representing an incremental improvement over existing methods.
The paper tackled the problem of fluid simulation reconstruction using generative networks, where traditional loss functions fail to efficiently minimize errors in high-frequency details. The authors introduced a frequency-aware loss function that improved reconstruction quality in mid-frequency bands, yielding perceptually better results with comparable training time.
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.