Frequency-Aware Reconstruction of Fluid Simulations with Generative Networks

Abstract

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.

arXiv commentSubmitted to Eurographics2020

References in corpus (6)

• BEGAN: Boundary Equilibrium Generative Adversarial Networks
• Frequency Principle: Fourier Analysis Sheds Light on Deep Neural Networks
• Accelerating Eulerian Fluid Simulation With Convolutional Networks
• Stable and Controllable Neural Texture Synthesis and Style Transfer Using Histogram Losses
• On the Spectral Bias of Neural Networks
• Image Super-Resolution Using a Wavelet-based Generative Adversarial Network

Cited by in corpus (4)

• Frequency Principle: Fourier Analysis Sheds Light on Deep Neural Networks
• Linear Frequency Principle Model to Understand the Absence of Overfitting in Neural Networks
• Deep frequency principle towards understanding why deeper learning is faster
• An Upper Limit of Decaying Rate with Respect to Frequency in Deep Neural Network