CIMS: Correction-Interpolation Method for Smoke Simulation

TL;DR

This paper introduces CIMS, a correction-interpolation method using deep learning to efficiently produce high frame rate smoke simulations with high accuracy, significantly reducing errors compared to traditional low frame rate simulations.

Contribution

The paper presents a novel correction-interpolation approach that employs a U-Net-based neural network with image modeling techniques to improve smoke simulation accuracy at larger time-steps.

Findings

Reduces mean squared error of large time-step simulations by over 80%.

Achieves 2.04 times higher accuracy than previous DNN-based methods.

Maintains computational efficiency with minimal impact on overall simulation time.

Abstract

In this paper, we propose CIMS: a novel correction-interpolation method for smoke simulation. The basis of our method is to first generate a low frame rate smoke simulation, then increase the frame rate using temporal interpolation. However, low frame rate smoke simulations are inaccurate as they require increasing the time-step. A simulation with a larger time-step produces results different from that of the original simulation with a small time-step. Therefore, the proposed method corrects the large time-step simulation results closer to the corresponding small time-step simulation results using a U-Net-based DNN model. To obtain more precise results, we applied modeling concepts used in the image domain, such as optical flow and perceptual loss. By correcting the large time-step simulation results and interpolating between them, the proposed method can efficiently and accurately generate high frame rate smoke simulations. We conduct qualitative and quantitative analyses to confirm the effectiveness of the proposed model. Our analyses show that our method reduces the mean squared error of large time-step simulation results by more than 80% on average. Our method also produces results closer to the ground truth than the previous DNN-based methods; it is on average 2.04 times more accurate than previous works. In addition, the computation time of the proposed correction method barely affects the overall computation time.