Abstract : In this paper, we introduce a geometric approach-based particle-grid hybrid framework for efficiently modeling and rendering realistic, opaque ice shapes containing air bubbles. To achieve this, we diffuse water temperature using a grid-based method and represent air bubbles within the ice shape using particles. To address the issues of noise and distorted dissolved air fields in previous methods, we propose the following improvements: 1) We synthesize the level-set of the ice shape into a spherical form to reduce noise and streamline the algorithm. 2) We propose an efficient method to control the termination condition of heat diffusion by utilizing the number of active particles and the initial dissolved oxygen level. 3) Unlike previous methods that calculated the dissolved air field only near air bubbles, we extend the diffusion to transparent regions, enabling the realistic representation of opaque ice shapes. 4) To accurately compute the directional properties of air bubbles, we introduce a level-set-based approach. 5) We extend the solver to control the generation of dissolved air fields and air bubbles based on water temperature. 6) Finally, we present a method to realistically visualize opaque ice shapes using Unreal Engine (UE). As a result, the proposed method achieves approximately three times faster computation speed compared to previous techniques while providing visually enhanced opaque ice shapes in real-time. This advancement makes it applicable to physics-based virtual ice shape simulations.

[paper]