Voxels on Fire Ye Zhao Xiaoming Wei Zhe Fan Arie Kaufman Hong Qin * Center for Visual Computing and Department of Computer Science Stony Brook University Stony Brook, NY 11794-4400 Abstract We introduce a method for the animation of fire propagation and the burning consumption of objects represented as volumetric data sets. Our method uses a volumetric fire propagation model based on an enhanced distance field. It can simulate the spreading of multiple fire fronts over a specified isosurface without actually having to cre- ate that isosurface. The distance field is generated from a specific shell volume that rapidly creates narrow spatial bands around the virtual surface of any given isovalue. The complete distance field is then obtained by propagation from the initial bands. At each step multiple fire fronts can evolve simultaneously on the volumetric ob- ject. The flames of the fire are constructed from streams of particles whose movement is regulated by a velocity field generated with the hardware-accelerated Lattice Boltzmann Model (LBM). The LBM provides a physically-based simulation of the air flow around the burning object. The object voxels and the splats associated with the flame particles are rendered in the same pipeline so that the volume data with its external and internal structures can be displayed along with the fire. CR Categories: I.3.5 [Computer Graphics]: Computational Ge- ometry and Object Modeling—Physically Based Modeling; I.3.7 [Computer Graphics]: Three-Dimensional Graphics and Realism— Animation Keywords: Fire Propagation, Distance Field, Lattice Boltzmann Model, Splatting, GPU Acceleration 1 Introduction Simulating fire phenomena is important in many applications such as entertainment, visual simulation, battlefield visualization, and even landscape design. However, the visualization and animation of fire is difficult. Extensive studies [Beaudoin et al. 2001; Chiba et al. 1994; Lee et al. 2001; Nguyen et al. 2002; Perlin 1985; Per- lin and Hoffert 1989; Perry and Ricard 1994; Reeves 1983; Stam and Fiume 1995] have been conducted using different approaches to model and render the dynamic behavior of fire. A good sur- vey has been given by Nielson and Madsen [1999]. Over a decade ago, Perlin et al. [1985; 1989] presented a noise-based method to model fire where fractal perturbation is used to simulate its turbu- lent movements. This approach is easy to implement, however it * Email:{yezhao, wxiaomin, fzhe, ari, qin}@cs.sunysb.edu Figure 1: Fire on a volumetric table. The underlying noncom- bustible metal frame is revealed once the wooden outer layer is consumed. cannot describe a fire front propagation or external effects such as wind. Reeves [1983] proposed the use of a particle system to model fire. The motion of the fire particles was affected by external forces, such as gravity. Due to the discrete nature of particles, a huge num- ber of them was required to achieve good visual effects. Chiba et al. [1994] combined a user defined vortex-based velocity field and a 2D fuel map to describe the movement of fire. The finite difference solver for partial differential equations was used in both the work of Stam and Fiume [1995] and the work of Foster and Metaxas [1997] to simulate turbulent gas and fire. Qian et al. [1998] used a front tracking method to simulate infinitely thin premixed flame surface, which is explicitly represented by connected marker points. Recently, Nguyen et al. [2002] presented a method based on the Navier-Stokes equations to model fuel with hot gaseous products. Using the level set method to track the moving flame surface they produced realistic looking turbulent flames. Unlike these studies, we focus in this paper on the modeling of fire front propagation and the burning consumption of objects rep- resented by volumetric data sets, such as the table shows in Figure 1. Splatting is used to render the flames resulting in a realistic vi- sual effect. King et al. [2000] first used textured splats in their work to avoid the computational complexity of large particle sys- tems. In our earlier work [Wei et al. 2002], textured splats were adopted as the basic display primitives and the Lattice Boltzmann Model (LBM) was introduced to model the interaction of the fire