Cumulus Cloud Synthesis with Similarity Solution and Particle/Voxel

Cumulus Cloud Synthesis with Similarity
Solution and Particle/Voxel Modeling
Bei Wang, Jingliang Peng, and C.-C. Jay Kuo
Department of Electrical Engineering, University of Southern California
[email protected], [email protected], [email protected]
Abstract. A realistic yet simple 3D cloud synthesis method is examined
in this work. Our synthesis framework consists of two main components.
First, by introducing a similarity approach for physics-based cumulus
cloud simulation to reduce the computational complexity, we introduce
a particle system for cloud modeling. Second, we adopt a voxel-based
scheme to model the water phase transition effect and the cloud fractal structure. It is demonstrated by computer simulation that the proposed synthesis algorithm generates realistic 3D cumulus clouds with
high computational efficiency and enables flexible cloud shape control
and incorporation of the wind effect . . .
1
Introduction
Clouds are among the most natural atmospheric phenomena in our daily life. To
generate vivid outdoor scenes, realistic cloud synthesis and animation is often a
key component in many 3D applications such as flight simulation, gaming and
3D movie production. Realistic cloud synthesis, albeit important, is a challenging
task due to the infinite variations in cloud shape.
Physics-based simulation gives the most natural-looking appearance of synthesized clouds. However, the physics-based simulation demands a high computational complexity to solve the Navier-Stokes partial differential equations
(PDEs). In contrast with the physics-based approach, another approach of lower
complexity known as the procedural method has been studied as well. Nevertherless, the procedure approach has less flexibility in cloud shape control corresponding to dynamics. We adopt the physics-based simulation of cumulus clouds
in this work and aim to lower the complexity. The goals and contributions of our
current research are summarized as: 1)Computational efficiency: by adopting
the similarity solution approach, we are able to catch the general characteristics
of clouds with a set of constant parameters without solving PDEs explicitly;
2)Realistic visual appearance: the proposed method is able to create the
general shape and amorphous appearance of a cloud as well as fine details such
as fractal substructures around boundary regions; 3) Flexible user control:
by modifying the setting of the constant parameters which describes the general
characteristics of cloud, users can control the cloud shape flexibly.
Technically, the above features are accomplished by the adoption of the similarity simulation approach as well as two powerful graphic modeling tools;
G. Bebis et al. (Eds.): ISVC 2008, Part I, LNCS 5358, pp. 65–74, 2008.
c Springer-Verlag Berlin Heidelberg 2008
66
B. Wang, J. Peng, and C.-C.J. Kuo
namely, particles and voxels. The particle system provides a natural model for
thermal-based cloud representation. In the first stage, each particle is used to
simulate the movement of a thermal. Then in the second stage, a grid structure
within a bounding box is utilized to model the accumulation and condensation
of water vapor particles.
The rest of this paper is organized as follows. Previous work is reviewed in
section 2. Three major components of the proposed cloud synthesizer are described in sections 3-5. Finally, experimental results and conclusion are given in
section 6 and section 7.
2
Review of Previous Work
Research on cloud synthesis, including simulation, modeling and rendering, has
history of more than two decades. In early years, methods of low complexity
were prevailing in the field due to the limitation of computing resource. With
the increase in computing power in recent years, methods of high complexity
have begun to thrive. Generally speaking, previous cloud synthesis methods can
be classified into two main categories: visual-based and physics-based methods.
Visual-based methods generate clouds using the amorphous cloud property
rather than the physical process of cloud formation. Examples include procedural
modeling [1], fractals modeling [2], qualitative simulation [3], and texture sprite
modeling [4]. These methods have attractive advantage with low computation
and easy implementation. They also provide the ability to change the cloud
shape through the parameters adjustment. However, modification of the cloud
shape is often achieved by trial-and-error, which has limited dynamics extension.
On the contrary, Physics-based methods [5] incorporate fluid dynamics in
cloud synthesis. They usually solve a set of PDEs in the simulation process (i.e.
the Navier-Stoke’s equation) so that their computational complexity is very high.
The computational cost was significantly reduced by Stam [6], who solved the
N.S. equations with simplified fluid dynamics. Thereafter, his work was applied
on cloud simulation with important physics process(phase transition) in cloud
formation considered by Harris [7], besides, Harris implemented Stam’s stable
fluid solver on GPU to achieve real-time simulation. Similarly, Miyazaki [8] used
the Coupled Map Lattice (CML) to generate various types of clouds. In Harris
and Ryo’s work, cloud simulation depends on initial water vapor condition and a
set of parameters such as viscosity. However, they don’t provide a straightforward
solution to cloud shape control, which is often a desirable feature.
When compared with previous physics-based method, our proposed cumulus
synthesis method does not solve PDEs explicitly, but adopt an analytic solution
with some parameters consistent to the Navier-Stokes equations, based on which
we had another cloud simulation work on decoupled modeling [9], and take the
water phase transition into account. By choosing the parameters properly, we
can generate different clouds efficiently.