Key Techniques Study On Natural Phenomena Simulation For Virtual Reality Environment

Natural simulation is one of the challenging open problems in computer graphics. The people in many fields, such as computer animation, virtual reality and computer game, show great concern about the development of this subject. The main reason is that every progress in this subject can benefit them greatly. In recent years, this subject becoming a hot topic over the world gradually. However, there are some difficult issues need to be studied further. Since nature is complex and physical simulation takes great burden to current desktop computer, it is too difficult to simulate nature in real-time. A new generation graphics hardware powered with GPU was introduced recently. Since this kind of device has features both on parallel computation and programmability, many researchers turn to its help to solve complex computation and rendering problems. In this dissertation, I studied several key techniques in natural simulation, which includes physical lighting, sky lighting and generation of both perlin noise and flow noise. The solutions of these problems are largely dependent on GPU techniques. Firstly, because both of Cook-Torrance lighting model and Ashikhmin lighting model are computational complexities and low rendering speed, a deferred lighting method and a lighting method based on texture decomposition are presented to solve above two problems separately. These two new methods can efficiently accelerate rendering computation. Secondly, some materials such as jade are translucent which can cause subsurface scattering when ray enters this kind of objects. This makes lighting computation become much more complex. In this paper, a new method based on GPU techniques is purposed. Compare with BSSRDF, it is simple and can render jade object in an efficient without trade-off result image quality obviously. Physically based outdoor rendering is one of most important method to acquire high quality realistic rendering results. When sunlight travels through atmosphere, it will be scattered due to molecule and aerosol. A new method to fit sky and sky perspective real-time rendering computations is purposed in this paper. Upon the method one cannot concern the eye point location. The main advantage of this method is to move complex computation to GPU. Perlin noise and flow noise are important algorithm in simulating nature. Unfortunate they are computation expensive somehow. In the first, a new method to compute Perlin noise in real-time is presented. This method moves large amount computations to GPU side and can be calculated in very efficient way. Then we use quaternion representation for solving arbitrary rotation problem in flow noise, and also move heavily computation to GPU side. At last, we simulate fire with Perlin noise based on GPU computation.