Budget Amount *help |
¥3,500,000 (Direct Cost: ¥3,500,000)
Fiscal Year 2004: ¥1,900,000 (Direct Cost: ¥1,900,000)
Fiscal Year 2003: ¥1,600,000 (Direct Cost: ¥1,600,000)
|
Research Abstract |
Mathematical fluid modeling is generally divided into two major categories : Eulerian and Lagrangian. The former formulates the fluid behavior by defining related physical fields, whereas the latter regards the fluid as a set of particles to solve the multiple body problem. The two approaches are accompanied with different visualization schemes : that is surface rendering (Eulerian) and point-based rendering (Lagranian). Meanwhile, hybrid simulation has recently been exploited to couple these two approaches for describing complex phenomena arising in many science and engineering problems. Therefore, there is an urgent need to establish simultaneous rendering of continuous and discontinuous objects. Pfister et al. proposed in 2000, a novel modeling technique, called surfels (surface cells), which relies on a point-based approximation to render complex shape of rigid body surfaces. In this study, we have extended the concept to visualize scattered objects as well by adding such features a
… More
s the radius and opacity distribution to describe the disk of surfel primitives, to yield a unified framework for representing physical fields and particle system. Then a prototype rendering system has been implemented on a standard workstation, where all of the traditional techniques such as surface rendering, volume rendering and particle rendering can be executed, and its feasibility has been proven with application to the following two specific problems : (1)Stratum disposal dataset (data courtesy of Fuji Research Institute) A particle-based model was used to simulate how nuclear waste infiltrate into a porous media over the long term. By using our system, scattered distribution of 200,000 colloid surfels with a relatively small radius is superimposed naturally on overlapped surfels to visualize inner boundaries of the porous media. (2)Breaking wave dataset (data coutesy of S.Koshizuka, University of Tokyo) Breaking wave is still known as one of the challenging topics in the research on interactive image synthesis of natural phenomena. In our study, the dynamics of 2D breaking wave was modeled using the Moving Particle Semi-implicit method. Then, a quasi-3D dataset of breaking wave was generated with parallel interpolation and local particle generation. We took full advantage of extended surfel primitives with varying radius and Gaussian opacity distribution to visualize the components of breaking waves, namely swell and riffle as well as splash, in a consistent manner. Less
|