![]() ![]() ![]() ![]() ![]() ![]() |
| Japanese page is here. |
SIGGRAPH '98 Exhibition |
Novel techniques of this project have been demonstrated at
SIGGRAPH '98 Exhibition
that is the world-largest conference for
computer graphics technology.
The demonstration is shown as follows:
Radiosity
The first demonstration is radiosity, which is an algorithm for computer graphics. Radiosity method first divides a given geometric model into meshes, and then calculates the converged light energy for each element of the meshes. In this demonstration, the bubble mesh method is used for and coarse elements in other regions, so that it represents photo-realistic shadows and highlights, while it avoids the increase of geometric data. This page also shows
![]()
Mesh deformation
The next demonstration shows the shape design technique, which applies the deformation of meshes.
This demonstration first generates a graded mesh inside a rectangular region, using the bubble mesh method. It then deforms the mesh by pulling up the five regions, which are filled by finer elements.
This figure shows the result of the deformation process. In this demonstration, the bubble mesh method is used for and coarse elements in other regions, so that it realizes the deformation process as users desire, while it avoids the increase of geometric data. This page also shows
Mechanical CAD/CAE
The final demonstration is the application of the square packing method to the mechanical CAD/CAE systems, that analyze safety, capacity, and strength of industrial products by using computational simulation methods such as FEM.
The above figures show an example of geometric model, which has seven melding spots, and a mesh generated by the square packing method. Finer elements are often required around the detailed features such as the abovementioned melding spots, since it often causes poor elements, but important numerical phenomena, around such features. The figures that the square packing method is suitable for various numerical simulations, since it flexibly controls the sizes of elements according to the geometric features or importance of numerical phenomena.
The above figures show that two meshes generated by the square packing method. These are generated by specifying different alignment directions in a same geometric model. The figures that the square packing method is suitable for various numerical simulations, since it flexibly controls the directionality of elements according to the requirements of numerical simulations.
This project develops a meshing software for large-scale geometric models of industrial products, which first automatically divides a given geometric model into some subregions, and then generate meshes inside each subregion. This architecture has some advantages in quality, computation time, parallelization, and so on. |
|
|
| Last modified 30 Sep 1999 |