Simplification of 3D Man-made Quadrilateral Mesh Model for Games
Date Issued
2012
Date
2012
Author(s)
Juan, Chih-Hsun
Abstract
Quadrilateral mesh has been a popular choice in model design due to the fact that it makes better simulation of deformation, and is more intuitive for texture mapping. Because the performance requirement in games, after producing a high-polygon model, we will need to produce another low-polygon model to make a normal map. This process is either done all manually or a user needs to define several rules on the model to get the low-polygon model. These methods not only take much time and sometimes is not so intuitive when a model is very complex. When making a high-polygon model the common way is to start from a basic model to do deformation, add elements, or make more detailed pattern. So these models have better structure than automatically converted or constructed quadrilateral models. There is also another situation when a user already construct the low-polygon model with desired structure and topology but need to adjust the amount of polygons. In these situations we need a quadrilateral based simplification method that preserves the original mesh structure and topology. Previous works usually focus on certain unstructured models that introduce some structure optimization method or do certain geometry enhancement in order to get a better local quality. These operations often destroy the special attributes of quadrilateral model and the results are not conform to requirements of common use. When the model size is very big these operations also slow down the simplification seriously. Another issue in games is level of detail, models use in games usually have separate part and too detailed features that cause poor result in level of detail operation. We use a voxel-based method to reconstruct models to solved these problems.
Subjects
Quadrilateral mesh
mesh simplification
Level of detail
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
index.html
Size
23.27 KB
Format
HTML
Checksum
(MD5):b39f771a1f80c6a9eb4e7f834cd9fbaf