Fast CSG voxelization by frame buffer pixel mapping

Abstract

This paper describes a fast algorithm for the volume conversion and rendering of CSG models constructed from both geometric and volumetric primitives. Using 3D texture mapping and frame buffer pixel operations, the algorithm can interactively generate a binary volume of the CSG model. The result can be used for volume rendering and other applications. Boolean operations are implicitly computed by a Point-Classification Map, and implemented by a hardware assisted frame buffer pixel map. The algorithm can be applied to any regions of interest of the model, thus provides a multi-resolution rendering solution through dynamic voxelization of the viewing regions. Since no pre-processing is required for any change of the CSG tree, it can be used as an effective rendering tool in a volumetric CSG modeling environment.

Graphics Based Path Planning for Multi-Axis Machine Tools

“…In this research, a voxel model is used to represent a part and blank. The part is voxelized directly from the polyhedral model and the resulting surface of the part is defined by the exterior voxels; but, this is not a requirement as there exists methods to voxelize part models directly from CSG representations [31].…”

Fault-tolerant defect prediction in high-precision foundry

“…In recent years, the evolution of graphics hardware has given rise to faster and more accurate voxelization algorithms [9], [8], [10], [11]. Thanks also to the emergence of programmable shaders (sets of code instructions executed directly on the graphic processing units), much of the computational burden has shifted to specialized hardware, so it is now possible to conduct voxelization and simulation operations in real-time [12], [13].…”

Real-time voxelization for complex polygonal models

“…These slices constitute the final volume. The algorithm was later extended to a wide range of 3D objects [10] [11] [12] and applications including 3D spatial analysis [1] and collision detection [2]. However, its algorithmic complexity is proportional to the slice number of the resultant volume.…”

“…This concept was first introduced by Arie Kaufman [21] [19]. Since then its applications in diverse fields have been broadly explored, including volume modelling [32], virtual medicine [22], haptic rendering [23], visualization of geometric model [33], CSG modelling [12] [7], collision detection [15] [2] [13] and 3D spatial analysis [1] etc.…”

“…The latter can be further divided into filtered [31] [28], multivalued voxelization [10] [16], object identification voxelization [17] and distance transform [27] [30]. As the form of processed primitives is concerned, there are methods for line [5], triangle [6], polygon [21] [31] [18], CSG [12] [7], parametric surface [19] [28] and implicit surface [29] voxelization. Concerning the structure of the volume, the result can be stored in the form of regular grid [32], general 2D lattices [27] or distance transform [34].…”