ISSN 2079-3537      

 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                                                                                             
Scientific Visualization
Issue Year: 2012
Quarter: 4
Volume: 4
Number: 4
Pages: 2 - 16
Article Name: ANTI-ALIASING IN THE IMPLEMENTATION OF PRE-INTEGRATED RENDERING FOR VISUALIZATION OF THREE-DIMENSIONAL SCALAR FIELDS ON THE GPU
Authors: D. Bogolepov (Russia), I. Bugaev (Russia), A. Belokamenskaya (Russia), V. Turlapov (Russia)
Address: D. Bogolepov
denisbogol@gmail.com
Nizhny Novgorod State University after Lobachevsky, Nizhny Novgorod, Russia
 
I. Bugaev
bugaevio@gmail.com
Nizhny Novgorod State University after Lobachevsky, Nizhny Novgorod, Russia
 
A. Belokamenskaya
alexandra.belokamenskaya@gmail.com
Nizhny Novgorod State University after Lobachevsky, Nizhny Novgorod, Russia
 
V. Turlapov
vadim.turlapov@cs.vmk.unn.ru
Nizhny Novgorod State University after Lobachevsky, Nizhny Novgorod, Russia
Abstract: We consider the possible variations GPU-algorithm Volume Raycasting, used in scientific visualization of three-dimensional scalar fields, in the case of pre-integrated classification for visualized datasets. A method of generating multiple beams per pixel is offered, allowing at the same time solve the problem of full-screen anti-aliasing (FSAA) and further improve the visual quality, achieved through pre-integrated rendering. It is offered to use sub-pixel ray coherence for optimization of calculations (tested version generates 5 rays per pixel). In addition to the front-to-back order of tracing we study the back-to-front order, which can be useful if you want to trace the full way without an early ray termination. A test implementation of the methods is executed on the high-level shader language GLSL. A high quality and real-time performance of the techniques are demonstrated on the six examples of datasets, the most difficult of which was the Stanford Bunny (512x512x361, 16bit). It is used the GPU NVIDIA GeForce 560 1GB under operation system Ubuntu Linux 10.10 with the video driver NVIDIA Linux Graphics Driver 275.36. Visualization of datasets is made for both post-classification and pre-integrated classification modes, and both forward and reverse algorithms of color calculation are used. The reasons for the effectiveness of pre-integrated rendering are discussed.
Language: Russian