МОДИФІКАЦІЯ МЕТОДУ А-БУФЕРА ДЛЯ ЗГЛАДЖУВАННЯ ТРИВИМІРНИХ ЗОБРАЖЕНЬ

Abstract

According to the A-buffer method, a dynamic list of polygon fragments that cover a pixel is used for antialiasing of three-dimensional images. The A-buffer method uses less memory than the additional sampling method, but the need for dynamic memory allocation complicates its hardware implementation in computer graphics systems. Using a fixed number of fragments per pixel of the image simplifies the hardware implementation of the A-buffer method, but memory is used inefficiently because the number of fragments to calculate the pixel color intensity can differ significantly for different pixels. It was proposed to use a combined static-dynamic principle of memory organization, which is to introduce a common fragment buffer. New modifications of the A-buffer method have been developed, which use: - a combined static-dynamic approach to memory organization, which consists in the introduction of a common buffer for storing redundant fragments. The proposed modification has a simpler hardware implementation compared to dynamic memory allocation and requires less memory to store fragments of polygons than a fixed allocation; - a new approach to reducing the error of color intensity calculations, which occurs due to the rejection of fragments of polygons that did not have enough space in the fragment buffer. The essence of the approach is that the two most distant fragments from the observer are combined into one. The block diagram of the rasterization subsystem of three-dimensional graphic objects is developed, in which the modified A-buffer method is used. A computer program for modeling and testing smoothing methods based on the Abufer method has been developed. The simulation showed that the use of the proposed principle of memory allocation allows to reduce by 1.3-3.1 times the amount of memory required for smoothing three-dimensional images using the A-buffer method, compared with fixed memory allocation. The simulation results showed that the proposed solution, based on the combination of redundant fragments, provides a smaller value of the normalized standard error compared to the approach in which excess fragments are discarded. When using the new method to process polygon fragments, smoothing even complex scenes with the A-buffer method can be performed using only three fragments per pixel of the image, which reduces the amount of memory required to store fragments of polygons.