Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Generation of Filtered Noise for Computer Graphics

IP.com Disclosure Number: IPCOM000035144D
Original Publication Date: 1989-Jun-01
Included in the Prior Art Database: 2005-Jan-28
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Todd, SJP: AUTHOR

Abstract

In computer graphics, realistic solid models require texturing which is implemented by generating spatially filtered three-dimensional (3D) noise. In the technique disclosed, one-dimensional (1D) noise tables are first formed. Then plane waves in three space based on these tables are generated. The wave fronts are perturbed using further 1D tables, and the perturbed waves are added to confuse directionality. The use of different 1D tables in the generator of 3D noise prevents visual symmetries in the textures produced.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 1

Generation of Filtered Noise for Computer Graphics

In computer graphics, realistic solid models require texturing which is implemented by generating spatially filtered three-dimensional (3D) noise. In the technique disclosed, one-dimensional (1D) noise tables are first formed. Then plane waves in three space based on these tables are generated. The wave fronts are perturbed using further 1D tables, and the perturbed waves are added to confuse directionality. The use of different 1D tables in the generator of 3D noise prevents visual symmetries in the textures produced.

A known texturing technique requires a spatially filtered 3D noise function. This is most conveniently implemented by a sampling process that returns the function value for a specified point in space. A typical picture generation requires of the order of one million such samples. This article discusses an implementation of such a sampling function based on 1D lookup tables. Each sample requires about ten references to the tables, and the tables require about 4096 bytes.

Previous techniques include: Very large 3D tables; Many lookups into a smaller 3D table generated, for example, by Fourier analysis. These lookups require bicubic interpolation to prevent poor results; Random midpoint displacement, which gives poor results; Very many successive random additions

The principles of the disclosed technique are: generate 1D noise tables using, for example, Fourier analysis, generate plane waves in three space based on these tables perturb the wave fronts, using further 1D tables add the perturbed waves and optionally,...