|
Dear DDJ,
"The software field is one of constant reinvention," write Richard Stallman and Simson Garfinkle in their passionate "Against Software Patents" (Communications of the ACM, January 1992). No doubt this massive obscure heritage leads one to suspect he has stumbled onto something new more often than is the case.
The method described by Dean Clark in "A 2-D DDA Algorithm for Fast Image Scaling" (DDJ, April 1997) falls into this category. It describes a method that was well known in the circles in which I traveled almost a decade ago, when I wrote ("Full-color imaging on amplitude-quantized color mosaic displays," SPIE Vol. 1075 "Digital Image Processing Applications" (1989) p. 199ff)
Scholars may notice that single-branched error propagation is isomorphic with the rendering of straight lines using Bresenham's algorithm and modern methods of rescaling binary raster images....
As Dean's well-written article points out, nonbinary images can obviously exploit the method for the sake of speed at the sacrifice of image quality, relative to interpolation methods.
R.I. Feigenblatt
docdtv@peachlink.com
Dean responds: Thank you Mr. Feigenblatt. Your quote of Stallman and Garfinkle is almost painfully true. While I did "invent" the method presented independently (though more like eight years ago, not ten), I'm certainly not surprised that someone else thought of it first. Even so, the only other place I've seen it presented is in one of the Graphics Gems books -- and not until after I'd already submitted the article to DDJ.
......