Wavelet is a program that demonstrates “lossy” image compression. Graphics formats such as GIF files compress the image while retaining all of the picture’s information. GIF files are often about one third of the original size. Other formats, such as JPEG, achieve better compression through methods that throw away part of the data.
In real pictures, there is a lot of small detail that the human eye isn’t quick to notice. JPEG represents a picture as a sum of cosine waves of various amplitudes and frequencies. High frequency data, that means rapidly changing features, are eliminated while preserving the overall picture. This works well for continuous tone pictures like photographs but not for cartoons or other images with sharp contrasts in colors and shades.
Wavelet represents the picture as a sum of wavelets. Cosine waves are shaped like /\/\/\/\/\/\. Wavelets are shaped like ---/\/\---. They work much better at preserving local contrasts. Thus, 80% of the picture’s information can be eliminated and still not be noticed by the casual observer. To learn more about wavelets, read “Numerical Recipes in C. 2d Edition”.
Wavelet is not meant to create a new graphics standard. Rather, it lets the observer decide how much information to eliminate via a scroll bar, and then rapidly see what the picture will look like compared to the original. For better performance, the resulting image is displayed in gray scale. Thus, I recommend that you set your monitor to 256 gray scale.
Wavelet shows 4 windows. The first holds the source image. The second shows what the picture represented by wavelets looks like. The third is the resulting gray scale image. The fourth window controls the amount of information thrown away. You can print out any of the 3 picture windows by selecting it and then choosing Print from the File menu.
In the Wavelet ƒ folder, you’ll find Wavelet, Wavelet FPU, this README file, and a sample PICT. Wavelet requires System 7 with 32 bit Color Quickdraw on a 68020 or better processor and also 2 Megs of available RAM. Sorry, no native PowerPC version yet. Wavelet FPU works more than twice as fast on machines with a math coprocessor. You can only read in PICT files. Graphics Converter available from info-mac can convert any graphics file to a PICT file.
Wavelet is my first Macintosh program. I hope to produce more. Feedback is very welcome! If you really like it, please send me $5.00 so I can purchase a PowerPC compiler to make this a much faster, native version.
