Trying to describe byte values different then the common binary number system I wrote Mtcc.
Mtcc take by default 128 bytes and place them in a 16 x 16 matrix like a bitmap. Not used matrix rows are excluded and same with the row/column values behind the last value in a row, this last one only if there is any gain. What is left from the matrix is stored with the shortest 128 coordinates.
It run under Windows as command line mode 32 and 64-bit and need Framework 2.0 or higher.
Command line syntax:
Compress: mtcc c input output
Decompress: mtcc d input output
Memory usage is very low, gain is also very low, probably the worst compression algorithm ever made.
There is an undocumented parameter when you put a space and number behind the output parameter the amount of matrix coordinates can be controlled, sometimes 256 can give better results then the default 128.
There is no error handling and because it's more an encoder then compressor, it's probably possible to decompress any random input file without an error.
A future idea is to split the output optional in two parts, a matrix and coordinate part, where one can be transfered secure and the other not secure to reconstruct original at destination side with both parts.
Download
http://www.metacompressor.com/download/mtcc.zip