(almost) unlimited compression without data loss: where am I wrong ?

3 posts / 0 new
Last post
take two files with the same size, the first being a text file while the other one is data file;
compress them using winzip;
the first file will be severely compressed, while the other file will eventually (with luck) gain some  space;
then let us take  a third file, of size higher than that of the second file, but composed of text;
this third file will be of smaller size than the second one when it will be compressed.

here's what we have got:
 two files with same size prior to compression will have different sizes after compress.
one file of greater size than another will bear better compression ratio after compression.

so I wondered myself saying:

the algorithm lempel-ziv huffman (i.e. winzip) must work in a given way... that I don't know how but,
with taking a binary file (hard to compress) and giving it some fresh (i.e. adding bytes so as to encourage compression the same way as with a bigger text file) would theorically permit to win some space (be it only some bytes in a one MB file) so as to let us recompress a second time... a third time, then a thousand times which would lead to a drastic diminution in file size.

so, where am I wrong here please??? 
Why don't you try it and find out.
http://guild.medialoungeca.com/index.php?action=forum The Guild I'm apart of. We're in WOW, STO, Rift and soon Star Wars feel free to register and hang out. http://sparkster11.deviantart.com/ my deviantart Wheelman of the House of Trolls, "I love it when you watch" Carrier of Section 2, 3 and 6 cargo. Resident Driver Stud God of Transportation and Lust.
You obviously do not know what you are talking about.  Instead of thinking about it, study it.  Each file is handled in a different way.  The compression process works along a specific system, but how each type of file is put together ends up with how much space is conserved.

Now, you are intending to alter the file to enable further compression.  Okay, but how do you write code to recognize the alterations?  There is then the one-byte trick.  Sit down and figure out how to uncompress a single byte into a full file.  It can be intellectually fun, and should teach you a lot of the fundamental process.  You might come up with something better, but you will probably just realize why certain programs continue to survive.