You probably don't believe me, but if you're at all interested in my new "CAR" compression alogrithm, check this out:
The strange thing is, it works better on compressed files!
Zipping an MP3 file gives you 99% of original, but check this out!
**** TESTS ON UNCOMPRESSED FILES ****
TXT File Example
TXT File: 1,318,671
Savings: 1,308,940
CAR File: 9,731
Percent: 0.7%
WAV File Example
WAV File: 8,362,354
Savings: 8,323,477
CAR File: 38,877
Percent: 0.5%
EXE File Example
EXE File: 216,064
Savings: 213,336
CAR File: 2,728
Percent: 1.3%
**** TESTS ON ALREADY-COMPRESSED FILES ****
MP3 File Example
MP3 File: 4,961,773
Savings: 4,945,669
CAR File: 16,104
Percent: 0.3%
MPG File Example
MPG File: 5,976,068
Savings: 5,946,909
CAR File: 29,159
Percent: 0.5%
If you didn't see it first time, I compressed an MP3 file from 5 meg to 16kb.
What CAR actually does is obviously a complete secret, but I'm really really excited about it! I've been thinking of how to do it for years - but now, yay! (I figured it out playing around in QB, of all things!).
What I want to know is basically are there any sites that are relatively easy to understand that tell you how to do:
* Huffman Compression
* LZW Compression
* "Textbook" RLE Compression (I only know PCX's RLE)
I know that you use binary trees and nodes and so on but I have no idea for a software implementation!
Anyways you probably don't believe me, but I just wanna try to make the compression better.
Thanks from a very very excited
Kieren Johnstone
Name:
Anonymous2006-04-26 17:18
This is how it's done:
First, you must realize that zero is nothing. Nothing is useless, so I strip all zeros out of the binary data.
After that, all that is left is a very long string of 1's. This string is compacted to a single 'one', along with a number indicating it's length. Then I add my signature and some other overhead, including a bit of random data to give a unique hash, thus ending up at between 2-16kb of compressed data.
I make the filename store the contents of the file
It takes up 0 bytes.
Name:
Anonymous2006-04-27 18:50
I take the filename and store it in an NTFS alternate data stream, then truncate the file at 0 bytes. It will appear that the file is zero bytes, yet the data can be retrieved if you know the key to access the data stream.
I wrote a program that looks at the amount of free disk space, multiplies it by 2, and then outputs it.
Double disk space!!! No effort FTW!
Using similar technology, I wrote a program that multiplies ANYONE's reported bandwidth by a factor of 2000. It's like I have a fucking T3 up in here W00T!
Name:
Anonymous2006-04-28 0:31
T3's are really slow these days, aren't they?
I have a friend who swears by his 56kpbs OC192 though. That must be awesome.
Name:
Anonymous2006-04-28 1:30
Using proxies that doesn't make any diference
Name:
Anonymous2006-04-28 3:11
You're all idiots. The very fact that you saved the file means that you used the directory tree, thus, your file isn't 0 bytes because your file also takes in account, oh oh, the file name itself. I got a better compression algorithm, DTF. Delete the file, 100% savings, and the best you guys could do was 99%? psh.
Name:
Anonymous2006-04-28 3:41
I've got better, move the file to your googlefs, box.net, or openomy drive, HAHA! Eat that!