Data-Compression Algorithms

[Lior]

Hi all,
I wanted to ask you if you know some forums or known websites for lossless data-compression algorithms.
Second, I wanted to ask you:
If I invent a new compression technique which can compress even ZIP and MP3 files dramatically, could I make out of this useful algorithm any money? or openning a start-up for this compression algorithm?

what do u think?
10x.

[HarryW]

Yes, you could make a sh**load. The only reason dial-up modems can work at 56k and not 2400 baud is because of compression. More compression => more bandwidth so long as you can do the maths fast enough.

If you patented the technique (which I think you can do but I'm not sure) you could get royalties from anything that uses it.

Of course, you have to find one first ;) Good luck, I think you'll need it.

[Guv]

I hate to rain on your parade, but I doubt that you could make money on a compression algorithm.

The methods currently in use are damned efficient. It is probably impossible to squeeze out more than another 2 percent. With only a small increase in the compression, it would be impossible to out market the companies already selling the existing software. The existing companies are firmly entrenched. A lot of the users send files to each other. For example: I would not use new software until a lot of my friends were also using it.

You might be able to find a special niche in this market. Instead of going for a general purpose compression algorithm, perhaps you can discover a way to dramatically compress some particular type of file.

Suppose you could compress Microsoft Word or Corel WordPerfect files to 25% of what Pkzip does. Microsoft or Corel might pay you to use it as an add-on feature to their software. If you got a dime for every copy of Word or WordPerfect sold, you should be happy.

There might be some widely used graphic format that you could compress better.

When compressing text files, the existing software might be biased toward compressing English text. Perhaps you might be able to devise a method that is much better for some other language, and find a niche market.

[kedaman]

I haven't heard a word about the VQF format since 2 or 3 years ago i first heard about it, it's a format that compresses audio a lot better than mp3's and even preserving better quality, though at the cost of more cpu usage. Even that revolutionary compression didn't become popular because that the effeciency of mp3's were enough and so widely known that nobody did care.

[Lior]

You are right about the VQF format, but the only difference is that my new compression algorithms compresses even ZIP files. you see, it's the beginning of a revolution. most of the net uses ZIP for compressing files, with my new algorithm, it can become a standard!
VQF compresses only audio, so even if he does it better than MP3, it cant compete the popularity of it and as you said, the MP3 compression is enough in most cases.
but when I am talking about lossless-compression-algorithms that means that it can compress any sort of imformation: images, audio, video, text, anything..
this means that it can compress even its own compressed files!!! I know it sounds impossible, but here comes all the idea of it.

What do you say, guys?

[kedaman]

Happy coding!

Lior,
If your compression program compresses a file, you should not be able to then again compress that file with the same program. The reason for this is, you already compressed it as far as you could the first time. If there was any way to compress it further, than your compression program is inefficient and missed some things. There should be no reason to keep running a compression program on the same file, once through should compress it as far as possible.

[Lior]

My algo supports "deepness feature"
which lets the user choose how strong to compress it.

[T-Bone]

Lior,

With compression you will find that the way most compression algorithm's work is by looking for repetetive paterns within the raw data. Which means that once it is compressed the repetitive pattern is no longer there.

In short the more something is compressed, the more RANDOM it becomes. And of course the more random it becomes the harder it is to try and rind repetitve patterns!

[SteveCRM]

Good luck Lior!!!!!!! Keep us posted on your progress.

P.S. how would you go about doing this?

[Lior]

Well...
the algorithm is very long and complicated, based on high math involving bits-calculating mixed with probabilities calculations.
The algorithm is designed for lossless data.

[SteveCRM]

I mean how do you get the files data and stuff like that.