I've been thinking about this for a while (10 minutes).

Compressing a file has a limit, so say if i use the best compression method on winrar on a file i'll maybe get it down to 50% of the size right.

Here's how I've been visioning a new type of compression:

Say you have a file made up of all its little bits of data. A new compression language could be written that basically accounts for most strings of code that make up files. So for example in a jpeg image you can write a small code that basically says the colour of the pixel and its location. Then just write a language that can express all of that data in maybe 3 characters (colour, X, Y).

If the compression tool takes all of the pixels and turns it into these 3 character strings and saves it into another file; once you reopen the file it can "decompress" the code and rebuild the file with all of the colour and position info.

Its very hard to explain, but i'm sure it will click in someone's mind and they can help me explain it.

Think of short hand writing, in computer form. A squiggle with a dot here means a "hello how are you". Same principle, really.

Thoughts?

(no drugs taken or used in construction of ideas)

And by the way - if this is actually how compression works then slap me, if not and it in theory works - Idea property of Jack McIntyre 29/09/2008.

1 character = 8 bits (in ASCII)
3 characters per pixel = 24 bits per pixel

1024 pixels * 768 pixels = 786432 pixels

786432 pixels * 24 bits per pixel = 18874368 bits = 2359296 bytes = 2.25 MB

Tell me again how this is an efficient compression algorithm?

Quote from Forbin :

1 character = 8 bits (in ASCII)
3 characters per pixel = 24 bits per pixel

1024 pixels * 768 pixels = 786432 pixels

786432 pixels * 24 bits per pixel = 18874368 bits = 2359296 bytes = 2.25 MB

Tell me again how this is an efficient compression algorithm?

The 3 character idea was just a bit of an "upgrade" from my original idea. I first thought that 1 character could be used for everything in the picture, although that would mean an infinite language would need to be created for it to work.

I think one thing I havn't grasped about the idea is that a bit of data has a virtual "mass" and can't just be reduced.

I think I get what you mean. So certain common strings of code aren't contained in the actual zip, rather in the compressor/extractor application?

So if say the string 01010001010100101010100001010101010101010 (to pick something at random) happens in a lot of zips to occur quite often, then this is just stored in the zip as a scrap of code meaning 'use string x stored in the zip application'

It would mean a big install for the zip application and would probably mean theings were quite slow to zip-unzip, but I can't see how it wouldn't work to some extent. It'd depend a lot on whether certain strings occur long enough in an 'average' file.

That is pretty much how an image works anyway... sort of.

Lets take for example .gifs (it was ~6 years ago when I learnt this and I can only remember how .gifs work.).

You have a 100x100 image of vertical stripes. The code in easy to read for that image would be this:

Redx100, Blackx100, ... , Redx100, Blackx100.

Of course if you rotate that image 90* it'll be

Red, Red, ... , Red, Red
Black, Black, ... , Black, Black
Red, Red, ... , Red, Red
Black, Black, ... , Black, Black
etc

And that just eats space. Naturally not vast amounts, but there you go, a little education.


1,022 bytes

1,103 bytes

A dutch guy had a very good compression idea that saved a lot more then the regular compression methods. He got investors and started working on it, but died along the way before it was launched.

Someone tried a few years later to discover what exactly this new compression was (from old-news and interviews with the deadguy). He found out the new part was he combined compressed files which did a lot for the filesize (it was more then that, but that was the biggest factor in increasing the compression rate) because then there is a lot more to compress. Only problem is you can't use that online. You can't send one file because you need the whole archive.

Quote from marzman :

Only problem is you can't use that online. You can't send one file because you need the whole archive.

Isn't that basically how .tar.gz files work? The files are placed inside a .tar container, and then the one big file is compressed?

Quote from wheel4hummer :

Isn't that basically how .tar.gz files work? The files are placed inside a .tar container, and then the one big file is compressed?

it was more then that, but nobody know exactly what

http://en.wikipedia.org/wiki/Jan_Sloot

Like dictionary compression (can't remember the exact term).

The problem with data structures and algorithms etc is most of the good ones have already been found. Don't quote me on that in 100 years time

This certainly qualifies as a lerts theory in that you didn't do any research before posting your idea, but the standard of your explanation is too high.

There is no mention of aggressive vampires, either.

This is a bad copy.

It makes too much sense. Run it through a translator to something like Russian and then back to English. Then it'll be heading the right direction.

Quote :

(no drugs taken or used in construction of ideas)

This is where you went wrong Jack.

To qualify for lertsness you must also mention that this idea can save the world, and extend human capabilities 100-fold, but that it is kept secret by some sinister force.

Quote from Lerts Logic Version :

I it sees that it thinks about this during some time (10 minutes). The compression of a file has a limit, so it says if I use the best method of the compression in winrar in a file i' perhaps it obtains it down until 50% of the right of the size.

Here's how I a new type of compression see it envisioning: It says that you have a composite file of all small small pieces of data. A new language of the compression could be written that basically explains the majority of the sequences of the code which they compose archives.

So for example in an image of the JPEG you can write a small code that basically says the color of the pixel and its location. Then as soon as a language writes that can perhaps express all that the data in 3 characters (color, X, Y).

If the tool of the compression takes all the pixels and it gives return them in these 3 sets of characters it saves and them in another file; once you are again the file can "decompress" the code and reconstructs the file yet the color and places the Info.

Very hard his to explain because of lizardmen overlords, but i' m sure that will make sense in someone the mind and they can reach the next level of enlightment. It thinks about the short writing of the hand, about listing of computer. Squiggle with a point means "here, hello how he is you". The same principle, really. Thoughts?

Here, that should do it.

LOL the last paragraph is gold.

I dunno about the "technical" side, but there is some pretty high compression methods available.

I remember extracting some compressed files, they were maybe 5 Mb on size but when uncompressed they had a full .ISO file in, few gigs on size.
If I remember correctly they even had the .rar extension and I extracted them with WinRAR.

Wow, that can only mean that the .iso file was mostly empty though. I can also make a several gig text file just containing "0" over and over and when you compress it it will be < 1KB, simply because you can write it as "1000000000 times the character 0" (aka run length encoding, one of the simplest compression methods available).

Quote from Becky Rose :

This is where you went wrong Jack.

I knew i'd missed a step out!

Quote from Becky Rose :

Here, that should do it.

I'm glad i posted a load of nonsense and looked silly, because that reply made me howl with laughter.

By the way this is a 100% genuine idea that popped into my mind - i'm not taking the mick out of lerts its just that after i tried to explain it to people on MSN it sounded wacky enough.

All we need now is a rar that stands up by itself.

Quote from Crashgate3 :

I think I get what you mean. So certain common strings of code aren't contained in the actual zip, rather in the compressor/extractor application?

pushing tons of information into sidebands doesnt compress it in any way

Quote from mcintyrej :

All we need now is a rar that stands up by itself.

You mean a self-extracting .RAR file which doesn't exist? It's already built into WinRAR ("Build SFX" iirc).

For compressing bigger files, I like to use 7-Zip (admittedly set to create a self-extracting file because no-one else has it installed), as I find it's the best - LFS is also compressed in this way

Quote from Jakg :

You mean a self-extracting .RAR file which doesn't exist? It's already built into WinRAR ("Build SFX" iirc).

For compressing bigger files, I like to use 7-Zip (admittedly set to create a self-extracting file because no-one else has it installed), as I find it's the best - LFS is also compressed in this way

No, I mean install a gyro into the rar file so it stands up like the brolly or pen (cant remeber) that lerts designed.