Re: Chaos Theory - Yes!

St. James (jimmy@nospam.socs.uts.edu.au)
Thu, 18 Jul 1996 11:56:51 +1000 (EST)

At 11:35 am 07/18/96 +1000, you wrote:
>> They have created a new image format which uses the method of fractal
>> compression to make image files so small, you no longer need to wait yonks
>> for graphics to download from the internet. It compresses image files by
>> a factor of 50:1
>
>I don't have a web browser on me at the moment, so I can't look at the web
>site, but there are a couple of things you should think about. Last I
>heard, fractal compression was very good at compressing scenes that had
>recursive patterns in them, i.e. fractals. The thing about fractals is
>they don't take much information to transfer them anyway. On images (or
>data) that doesn't have this property - including most data files your
>would which to move and most pictures, fractal compression works no
>better, and often worse then normal techniques.

That 50k compressed file is 50k of sets of affine transformations, also
probably compressed (lossless). They can do it automagically nowadays, but it
is very asymmetric. Which is okay really, they take a while squishing and we
take next to no time expanding. And you can zoom in to any level you wish.
The notion
of the pixel is irrelevant, but artifacts due to the chunks they build the
encoding
around show up at high magnification. The artifacts are kind of funky in
their own
way and no doubt we will start to see them used in design within the year.

>Sorry, this wouldn't work (IMHO). There are theoretical limits to how far
>you can push lossless compression (which is what you want for data, not
>images). An encyclopedia isn't fractal in nature (self similar at
>different scales), hence would not compress using this technique.

That's right kids. Don't use this on text. IMO you would only use it on
images that
have greater than screen resolution.

Then there is wavelet compression...

jimmy