Compression - Preprocessing techniques in OverZip and Precomp

Author

Preprocessing techniques in OverZip and Precomp

ill Bill

2006-09-21, 6:55 pm

The straightforward preprocessing strategy of the Precomp program
that decompresses compressed ZIP streams in
PDF,SWF,ZIP/JAR, and PNG files, and then
uses the decompressed files instead of the compressed ones
as input to better compressor (like uharc)
may be compared with less straightforward strategy
of OverZip.

The OverZip program transforms *compressed* streams
in archive files (without decompression of the streams)
so that the files can achieve better compression.

Obviously, the first approach gives smaller output file.
But the second is faster and more universal -
you don't have to know the exact algorithm
used in the compressed streams, you can
handle JPEG streams, etc.

So, it would be useful to have some hybrid
of OverZip and Precomp. The authors are requested to cooperate.

schnaader@gmx.de

2006-09-22, 7:55 am

ill Bill wrote:

> The OverZip program transforms *compressed* streams
> in archive files (without decompression of the streams)
> so that the files can achieve better compression.
>
> So, it would be useful to have some hybrid
> of OverZip and Precomp. The authors are requested to cooperate.

There were some people that suggested similar methods for Precomp, too.
It is on my to-do list, and will appear in one of the newer versions.

There will be a "secure" mode, which will check streams more carefully
and only process known streams like PDF, PNG, ZIP with headers and
footers that can be detected for sure.

This will often (not always, see discussion about ohs.doc for this)
lead to worse compression in the actual lossless mode (because "real"
zLib-streams with 2 byte headers and unimplemented header types can't
be processed), but will be useful for a "lossy" mode.

This lossy mode will do what you described, it will recompress streams
with maximum compression without the possibility to get the original
file back.

Greetings,
Christian "schnaader" Schneider
---
http://schnaader.info
Damn kids. They're all alike.

ill Bill

2006-09-24, 6:55 pm

schnaader@gmx.de wrote:
> This lossy mode will do what you described, it will recompress streams
> with maximum compression without the possibility to get the original
> file back.

OverZip claims lossless compression, even for archives
with mp3 and jpeg files. This program recompresses
RAR archives, beating RAR's maximum compression mode,
and gets the archives with the original files back.
This is true for all my tests.

schnaader@gmx.de

2006-09-25, 6:55 pm

> OverZip claims lossless compression, even for archives
> with mp3 and jpeg files. This program recompresses
> RAR archives, beating RAR's maximum compression mode,
> and gets the archives with the original files back.
> This is true for all my tests.

I just had a look at OverZip. It uses external programs like UHArc or
WinZip to decompress, and you need to specify where these programs are.
I didn't even get it to shrink a JPG file.

But far the worst is it uses Precomp! OverZip is a 30-day-trial
version. This is commercial use. Precomp MUST NOT be used in commercial
programs, but is only "free for non-commercial use", as stated clearly
in the copyright message.

Looking at your e-mail adress, it seems that you are one of the
authors, Serge Preobraz. If you're not, correct me. But if you are, I
advise you to take Precomp out of OverZip as long as you don't have my
permission.

Thank you,
Christian Schneider
---
http://schnaader.info
Damn kids. They're all alike.

Sponsored Links