New Lossless Zip Algorithm Compresses JPEGs by 20%! Exclusive Interview Explains the Process

norman

The ubiquitous Zip file compression format has been a staple of PC users since it first made its debut as PKZIP in the early 90s. Back then, the size limitations of floppy disk media and the painfully low-bandwidth dial-up connections made file compression a complete necessity. The Zip format today, while still popular, has largely been eclipsed by RAR compression, which has offered slightly better compression at the cost of archiving speed. That’s why we were so surprised to hear that WinZip 12, which launched yesterday, boasted an unbelievable 25% compression ratio for JPEG images – without sacrificing image quality. Ever the skeptics, we put the new software to the test, and grilled WinZip’s VP of Development about how this new algorithm works.

We decided to test WinZip 12 on a large folder of high-resolution JPEG photos, each ranging from three to four MB, and totaling 1.13GB (or 1,217,671,445 bytes). This photo gallery (taken at this year’s PAX), was made up of 301 photos, each at 3888x2592 resolution, and taken straight from our Canon 40D camera.

Dragging the files into the WinZip program window, we were shown a pop-up with typical configuration settings, including the target destination and encryption options. The one notable difference was the presence of a “Change Compression” button. This opened a new window with two distinct compression options: “Legacy Compression” and “Best method for each file type”. We decided to give both a try.

With Legacy Compression, our newly compressed file ended up as 1.13GB, or 1,214,464,603 bytes. The total space savings displayed was 3.10Mb over 300 images, or about 1%. Looking at each individual file in the archive, we saw an average of 10KB reduction for each photo. Not particularly impressive, though at least the zipped file wasn’t larger than the original folder (which we’ve seen occur with poor compression algorithms).

“Best Method Compression” faired much better. Changing the compression setting to the new optimized algorithm churned out a 927MB file (972,699,916 bytes), or an approximate 20% size reduction. Color us surprised. We unpacked the new Zip file to examine the JPEGs and couldn’t find any difference between those and the originals – no new compression artifacts.

But here’s the catch (and there always is one). Compressing with the new algorithm took almost four times as long on a dual-core Pentium D machine than with Legacy Compression. Also, the new Zip file was only compatible with WinZip 12 – it wouldn’t open with Windows XP or Vista’s native archive browser, let alone on OS X’s built-in Zip app. That means that these files are useless unless you pay for WinZip 12. We contacted WinZip to find out exactly how the algorithm works, and how they expect to get the internet community adopt it.

Interview with WinZip’s Bill Richard (VP Development) and Shawn Cole (Director of Product Management)

WinZip 12 is able to significantly compress JPEGs. How is this possible?

Bill Richard :  Essentially, we reduce the file size of the JPEG file when we zip it by recompressing the JPEG file’s lossless compressed data using advanced compression methods that do a better job than the original compression methods defined by the JPEG specification.

A very important point is that WinZip 12 will compress JPEG images with no loss in image quality or data integrity.  This means that image files you put in a Zip file will be exactly the same, bit-for-bit when they are extracted.  This is a critical feature for an archive tool and WinZip takes lossless compression very seriously.  Full details of our new lossless JPEG compression method will be posted on our web site shortly.

What’s different about the compression algorithm this time around? Is it a new algorithm?

BR: Zip file compression has always used multiple compression methods. With WinZip 12.0, we added support for two new compression methods. The first is JPEG compression which has been developed by WinZip working in partnership with one of the original authors of the JPEG standard.  The second is the LZMA compression algorithm and is good for many types of files we tested including DOC, XLS, PPT, EPS, CDR, DWG and many more.  This was already defined in the Zip format appnote.txt file (the open specification for Zip).

These methods join others that we have introduced in past versions of WinZip like PPMd, bzip2, and WAVPACK to give WinZip a comprehensive tool kit that can now make smaller Zip files than ever before. To use these in WinZip 12, select ‘Best’ from the compression method options when adding files to a Zip file and WinZip will pick the optimal compression method for each file type you zip.

Does Winzip 12 only make improvements for JPEG compression? Are other image or document formats compressed better as well?

BR: WinZip 12 adds the LZMA compression method which provides significant compression improvements in both size savings and speed for many common file types.

Better compression seems to require a lot more computational power. How does Winzip approach the tradeoff between compression efficiency and heavy CPU usage? Which is more important?

BR : Our benchmark is the end user tolerance threshold. This is the point in which most people feel the additional compression savings is no longer worth the time it takes to achieve the extra savings.

How compatible are the new  Zip files with other compression software? How about older versions of Winzip?

BR: This is always a good question and one users need to think about when making Zip files.  The newer compression methods available in WinZip and other compression utilities are not backward compatible with older Zip tools or tools that have not kept up with the Zip file specification.  All versions of WinZip will inform you when a Zip file is compressed with a method that WinZip version is not familiar with. We publish the specifications for new compression methods we introduce so that other compression utilities can add support. In summary, recipients of Zip files that include state-of-the-art compression methods will require a compatible Zip utility to extract the files.

Are there plans to release a shareware version of Winzip 12, or a version that’ll let users decompress (but maybe not compress)?

Shawn Cole: WinZip is and always has been shareware but it has never been free beyond the evaluation period. Anybody can download WinZip and use it for free for 45 days while they decide if it is useful to them.  We are always evaluating the best ways to package the amazing features in WinZip but won’t be making any changes here in the foreseeable future.  So far, our users seem pretty happy with their investments in WinZip.

Does Corel [the owners of WinZip] have any plans to license its algorithms or any proprietary software?

BR: It has always been and remains WinZip’s philosophy to promote Zip as an open archive.  In that spirit, we have already published all our compression and encryption methods along with any enhancements we have made to the Zip format.   This ensures that the Zip file format and the community that support it can benefit from these improvements. This in turn, benefits users around the world.  An example of one benefit of this openness is that Zip files have become ubiquitous on the internet and because it is an open format, many virus scanners and search indexers have added the capability to “look” inside a Zip file.  Proprietary or little-used archive formats don’t have that advantage.

The 3rd party zip market has been pretty stagnant since Windows started incorporating native Zip compression and as internet bandwidth gets cheaper. How is Corel changing WinZip’s business model to adapt to the new computing environment?

SC: The market for robust, reliable Zip utilities like WinZip has not been affected as much as you might think.  We don’t see a stagnant Zip market and you can see that we have released several major versions over the years since Microsoft added built-in zipping to Windows.  Think of it like this, Windows’ Zip compression and decompression features are like WordPad.  It’s there, it’s free and sure you can write documents with WordPad but who chooses it as their primary word processor?

For people that need compatibility with all the archives they are likely to receive, or who need advanced compression methods or strong AES encryption, WinZip has clear advantages.  To compare compression performance, just try zipping up your documents with Windows and then do it with WinZip 12.0 using the ‘Best’ compression method and see the difference.  For even more power, check out our WinZip Pro Edition to see additional tools designed to automate many of the tasks that occur before or after zipping.

How much more compression do you think is possible with further R&D? Are we close to or have we reached the limits of file compression?

BR: One should never say never. There are a lot of really smart people in the compression industry continually working to improve existing ideas and develop new ideas. And, as computer performance continues to improve, so does the possibility to use new compression methods that only a few years ago would have exceeded the end user tolerance level. We do not feel the limits have been reached and we will continue to seek out ways to benefit the Zip format and the people that use it.

Around the web

by CPMStar (Sponsored) Free to play

Comments