Alas poor JPEG…

You might have expected Microsoft’s recent announcement of its new continuous-tone bitmap format, Windows Media Photo (WMPhoto), to be greeted with universal acclaim. After all, JPEG is such a central workhorse that a new, improved “JPEG-killer” must surely be good news. In fact, my initial reaction, along with that of the majority of professional designers, was shock, horror and anger.

To understand why, it’s necessary to look back at the history of bitmap formats to appreciate what a major role JPEG plays. Back in the 1990s, when PCs first became capable of handling 24-bit colour and continuous-tone photographs, a serious problem arose: in photographic images, the colour of neighbouring pixels varies almost randomly, which gave the then prevalent compression algorithms RLE (run length encoding) and LZW (as in ZIP files) almost nothing to work on. With the spread of desktop scanners, digitised photographs became common currency among designers, leading to file size becoming a huge issue (don’t forget storage and comms bandwidth were both limited compared to today).

A white knight rode to the rescue in the form of the Joint Photographic Experts Group (JPEG), which invented the revolutionary JPEG compression scheme. JPEG converts 24-bit RGB data into a different colour space called YCbCr (Y handles luminance and the Cb and Cr channels colour), then downsamples the Cb and Cr components, usually by a factor of two horizontally and vertically, since the human eye is more attuned to variation in brightness than in colour. It then chops up each channel into 8 x 8 pixel tiles and applies a Discrete Cosine Transform (DCT) to convert these into a “frequency space”. The eye isn’t particularly good at distinguishing the exact strength of high-frequency brightness variations, so this permits quantising of the results for a more or less aggressive trade-off of file size against visual quality. The JPEG approach is “lossy”, which means the compression isn’t fully reversible to the original pixel values, but it was capable of such high compression ratios that it became a huge success. It typically achieves between 10:1 and 20:1 compression without any visible loss of quality, 50:1 with moderately visible defects, and right up to 100:1 if all you need is a low-quality image.

JPEG had another major advantage in that it was and is free to use (not that you’d have known at the beginning when early JPEG implementations were proprietary and expensive). Eventually, Thomas Hamilton developed the standardised JFIF file format (JPEG File Interchange Format, now almost universally referred to simply as JPEG) and placed it in the public domain. Just as important was the work of Tom Lane and the Independent JPEG Group (IJG), which created an open-source software implementation that ensured that JPEG was taken up by bitmap-editor vendors and, most crucially, by Netscape in its groundbreaking Mosaic web browser. By the time digital cameras became commonplace in the mid-1990s, JPEG (or strictly JFIF) was firmly ensconced as the ideal royalty- and licence-free, cross-platform, cross-application file format for handling photographic images, both locally on the desktop and on the web. In 2001, Microsoft finally recognised the central role JPEG played by adding native support to the new GDI+ on which Windows XP was built.

JPEG does a brilliant job and has established a hard-won universality, so what does WMPhoto have to offer? In Microsoft’s announcement, two features were promoted heavily: its lossless compression capability and its generally superior lossy compression. Clearly, these are both real benefits, and lossless compression is compulsory in certain circumstances – for example, in medical imaging. It’s also welcome when you need reassuring that you have the best-quality image available, and it avoids the problem of degrading quality caused by repeatedly opening, editing and closing a JPEG (which is why you should always save to a lossless format when working with JPEGs). Clearly, if JPEG’s great strength is its lossy compression, any format that can improve on it has to be impressive, especially when the type of degradation it introduces is less noticeable than JPEG’s strikingly blocky artefacts. If WMPhoto delivers on the claims made for it – halving the equivalent JPEG file size while maintaining quality – it would certainly be extraordinary.