The venerable JPEG format for multimedia images has been an industry standard for some 20 years. Now, the search is on for sharper, more versatile image-coding technology, and Microsoft thinks it has developed it.

By Laurie Rowell

How does Microsoft, a company eyed warily by competitors and open-source advocates alike, deliver a new information technology standard to the industry free of charge and without any strings attached? Especially when that standard promises to change how we see the world, via the Internet and other multimedia sources, for years to come?

Robert Rossi, principal program manager lead for Microsoft’s Core Media Processing Group, took a crafty approach to this challenge. He stood in front of the Joint Photographic Experts Group (JPEG), the group that set the JPEG standard, at an American National Standards Institute (ANSI) meeting in February, and said, in essence, “Here is your JPEG1 successor.”

The new format, originally called HD-Photo and renamed JPEG XR (extended range) was developed at Microsoft entirely. “Internally, a lot of people would question why Microsoft gave the technology away free of charge.” Rossi said. “The truth is that an image-coding technology doesn’t survive if it has a cost associated with it, because it needs diffusion.”

“We felt it was important from a political standpoint to go to that group (JPEG) first, even if they turned us down,” said Rossi. He said a straightforward approach to the computing community was necessary to ensure that it couldn’t be claimed later that Microsoft had developed a technology that replaced JPEG without first submitting it to the JPEG organization.

Yet there were serious questions about whether the group would accept Microsoft’s new format, which would be competing with that organization’s own JPEG 2000 standard, originally slated to be the replacement for JPEG.

JPEG 2000, however, has run into difficulties. “They did not realistically look at the limitations of what could be put in an embedded processor for a digital still camera running at 100-200 MHz, and at what cost,” Rossi said.

He believed there was never any question about where the superior technology lay. “One of the advantages that we’ve had all along has been outstanding engineering, including the excellence in tradeoff decisions that make for a successful design,” he said.

The team has made presentations at three JPEG conferences this year. It will show the technology next to camera makers in Kobe, Japan, but that won’t be the end of the process. “It takes a year and change to pull this off,” Rossi said.

Most of us will have to wait until new chips are developed and begin appearing in digital cameras for consumers in a couple of years, but photographers who use high-end digital SLR cameras and capture images in RAW file formats are using it now.

Compression in most digital-image formats, including JPEG, is achieved by discarding data that the human eye doesn’t see. This “lossy” compression is great if you need to conserve space, but if you want to use image-editing software like Photoshop to improve, augment, or repurpose an image, you would receive better results if you have the data that was discarded in compression. For that reason, many photographers have insisted on keeping “lossless” images, in which nothing is discarded.

In the past, to obtain both lossy and lossless images from a single camera required two CODECS, something that was impractical because of the number of wires and gates it demanded in a custom silicon chip. However, two separate CODECS are not necessary for JPEG XR. “We designed one algorithm where the connections among processing components inside the chip are the same for both the lossless and the lossy CODEC,” Rossi said. “That’s a big value, and that’s a thing that a lot of consumers want.” They get the options of efficient lossy compression and a lossless image for archiving.

JPEG XR heralds changes in the color space, too. JPEG images currently use 8-bit component color to define the RGB color channels. The colors that can be encoded using this model are a significant subset of what the human eye can perceive. JPEG XR, however, allows for much broader color ranges by enabling the efficient representation of 16- and 32-bit component colors. The resulting images will offer more vibrant and more realistic color representation and will access the capabilities of emerging displays and printers.

Printer manufacturers are already scrambling to meet the demands of this new technology. The familiar four-color ink combination called CMYK (cyan, magenta, yellow, and key/black) won’t be adequate. Companies are now developing printers based on models of five to eight colors and devices that will employ some of the newer print media, Rossi said. “There is new trans-reflective material that will be lit from both the back and front … and organics.

“Organics are conducting polymers or plastics, an emerging technology,” he said. “It is possible for organics to be printed by a device similar to a laser printer, with the resultant image stimulated to emit light either by a voltage source or by backlighting. These are prints that actually emit light instead of just reflect light.”

Rossi says a new era of digital photography is emerging, one he’ll be describing for Japanese camera makers at the upcoming meeting in Kobe. “Microsoft will be taking the lead in defining … the future trajectory of digital photography from an information technology perspective. That could result in billions in revenue.”

So how did a small team in the Core Media Processing Group develop and deliver a format to replace JPEG?

“It took five years, end to end, for this to go from an idea to being proposed as a new international standard,” Rossi said. “That is a big investment by Microsoft, but it’s not a lot of time when you consider you’re replacing a technology that’s been in place for 20 years and is in billions of product shipments.

“Between image capture and rendering, JPEG is in almost every multimedia product except television,” he said, adding that aside from Internet protocols such as TCP/IP, “it’s arguably the most successful information technology standard ever.”

When Rossi speaks of members of his team and contributors from Microsoft Research, such as Distinguished Engineer Rico Malvar, as “gifted,” it’s clear he feels he’s phrasing things conservatively. “This kind of performance comes from a team that took a long time to put together,” he said. “You develop these types of competencies when you’re able to put the correct group of people together … and give them the appropriate amount of time. Then you can do big things that change the industry.

“It’s unbelievable when you know that you went against a giant group of people and you successfully proposed a successor to one of the most successful and longstanding standards in information technology,” Rossi said. “That’s quite a feat.”