Microsoft leads effort to solve photo metadata problem

Computerworld - Have you ever been vexed to find that the titles, keywords or ratings you painstakingly entered to organize your digital photo collection disappear when you move them from one software (or service) to another?

Or were you puzzled when the data created when you originally took the photo, such as the exposure, date/time or GPS location end up garbled or missing?

That's not surprising, according to Josh Weisberg, director of Microsoft Corp.'s rich media group. Despite prior standardization efforts, interoperability of photo metadata remains dismal.

"There are several existing standards, but they aren't talking to each other," he said.

Those efforts have failed, he said, because they have been led by vendors in one link of the digital photography chain -- camera manufacturers, or photo software makers -- that didn't consider the needs of other parties.

As a result, there are six different standards for storing something seemingly as simple as photo captions, he said.

Microsoft is leading an effort to fix this by creating a single specification that will, it is hoped, eventually unify all of the existing standards out there.

Announced today at the Photokina trade show in Cologne, Germany, the Metadata Working Group has six corporate members, all leading players in their respective areas of imaging, including Adobe Systems Inc., Apple Inc., Canon Inc., Sony Corp., Nokia and Microsoft.

So far, the group, led by Weisberg, has put out guidelines on how to treat eight key metadata fields. The guidelines are aimed at makers of cameras and cameraphones, software vendors, and Web services and search engines such as Flickr and Google.

They include fields for keywords, descriptions, date and time, location (with different fields for where the photographer was and where the subject was), orientation (i.e. is the photo meant to be displayed vertically or horizontally), rating, copyright and creator.

The guidelines also ask device and software makers to ensure that no metadata is ever deleted without explicitly asking the user, Weisberg said.

The specifications do not create new standards, but build on top of existing ones such as Adobe's XMP (Extensible Metadata Platform) or Exif (Exchangeable Image File).

In other areas, such as office documents, the trend is to use human-readable XML formats such as Office Open XML (OOXML) and OpenDocument Format (ODF), and not to store metadata in hard-coded fields but to embed it -- albeit invisibly -- along with the text or data.

Why not similarly embed these data fields in free-flowing XML and require software and services and search engines to figure out how to pull it out?

While that may work for "ad hoc" data such as captions or tags, Weisberg said that approach isn't up to snuff for highly technical, mathematical data such as GPS coordinates, altitude readings or compass headings. It would create more work for developers, who "would need to write a bunch of code to interpret that data," he said.