“Image Regions” are introduced in new version

The IPTC has released the IPTC Photo Metadata Standard 2019.1 This new version of the fundamental standard for descriptive embedded metadata features Image Regions, which users, particularly including application developers, can use to define an area within an image.

Let’s say you want to tag faces with names that your AI recognition software assigns to them. To do that, you’ve got to carve out a shape in the photo and tell your program – and the rest of the world, ultimately – just what is depicted within that area of the picture. 

Old hat, you say? There are already programs doing this, you say?  True enough. There are programs out in the wild that tag faces and do that sort of thing. Until now, every developer has had to make up his or her own scheme for defining those regions and the results have not been portable between applications. 

Now, those days are (can be) over. You can mark up a picture of your board of directors in one program, send the photo around the world, and a user on the other side of the globe can use a different program to read the picture and, like magic, your CEO remains your CEO and doesn’t swap heads with the guy from Human Resources. Glory be! Standards! This is what they do.

 

Standards or chaos

Think about star ratings and color labels. Both of those features started out as proprietary implementations in photo editing software. The IPTC was able to wrestle star ratings to the ground and standardize them. So now I can assign a rating in Photo Mechanic and read it in ON1, or Lightroom, or Capture One, or XnView, or….. Just about any editing program, so long as the developers have taken the effort to make their programs industry standards compliant. What a wonderful world stars live in.

Color labels, on the other hand, haven’t faired as well. They started out as a proprietary feature in Photo Mechanic and just sort of evolved from there. Developers have informally arrived at a consensus on at least where to store label information in a photo, but that’s it.

An application can be configured to play nicely with color labels from some other app – if the developers agree on a common set of colors and text labels for those colors. Or let their users customize their own copies of their programs. This means that a photo labeled in one app may or may not read correctly in a different app, or even another copy of the same app. Or, worse yet, they may appear to work but actually display something random.

It’s a rare week that goes by without somebody asking me to help unravel some color label confusion.

Download the new standard in PDF format here.

So, thank you, IPTC!

The IPTC has just thrown a lifeline to developers who want to move from the chaos and darkness of the insular past to a bright future where we can all just get along. 

So, developers, grab the standard and follow it! Make the world and your own app better. Keep those CEOs and Human Resource guys in their own heads. Be the star ratings and not the color labels!

The Image Region standard is pretty simple. There are fields to define the shape of a region, the coordinates of the region, the copyright holders of the region, and what’s depicted in the region, and so on. You can look at demonstration pictures on the IPTC site here, read the standard itself here, and read the Image Regions chapter of the IPTC standard users guide here, and the official IPTC announcement for the new standard here.. 

The always-proactive Phil Harvey has added support for the Image Regions tags to ExifTool (as of version 11.74) so you can use ExifTool to read and write the tags. But writing a front end that can put them to practical use is on you.

Has anything else changed?

Most of the readers of this blog are photographers, digital asset managers, or developers who write tools for them. We care a lot about the metadata fields that are defined in the IPTC standard’s Core Schema and somewhat less about the fancy stuff – like Image Regions – that live in the extended fields. 

What’s in the new standard that we might have to worry about? Nothing, basically. The Core Schema remains unchanged at version 1.2. For most of us, it’s steady rolling.

The new version of the standard and its user guide can be found on my Glossary/Documents page, here.

 

Are you using a program that utilizes the new Image Regions tag? Or should? Are you writing such a program? Jump in the comments and let us know. And reach out to your favorite developers and keep them on the standards-compliant straight and narrow.