Episode 12 - Embedded Metadata with David Riecks
Hello friends! Welcome back to the podcast all about metadata. I am delighted to bring yet another guest speaker in today to talk specifically about photo metadata. I think we’re going to see a good trend toward more and more guests as everyone seems to be responding well to it. I also LOVE how much you all loved the Z39 episodes. Leave it to metadata nerds to get all amped about a standard protocol.
So today I am joined by David Riecks. David founded ControlledVocabulary.com and has been involved in many digital image standards initiatives as well as being a speaker at numerous industry events and serving on various committees for photo metadata standards. And I love his history of how he got associated with embedded image metadata:
Some time ago, David had been working with custom-made text-only "databases" to save basic information about film based images, but then later he moved on to using Lotus 123. Now, at the time he would have called this "Image Management" since he was exclusively dealing with photos.
And in 1996, he discovered that information added to images in the "File Info of Photoshop 4" could be ingested into Kudo Image Publisher at the same time a thumbnail was created. This was the first inkling of Digital Asset Management. Kudo was impressive as the ingested info – the metadata – could be exported to Filemaker 4, massaged and brought back.
With that, he began extolling the virtues of "embedded metadata" using the International Press Telecommunications Council standards (which you probably know as IPTC). That led to volunteer work with the IPTC in revamping the IIM standard to work with XMP.
So, David has worked within the Photoshop Image Browser (which became Bridge), iView Media Pro, Photo Mechanic, Lightroom, Canto Cumulus, and more. And he’s been serving on the Photo Metadata Working group ever since the IPTC4XMP project concluded, and now is looking at metadata for audio and video files as well as photos. So David, thank you very much for joining us today.
You have had a 30+ year as a photographer, and now you're focus is on photo metadata and controlled vocabularies. In my experience, photography is probably the most common background for folks interested in photo metadata to come from, alongside library science. A lot of digital asset managers are either past photographers or librarians. And probably for the same reason that photographers are prevalent as digital asset managers, images tend to be the most common digital assets to move through their lifecycle with embedded metadata. Can you tell me, with your dual perspective of photographer and metadata thinker, about the value, process and longevity of metadata actually embedded into a photo?
Sure, well, the first thing is that photos are probably the most prevalent, but there are also a lot of audio files ever since mp3 players have become popular, a lot of those have embedded metadata in them as well. Both of these suffer from the same problem, which is that they don't contain words by themselves, so in order to find photos especially, you must tag it with some kind of text and in order to do that properly, you need to get down to the essentials of what something is about. That's something that I discovered early on with photographs is that they can mean something else to different people. So you need to use your own words to describe them, but then you also need to think like the person that is going to be searching for them. And I am building my own photo file and managing image collections for others, it dawned on me that the ways of storing press and negative are inadequate, especially when you have to deal with actual film, negatives, print, things like that.
But what I think is amazing about the digital world is that you can have multiple copies that are essentially identical. In the old days with film, each print you made was less than the negative, even if you tried to make a negative of the negative, it was never as good as the original. So there's really no degradation, and if you put a tag with it, then you could find it simply by searching for the keyword. Many of the operating systems will allow you to do that. Imagine if real life were that way. You just say the name of something and it pops up in front of you, that would be great. We have that option here, it just, it does take some time to put those keywords, those tags with them and my support of metadata is driven by that. Some people want to express ownership "I am the creator of the photo" others are more concerned about telling you what is going on in the photo or where it was taken, and those are all things that you can do.
that's great. I also find that it's pretty rare among digital assets, but maybe you've had a different experience, that embedding metadata is a common practice so much as it is with photo metadata.
I know that there are a lot of musicians that are struggling to embed stuff. Photographers by and large, I had worked with a group called the Stock Artist's Alliance, and a lot of the stock photographers there, their primary concern was "I wanted to have my creator and my copyright notice so that I can claim ownership" and for other people, well it depends on how you're using it.
Stock photographers are creating stuff and sending it out into the world, and other photographers may be more concerned about the historical relevance, and if you're a newspaper photographer, you're probably more concerned with the five W's and an H - the Who, What, When, Where, Why and How. In the old days too, I've been around long enough that I can remember an AP photographer that had one of those spinning drums with an ecoustic coupler for a phone, and they made a print and it typed out a caption and they taped it onto this thing and it spun around and sent it through the phone line.
Even then they had tp type a caption, and for a long time I know that people were really good about writing something on the back of the print. But when you get into digital, there is no back of the print, so you've got to learn how to embed if you want it to stick with it as it travels through things. But even then, one of the projects that I have been working on, starting with the Stock Artist's Alliance, but we picked it up in the IPTC and have been carrying it on. We've been doing a test every three years the social media systems and whether or not they retain the embedded metadata, because that is a big problem. You spend this time getting that stuff into the file and then you upload it into Twitter and, poof, all of it is gone.
Wow, that's a good point. Now you have a website called controlledvocabulary.com (there will be links on the site page!) that is dedicated to sharing knowledge about the use and value of controlled vocabularies. How do organizations who don't have their own controlled vocabularies for images define the right set of controls for their own image metadata?
I would say for the most part, they don't. Until it becomes painfully obvious that they should, it usually becomes phrased as "I wish we had started doing this five years ago. The problem - a lot of people that get involved with it, especially that I have been working with in the last decade or so, are single users or small groups of people, and a lot of the tools that are available for those users are very crude. Rather than full-blown ontologies or controlled vocabularies, at best what you're looking at is a hierarchical set of controlled keywords, so you've got five, ten, maybe fifteen different top level and then you break each of those down, so if you have animals, you might have a break between livestock and pets, and if it is pets, are they dogs or cats or ferrets, and you just keep digging own into the hierarchy.
And then, by and large, with the exception of maybe lightRoom, most of them will end up taking those hierarchical keywords and just parsing it down into a set of individual keywords and applying it to the image. That can create some problems, but it is better than nothing. On the higher end, some enterprise kinds of systems, there are better tools, but the costs of those are enormous.
And now you are an active member of the IPTC Photo Metadata Working group. Can you tell me more about that group, what it has accomplished and what it is looking toward in the future?
Well, first, a lot of people know this, IPTC stands for the International Press Telecommunications Council, and I have learned from my Google alerts that there are some other organizations that have the same acronym, so you can occasionally see some interesting things. I would have to go back a look to be sure. but I know back in the 80's there were some forward-thinking people there who were looking at, what are the distinct fields that we need to have in terms of metadata in order to store either, with an image - and it wasn't just images. They were looking at all news materials, audio, photos and things like that. What are the kinds of things that we need to know? The date it was taken, where it was taken, when it was taken, who took it, who wrote the caption, etc.
So, sometime later, not sure exactly when on the exact date, some time in the late 80's, early 90's. Adobe took some of the 18 or 20 top fields that were in the IPTC schema and applied them to the version of PhotoShop, which would have been PhotoShop 4. That would have been the early-mid 90's or something like that. That's when a lot of people first began to notice them because there was this file, and the file info and you could fill out all that information there. And again, one of the groups that I was working with then, The Universal Photographical Digital Imaging Guidelines Group - one of the things that we noticed right off the bat is that when Adobe introduced their Save for Web feature, initially it was stripping all of the metadata out of the image. We had been so used to - you scan something, or you have a digital file you're working on or it is a PhotoShop file, and you save it as a Tiff or save it as a jpg, all that information is in there. But then you started using Save for Web and everything was gone. So we lobbied to change the defaults on that to at least save a core set of metadata. And then if you changed the pull-down menu, you could have it save everything as well.
I got involved with the, when Adobe was working with IPTC is about 2004. They were updating the way that PhotoShop stored some of that IPTC metadata. It was a little wonky. The Image Resource Block is what Adobe called it. It was so wonky that, if you used regular English you were generally OK, but if you were using French or Spanish where you have little accents or umlauts or something like that, if you put in the metadata on a Windows computer and then saved it and opened up that file on a Mac, it would have characters you couldn't recognize wherever those little accents were. And same for vice versa, if you saved it on a Mac and opened it up on a Windows computer, you would have some problems. And so Adobe wanted to get rid of that and that became their who XMP thing. So they were basically rewriting how the information was stored. That's how I got pulled in, based on some of the stuff I had on the controlled vocabulary site.
Yeah, I was wondering about that. Does IPTC collaborate with the makers of EXIF or XMP or any of that to avoid overlap between these extremely high-use schemas? Do you make use of other schemas in your own work with photo metadata?
Well, I am involved with another group called Plus that has a set of terms that are stored within XMP, and a large number of those are included as part of the IPTC extension. So the 18-20 fields that are standard were in PhotoShop way back. When those first got converted over, that became something called the IPTC Core. Because there was like Dublin Core and everybody else was using Core so those were the Core. And then after that, there were a lot of people who came forward and said "hey this is great, but we're doing cultural heritage so we need this and that" and so, then those other things got added to a section called the IPTC Extension. And they're all written in XMP.
Now you go back to EXIF, the one distinction there, if you talk about technical metadata and administrative metadata and descriptive metadata. By and large, what is in IPTC would be considered descriptive metadata or administrative metadata. And EXIF is largely technical, although there is some overlap. And there is another group, they are not very active now, they may be in hiatus, they were called the Metadata Working Group, and they had representatives from Cannon and Microsoft, and I think it was even Nokia - because at the time, the company making the most cameras in the world was actually Nokia. They were small, and they were usually used to make calls, but they had a camera in them.
And so that EXIF information, one of the things that the Metadata Working Group, one of their main concerns was, we've got some fields in EXIF that are also in IPTC and also in XMP, and so if you're making software, when you save something that was put in XMP and also was in EXIF, you need to make sure that you update that as well, and so there were a lot of synchronization issues.
Great. Last question. How do people who are not familiar with embedded photo metadata get more information?
There is the ITPC website. ITPC.org. I also maintain a website we started under the Stock Artist's Alliance called Photometadata.org. Both of those have a lot of information about how to apply metadata. The IPTC site and my controlled vocabulary site both have lists of software for embedding metadata. There is also a site that we put up called embeddedmetadata.org. That was really pushing the whole concept of - once you put stuff in there, well it's like the "do no harm" idea. You have this metadata in the file, so you shouldn't be building software that is just going to strip that out. It gave some guidelines for things like that.
And along with that, that is where we first started doing the testing of the social media sites. IPTC has also done some testing of various software that can be used to read or write metadata. So those are all good places to start if you want to find out how to apply stuff. And those who are interested in video should be aware that IPTC has been working on something they call the Video Metadata Hub. It started in about 2008, but has only come out in the last couple of years. And what this does, it is not so much about embedding metadata. What it is, is coordinating and mapping the various metadata. The European Broadcasting Union has a way of embedding, or at least storing video metadata, but so does PBCore, and all of these other different groups. And so we've tried to create something that has all of these different fields in it and mapping across them and then if you're working with PBCore and you're moving things to EBUCore, then you've got a map for which fields match up with which fields.
You said that exists or that's coming?
No that already exists. We just know from working with various vendors that one of the problems is that if you give an engineer a mapping chart, it can be done in very short order. But when they have to sit down if this is the same as this, then you have to have a committee, and then months go by. And this way we've already figure it out: this is the same as this. And we've worked with the other groups to smooth that over and figure it out and maybe change the wording of our definitions and things like that.
I am going to be looking into that very soon ,actually
Great. That wraps up my questions. Thank you for having this conversation with me today.
No problem. Glad to add to the knowledge that's out there.
And then, did you have a reading recommendation?
I was hoping you would ask that. I have a few. The first is
"The DAM Survival Guide: Digital Asset Management Initiative Planning" by David Diamond
This is a book for those who are managing large photo collections (or other types of digital content) and need help in organizing not only the content but the PEOPLE involved in the initiative.
The second is "Personal Archiving: Preserving Our Digital Heritage" Edited by Donald T. Hawkins.
This is kind of an overview, but for someone who is trying to figure out how to manage all the various digital content they have or create, you are bound to find some good starting advice.
And the last one is "Everything is Miscellaneous: The Power of the New Digital Disorder" by David Weinberge. His central premise is that there is no universally accepted way of classifying information; but he really gets you to think about ways in which you could make stuff more accessible. Weinberger notes that "our homespun ways of maintaining order are going to break—they're already breaking—in the digital world." What happens when we liberate the metadata from the physical manifestation of that information? In Weinbergers mind this is the "3rd Order of Organization" where we are no longer bound by physical constraints.