Last October I wrote about the importance – and the difficulty – of embedding descriptions into digital photos. As a test, I ended the article by asking readers to download a photo into which I embedded a quote from Benjamin Franklin. I also asked readers to let me know if they could see the quote and, if so, to name the program they used to display the quote.
Here are some results and conclusions from that test.
• Over 80 readers responded
• Almost everyone could see the quote
• Most of the respondents used photometadata freeware to display the quote
So, in general, the test confirms that there are few constraints — technical or financial — to displaying photometadata. Phone-based and standalone digital cameras can display some basic photometadata (though I don’t think it displays descriptions) but you can’t add text. And, of course, you can display and add metadata to a digital photo on a computer with the right software and a little effort.
One other result took me by surprise. But I’m going to save that for the end of this story and use it for a new experiment that will tie in with this photo of children.
Asking people to add descriptions to their digital photos is like asking them to write descriptions on back of their paper photos. Doing so (writing on paper photos) is a good practice but most people don’t do it, including me. It’s time consuming, you have to wait for the ink to dry until you can stack the photos and so on. But someday these descriptions may help jog your memory or help other viewers understand the content of the photos. (“Oh, that’s grandma when she was a teenager.”) Similarly, adding descriptions to digital photos — photometadata — is a good practice but it is also a chore.
Mike Wash, the CIO of NARA, was an Kodak engineer for about 30 years. While at Kodak, Wash was responsible for many of the automated developments that we take granted on cameras today, like autofocus, automatic light adjustment and technical metadata recording (such as the date and time the photo was taken and the light and shutter-speed settings).
I asked him if it was possible to have a feature on a digital camera that would enable users to easily add metadata. Wash said, “I think that nearly everyone would agree that some sort of data associated with an image is valuable. But the hardest part is going to be dealing with the variable nature of what type of information you would use. Creating an ‘all things to all people’ type of one-touch metadata entry is going to be pretty hard.”
Wash added, though, that “documenting information about the photo” is one of the features his team of engineers at Kodak had on their list of unmet consumer needs. However, people will not want to allocate any time to the task of adding metadata; for example, typing a description into a camera or a smart phone would be awkward, time consuming and unappealing. “It has to be automatic,” Wash said. “Beyond easy.”
The idea that makes the most “beyond easy” sense is also a bit daring: voice-to-text software. With the photo displayed on your camera, you click a button or open an app, speak a few words about the photo into the device and the text of your words gets embedded into the photo file. Both Mike Wash and photometadata evangelist David Riecks support the idea.
Wash said, “Voice technology has improved a lot over the past few years. And that would easily translate into a low-cost feature.” Wash stressed that the cost threshhold is crucial to public acceptance. He mentioned a feature of the Kodak Advantix system that didn’t get developed because the budget to put it into a camera, at .25¢, was too much. Wash said, “You’re aiming for tenths of pennies here and there because all of that equates into less margin or too high a price to dissuade a consumer from buying your product.”
So we have the technological capability to add a voice-to-text photometadata function to digital cameras, and we have a need to embed descriptions to our digital photos, but –- for practical and economic reasons –- it must be cost effective for manufactures to develop it. It is likely that once consumers know that it is possible and easy to do, and they see the value of added descriptions in helping organize and find their photos, the market will favor the camera manufacturers who make that feature available. In a short time it could become a standard feature.
One more word about Mike Wash’s support of adding descriptive information to photos. It’s not just an interesting “cocktail conversation” idea for him. He actually has two vintage 100-year-old Autographic cameras. One sits on his desk. Kodak built the Autographic to enable users to write descriptions onto photographic film. After you take a picture (using special Autographic film) you open a little door in the camera back and you can write a note that would come out between frames. Wash said, “You just hold the camera with the little window up to a bright light for five or ten seconds and close the door. And it actually exposes your metadata onto the film between the negatives. It’s pretty cool.”
Now for the new experiment.
In the “find the photo description” test we conducted in the previous blog on this subject, a number of people found the quote embedded into the photo simply by searching their computers. I wasn’t aware that this was an option but, sure enough, current operating systems (Mac, PC and Linux) index any text they find in files, including photo files. Which means that you can search your computer for specific text (the computer will consult the index) and it’ll find the text, whatever it is, wherever it is and display — in the search results — the file in which the text resides.
This has some intriguing implications for photo storage. For example, if you know that you have a particular text in the photometadata (e.g. “Golden Gate Bridge”) but aren’t sure what the photo file name is or where on your computer you stashed the photo, you can still search and find that photo.
So here’s another test for you to try. In the photo at the top of this page, I embedded the following quote from the writer, Michael McLaverty, “When the heart’s cold, the voice of a child can warm it.” Download the photo file to anywhere in your computer (Either click and drag the photo to your desktop or on a PC: right-click and “Save As” and on a Mac: Control-click and “Save Image As”). Then search for the word “McLaverty.” (PC: Start menu > Search > Files or folders; Mac: Search).
Outcomes may vary, depending on your operating system and version, but in the “search results” you should see the photo of the children.
January 25, 2012 at 9:49 am
Found the photo in a heartbeat! Great info, thanks!
January 25, 2012 at 9:58 am
The exercise works particularly well if you have the freeware program Irfanview on your Windows computer. After the file is located through the search, Irfanview will show you the Exif or IPTC metadata.
The Nikon D3 and other pro cameras have a voice memo feature. But the memo is recorded as separate .wav files and there is no word recognition text nor is any speech information embedded in the file.
January 25, 2012 at 9:59 am
Thanks for the response and the information, Barry. Voice-to-text can’t be far away.
January 25, 2012 at 10:04 am
I found the photo in my search results and clicked on it, but not locate McLaverty’s name or the quotation. is there a next step to locating the quotation?
January 25, 2012 at 10:28 am
Worked like a charm on my Mac. Thanks!
January 25, 2012 at 10:38 am
David,
If you look at the comments on the bottom of the page of part 1 of this article, readers mention the software or methods they use to display photometadata. Some software is easier to use than others, some software is free while other software is not. In comment #2, above, photographer Barry Wheeler recommends the freeware program Irfanview. You may have to try several to see which one you prefer.
January 25, 2012 at 11:01 am
Just tried it on my PC at work. I found the picture, no problem.
January 25, 2012 at 11:26 am
Thanks for testing it, Margaret.
January 25, 2012 at 11:37 am
Found it easily with Windows 7 search. I opened the image using Windows Photo Viewer. Right-click on the photo and select Properties to see the quote. Very helpful – thanks!
January 25, 2012 at 1:53 pm
I found it. Great. Thanks!
January 25, 2012 at 4:37 pm
Found it easily with OS X Lion Finder search. Only other metadata is creator as Library of Congress.
On the topic of ‘automatic tagging’, sometimes if something is widely-known, what you see in the image may tag the image, either by machine or human eye. So Golden Gate bridge–if read by human as text (say a scanned postcard legend), or image recognized by machine vision (Google goggles?), could get tagged by info. like map coordinates.
In one case, I wanted to find the source of an special image. OK, this was an illustration, not a photo, but the metadata issues remain: who is the creator? How to find out if distributor embedded no metadata in the file on the web? I did a search using the text as visible in the illustration–luckily this visible text was unique and popular: “I am a Great White Shark”. Found it on the web right away (sadly, no-one seems to know the creator).
The visible text was not part of metadata, but now I put that in the Description field and I can more easily find the image file again using a computer search, just like the case with your quotation in the photo of the children.
January 25, 2012 at 6:29 pm
Thanks for sending this along, Mike. I’m subscribed to this blog now.
As to the test, I’m on a Mac laptop with a fairly recent system. I found it quite easily in Spotlight and Find.
I also looked at the metadata in Photoshop and Lightroom. Again, no problem.
In Preview, you have to search a little for the metadata by going to Tools–>Inspector. However, the keyboard command is Command I, which is the same as the System if you have the file highlighted in the directory.
January 25, 2012 at 9:17 pm
Interesting that ‘Save As’ from within NetNewsWire on my Mac didn’t store the original file (note the difference in filesize as well as the lack of the relevant quote).
Apparently, the file sent via RSS is not the same file being served on this page.
Terrell
from RSS:
$ jhead Desktop/www_downloads/children_8b30595u_2501.jpg
File name : Desktop/www_downloads/children_8b30595u_2501.jpg
File size : 22062 bytes
File date : 2012:01:25 21:11:16
Resolution : 250 x 263
Color/bw : Black and white
from this page:
$ jhead Desktop/www_downloads/children_8b30595u_2501b.jpg
File name : Desktop/www_downloads/children_8b30595u_2501b.jpg
File size : 61139 bytes
File date : 2012:01:25 21:12:46
Date/Time : 2011:12:29 08:27:52
Resolution : 250 x 263
Color/bw : Black and white
Jpeg process : Progressive
======= IPTC data: =======
Record vers. : 0
Caption : “When the heart’s cold the voice of a child can warm it” Michael McLaverty
Byline : Library of Congress
January 26, 2012 at 11:59 am
Windows 7 Search found it.
With the advent of touch sensitive screens and handwriting recognition software the Autographic method may make a resurgence. GPS data and time/date stamp are already a part of some cameras, as well as some basic facial recognition software. Integration of these features plus the development of new tools could help amswer many of the “who, what, where, and when” questions that viewers may have.
The greatest obstacle that photographers have are themselves. When a photographic outing results in 500+ photos, then the task of writing captions seems insurmountable. Software exists to help annotate multiple photos at once but it might be better for the photographer to cull their images to reduce the workload, or, perhaps, be more judicious in selecting what they photograph in the first place and how they do so.
The Dagurreotype photographer had many limitations to contend with but made some beautiful images within those constraints. Who has the patience to view 500-700 photos of Disneyland, or any one location at one time?
January 26, 2012 at 10:30 pm
On a Mac, OS 10.7.2, browsing with Safari 5.1.2.
1. “right-click” on image and save to Downloads folder.
2. Immediate activate Spotlight search (control-space on most Macs).
3. Enter “McLaverty” into Spotlight search box which opens up at upper-right of screen.
4. Photo pops up within 2 seconds.
BTW, what’s the point of saying that metadata is readable only with computer when digital photos themselves have the same limitations? (Printed photos are no longer digital, of course.)
January 27, 2012 at 9:33 am
I searched the desktop on a PC. Both the title of the photo and the photo showed up.
January 27, 2012 at 10:09 am
John, you’re correct about the photometadata and I rewrote my statement accordingly. Some photometadata is available on digital cameras, both phone-based and standalone cameras.
January 27, 2012 at 10:23 am
Windows 7 Search found it.
By the way, hovering over it with the mouse displays as Title the quote and its author.
January 27, 2012 at 10:31 am
Wow. Even better. Thanks for that bit of information, Dan.
January 27, 2012 at 10:52 am
Mike:
I think that Mr. Wash may be focusing on the “cost threshhold is crucial to public acceptance” thought too much (which might be why Kodak is where it is today?) Apple has never shied away from pricing their devices at the upper end of the price scale and they have enough cash that they could buy the country of Greece at the moment.
So I think there is greater potential for some advanced methods of adding descriptive information at the time of capture than some might think.
A little more thinking outside the box might help to “connect the dots” that are already available. There are already voice to text features in smartphones now (Siri on Mac, and other options on Android). There are also features available now in some of the new pro digital SLR cameras that allow for input of data (via cable or even Wifi) to add info to a photo at the time of capture (many GPS devices do this right now). What is needed is a way to take direct user input and move it to a place in the camera where it can be used without great effort.
1. Talk to a “voice recognition” app in your smart phone, or type in directly.
2. Edit text if necessary.
3. Press button to load this into a metadata template in the camera.
4. Press shutter.
5. Information is written into the image file in one of the various metadata containers.
6. Lather, rinse repeat, until the option to add input is turned off (or have a toggle switch at input note whether this is for one shot only, for a specific time period, or to be used until turned off).
The only downside to this method is that the descriptive info must be prepared and put into the camera in advance of taking the image (at least for “in camera” use).
There are, however, already ways to record info after the fact that already exist. As Barry Wheeler noted there are “voice annotation” features available on some digital cameras. I have a microphone that is built into my Nikon D2X that I purchased in 2005. I can press a button after I take a photo and speak into the mic and record a message (you can, if you wish, even set it up to ALWAYS record what is said after the shutter is pressed, but I don’t use it often enough to make that worthwhile). However it doesn’t put that info into the image itself, it simply records a WAV file and tags it with the same filename and date stamp. You have to play it back later and write the info into the photo metadata in order to embed it so that it travels with the file. This is most helpful if you want to make it searchable, for later “discovery” purposes.
David
January 27, 2012 at 11:33 am
In regards to Terrell’s comment about the image delivered via RSS not having the embedded info, this doesn’t surprise me in the least. Even though the pixel size (resolution) remains the same, the size of the file on disk has changed, so something happened when the image was delivered via RSS.
Many images that are uploaded or “shared” via social media or photo sharing sites suffer the same purpose. See the results of an ongoing survey of these sites/services at http://www.controlledvocabulary.com/socialmedia if you are interested in seeing which services don’t respect your photo metadata and which DO preserve it. One hint for many is that the most flagrant is facebook. They not only “strip” your image of descriptive and rights based information, they also downsample it to a lower resolution. So don’t think that these services are a substitute for managing your own digital library or you could be in for a shock.
David
January 27, 2012 at 11:37 am
It doesn’t surprise me at all that the results from your users conducting this test are finding that their results vary by computer OS and version (not to mention that it could vary even more if you had used a metadata field that was newer, or not common to all applications.
Assuming the information is in the caption field, Spotlight on the Mac OS should pick this up as far back as OS 10.4 (10.5 for sure). Mac OS 10.6 added support for XMP (at least thos fields within the IPTC Core) so that would find it regardless of which field was used (see http://www.photometadata.org/META-101-metadata-types for details on the varous schemas). However, things are more murky on the Windows platform. I’d had to do some more checking, but you won’t see this information within the Windows XP at an Operating System level — you can with other applications like irfanview, or MS Pro Photo Tools, and other freeware. It may on Windows Vista, but do know that this info is visible within Windows 7. If you install the free application Picasa, then you can search for this information there an it will be found (the only fields you can “see” however, are the “caption” and “Keywords” fields — and only for files stored as JPEGs). I think Picasa may also into the older “legacy” form of IPTC-IIM.
Some of your readers might also be interested in a program that the International Press Telecommunications Council (IPTC) has launched, in cooperation with a number of other partners. It’s called the the Embedded Metadata Manifesto (http://www.embeddedmetadata.org/) and was developed to promote best practices in regards to embedded metadata for all digital objects (not just image files).
David
January 27, 2012 at 5:06 pm
Mike:
BTW, the drop-dead simplest way to “peek” into the metadata of a file online (or even locally) is to use the online version of ExifTools, which can be found at http://regex.info/exif.cgi
The following link was generated by that tool specifically for the photo in this post.
http://bit.ly/yNYPe6
You can also see that you are still using Photoshop CS4 for Windows…. isn’t it time to upgrade?
David
January 28, 2012 at 11:35 am
The number of times I’ve had a clever caption when I took a picture, then later could not remember it. With a result of ‘why did I waste bits on this one?…’
I don’t care if there is voice to text — (not really, I do care. Make it easier.) I would love be able to narrate a sentence or two when I take the picture and have that as a sound recording attached to the picture.
Implementing it:
Sound recorder goes on when the shutter button is half depressed. If a picture is taken, the sound byte is attached to that picture. No picture, then it’s an indpenedent voice memo between the previous frame and the next frame. This would mean that you would have a set of sound bytes interleaved with your images, like mixing two decks of cards.
Back at the computer you could:
Turn voice to text
Discard, or attach to one of the frames.
Given that any camera that can record video already records sound, adding this should be a software mod for such a camera.
Nikon, are you listening? Put this on any camera that has a microphone.
Hello? Hello? Nikon?
Nikon never listens to me. Sigh.
January 29, 2012 at 1:43 pm
Worked as quick as a flash. Thank you for this column and information.
January 29, 2012 at 3:05 pm
I found the photo also … Didn’t know you could drag a photo onto the desktop.
January 30, 2012 at 7:03 am
Hi,
even Windows XP’s search tool works.
Kind regards,
Guy
January 30, 2012 at 8:09 am
Sherwood. In this list of comments, #2 and #20, photographers Barry Wheeler and David Riecks mention a few cameras that record audio. Also, Riecks gives the contact information for camera manufacturers on http://bit.ly/sbNrU0. I’m sure they’d love to hear from you.
January 31, 2012 at 7:50 pm
I downloaded the photo to my hard drive by dragging it to my desktop; I use a Mac PowerBook G4 and my OS is OS X 0.4.11. My browser for this exercise is Safari 4.1.3. I used Mac Spotlight to search for “McLaverty” and got the photo on my hard drive immediately. Thanks for this tip! My iPhoto library just got much more user-friendly since I can search for photos via metadata without needing to have iPhoto open at the time.
January 31, 2012 at 11:26 pm
I found the text without difficulty using Windows search, however this does nothing to solve my other concern, about the fragility (or instability of representation) of metadata attached to digital images.
For example, I opened this image using IrfanView version 4.28 and found the quotation in both the Exif and IPTC metadata. Then I saved the image in PNG format. The Windows search found the PNG version also, but upon opening it with Irfanview, neither the Exif and IPTC metadata were accessible, and there seemed to be no available change to the settings to make it available. Then I saved the PNG as a JPG again. Once again, Windows search found the item, however upon opening the new JPG, only the IPTC data appeared, and the field which contained the quote in your original version was now blank (as were all the other fields, in fact). It appears that the text is being kept with the image during the save-as process, but distorted or shifted into a different header field that Irfanview does not recognize. No doubt different software would produce different results and with proper testing and techniques, this problem could be avoided. But to me, it highlights the challenge of properly curating digital photos, especially by untrained people using cheap or free software.
February 1, 2012 at 12:10 am
I dragged the photo from the blog page to my desktop and searched my Mac using spotlight. Found it immediately.
February 1, 2012 at 8:32 am
Jennifer, thanks for your iPhoto comment. This experiment proves that if we could (easily) add metadata/descriptions to photos, those descriptions would be viewable in a wide variety of metadata-displaying media.
February 2, 2012 at 2:29 am
I too have been on the hunt for the best way to embed metadata in images. The two stumbling blocks I came up against in software are: 1. IPTC seems to be the only standard used but it doesn’t have the fields I want (like ‘event name’), 2. I don’t trust the software will save the original image, EXIF, etc. untouched.
In the meantime, my workaround has become my standard workflow (sigh): basically I bung as much as I can into the filename! But I used delimiters that I imagined I could parse and stuff inside the image file at some point down the track.
An example:
DSC05396 2006-04-22 Chicago–Millenium Park~Inside The Bean–John Dave.JPG
ie.
Camera assigned id
Date (ISO 8601 thanks)
Location
Sub-location/event (optional, multiple)
Title (try to be descriptive)
Other keywords, usually names (optional)
Originally I deleted the camera ID, but then thought it could be useful to have a unique ID automatically in my ‘metadata’. …and then I bought a new camera which started from DSC00001 again, d’oh!
It’s crude, but the key things are: we already have tools to edit it, it automatically sorts, we already have tools to search it, and I can arrange photos hierarchically if I want using filesystem folders!
The downside is filename lengths are limited, especially if archiving to CD.
February 2, 2012 at 2:30 am
Errr, I put the delimiters in using angle brackets.
Here’s the breakdown again
Camera assigned id
–space–
Date (ISO 8601 thanks)
–space–
Location
–double dash, if needed–
Sub-location/event (optional, multiple)
–tilde–
Title (try to be descriptive)
–double dash, if needed–
Other keywords, usually names (optional)
February 5, 2012 at 7:57 am
In Windows 7, the quote shows up in both the Title and Subject fields. (Right-click on photo in search results, choose Properties, click on the Details tab.)
Scrolling further down in Details shows that the Authors are Library of Congress, gives the date created and the program used (Adobe Photoshop CS4 Windows), more info about the image (size, resolution, etc.) and fields for Camera details which are not filled in (maybe because it was scanned, not made with a camera). The Camera and Advanced Photo fields are editable.
November 26, 2012 at 12:22 pm
Tried it in Ubuntu 12.04 using Places->Search for files.
Entered McLaverty in the More options field (containing text) and it wasn’t found.