Outside Winnipeg

It used to be just about sharing photos with friends, but now its so, so much more than that.

Flickr announces Machine Tags, (via Mikel) which is formal support within Flickr’s database and API for what are also known as TripleTags, namespace:predicate=value.

Example:

geotagged
geo:lat=43.245
geo:lon=-83.211


The idea was to add some specific meaning to a set of data (photo) in a way that was accessible to the user through the user interface. Savvy types would add keywords to the EXIF Metadata, but this is tedious, and this metadata is often stripped out when using some image conversion utilities for resizing/cropping/etc. Therefore, soon after the popularization of folksonomy, users found a way to also add ’semantic’ data.

However, this usage, while working, was not inherently supported. People used it as a way to find all their ‘geotagged’ photos, or photos of forests, and so on. But parsing them and dealing with them was sometimes a pain from the developer perspective, and it also made a mess of your tag clouds or tag listings as you had all these one-time-use tags for the specific lat/lon point.

Now that Flickr internally supports machine tags in a good way, developers can really start using this as a way to attach arbitrary metadata to any data item (photo, etc.) Dan Catt gives a good discussion on what it all means and how it can be used.

I wonder if a Microformats-like impromptu standards body will pop-up to help formalize the namespaces and predicates. Why ‘geotagged’ and not ‘geolocated’, ‘geo’, ‘geomarked’, etc? While free-form naming is nice – so very extensible, being able to use a common vocabulary would also be nice.

Perhaps just accomodating current standards and applying them to the machine tags would be good without requiring any additional work. I’m thinking like hcard:fn=andrewturner, hcard:url=http://highearthorbit.com, hcard:phone=555-1212, etc.

You the developer

So, the end result to the developer is to check out the additional parameters to the Flickr search method in the API. There is now a machine_tags optional parameter that allows you to search by namespace, predicate, or value and use boolean constructions for these parameters.

Dan Catt even references Spimes as an excellent example of why machine tags can really empower devices, especially sensored/automated ones. And now that Flickr has no limits on uploads, you can create frequent update-rate sensored images.

When Flickr added built-in mapping of photos, many rejoiced. However, it can be tedious to go through hundreds/thousands of photos and dropping them on a map. The User Interface for the Flickr Maps is really great – however, with this many photos, it would just take forever.

In addition, your photos are only geotagged in Flickr – and therefore not easily usable outside the service. The better way to geotag your photos is to actually write the Geo data to the EXIF of the photo. Then the metadata is carried around with the photo itself (until you pass it through some mean, metadata chomping machine like Photoshop).

The way I geotag my photos is to first get the coordinates of photos:

1. Carry around a GPS and store the tracks as GPX files – then you can mesh the GPS with the photos using WWMX (Windows), GPSPhotoLinker (Mac), or various scripts in Linux (fend for yourself, but check the geowankers mail archive)
2. Mark GPS Waypoints – or lookup addresses of locations and use MultiMap to get the latitude/longitude of these points
3. Guess

After I’ve either meshed up my coordinates, or have a list of locations, I fire up iView Media Pro, or iPhoto, and use my Applescript scripts in addition to ExifTool to actually write the GPS metadata. Because photo editing applications (like the aforementioned Photoshop) are usually very mean and don’t restore geo-metadata on edit and save, I suggest you edit all your photos first, and apply the geo Exif as the last step before uploading.

Now that you’re going to upload your photos, you first need to make sure Flickr uses these geotags for actual mapping. Enable Flickr to read your Geo EXIF tags. If you already have uploaded photos with geo-coordinates in the Exif data, Flickr will add these to the map (after a short wait – queueing and all).

You can my Flickr Photo Map, and you should go take some photographs! (especially for Pentax Day)

The idea of geotagging photos is not new. With the huge swell of interest around mapping when the GoogleMaps API came out, very quickly people starting mashing up Flickr & Maps. Flickr even hired Dan Catt of GeoBloggers quite awhile ago.

YahooFlickr finally released their built-in mapping tools.

Here’s how their mapping works:

1. Go to “Organize” your photos, and then select the “Map” tag
2. Answer the Privacy question – location privacy will become more of an issue with the rise of social location tracking, geotagging, etc.
3. Flickr sees if you have any existing geotagged photos and places them – it appears to just use photos that have been tagged with geotagged and not read the location information if it is embedded in the EXiF. Hopefully that is something that will change
4. Drag photos from the bottom well onto the map. Make sure to place the round button below the photo thumbnail. You can then add more photos to this stack of photos at a location. Also, place photos by the subject matter, not where you took the photo from. If I’m looking for the Eiffel tower photos, I don’t want to see it placed down the Champs d’Elysee
5. Photos now have a “Map” link below them on the photos page. You can also explore photos near where your photos were taken – so very social

Overall it’s a very good user experience. It’s nice now that mapping is baked-right-into Flickr, which means quicker adoption rate, and easier to get going with your mapping (rather than being relegated to various mapping utilities and scripts)

A couple of things I hope Flickr adds very soon:

1. GeoRSS output in the feeds
2. Add Microformats:geo to the photo descriptions
3. Markup the location description of each photo in Microformats:adr format.
4. Parse the Geo Exif tags in photos

Its typical, you get a nice camera (or two), go on a couple of trips, and you’ve very quickly created gigabytes of photographs that you then need to sort through. Automator doesn’t quite have the ability to bring in Riya for marking the people in your photos, or rating their quality, so you spend your Sunday afternoon just marking, annotating, rating, deleting, aligning, color-correcting, sharpening, etc.

I’m using iView Media Pro, and always wondered about the Subject and Scene options in the Metadata. iVMP has the best metadata editing I’ve found in any DAM. There just seem to be perhaps too many metadata options.

Turns out there is this whole history to photography. In fact, there are 28 groups of terms called NewsCodes that are used to help fully annotate media. These NewsCodes might specify that a media archive is of “cartoon”, “criminal”, “nightclub”, or “derivative securities” and that the location was “underwater”, “aerial”, “rear view”, or “offbeat”. The IPTC NewsCodes offer a complete listing of the currently suggested vocabulary. And they’re ever so thoughtful to offer a Windows program for viewing and saving the NewsML files.

iVMP allows you to easily bring in Vocabularies like the IPTC NewsCodes, except in simple text format. They’re easy to manipulate, and I’ve formatted the Scene and Subject Code files that you can drop into (remove the underscores, or merge the files into your existing vocabulary files):

~/Library/Application Support/iView/Plug-ins/Vocabulary/Default/

Maybe I’ll put together a quick viewer or formatting utility if there’s interest.