Digital Media | Spotlight: Photography | Inside Lightroom | Blogs
Looking for the Right Level of Keywording
After using modern workflow tools, like Adobe Lightroom, for the last few years, I'm finding myself approaching a crisis point with keywords. When I first started really adding keywords to image metadata years back, I took the same approach many people use for Flickr tags. I threw as much information as I could into keywords, including human subject's names and company affiliations. I figured what the heck, it can't hurt. It's all just going into a database, right? Now, I'm not so sure.
The whole reason to keyword images is to help you, as well as other people, find what you are looking for later. If all you had were keywords, then sure, stuffing every bit of data you have into them would probably be a good idea. But, when you execute a search across your library in Lightroom (or Aperture for that matter), your search query can be executed against any searchable field, which includes titles and captions. Well written titles and captions that include the who, what, and where of an image should contain most of the unique data that you'll want to search for. For example, if I want to look up photos of Ze Frank in my library, having the string "Ze Frank" in the title and/or caption will be enough. Adding it in as a keyword doesn't help at that point and just clutters things up.
On the other hand, where keywords are perfect for the job is describing attributes about the photo that aren't in the caption, or which are generally applicable to a large group of images. For example, I took the photo of Ze Frank here at RailsConf 2007, so it's probably useful to use "presentation", "speaker", "humor", and "keynote".
Is less more? Or, is more more? Without a greater organization telling you what standards you should stick to, such as those presumably found in formal news organizations, it's kind of hard to know. I've looked a bit at controlled vocabularies, such as those used by the Library of Congress, and I can see quite a bit of value in them. On the other hand, I'm not sure that formal taxonomies are that useful to the majority of users out there. One of the big wins about seeing tagging used in many Web 2.0 applications, such as Flickr, is that emergent structure of freeform word usage can be quite valuable indeed. And of course, Google has proven the power of free-form text search.
One thing for sure, I've found my keyword browser has become utterly useless with my current style of keywording, and that's a problem. Starting now, I'm personally going to be scaling things back with a focus on making good captions and adding in keywords that compliment those captions with an eye to making sure that when I hit Command-F (Control-F for those of you that use Windows), I'll be able to type in a term and find the image I need, no matter whether that term appears in a title, caption, or keyword.
What's your take on the right level of keywording to use? I'd love to hear your two cents!
Anyone ever heard whether this is worth its money?
http://www.controlledvocabulary.com/help/cvkc-lightroom.html
I have a bit different body of subjects than you do (mostly just family), so that colors my response, but here goes:
I find keywords per family member critical to tagging and later finding people in photos. I talked about the overall rationale and my larger thoughts on keywording in http://tomdibble.wordpress.com/2008/01/15/aperture-workflow/ -- I won't bother repeating the bulk of that here.
With respect to people, I agree that for the "keeper" photos the "work" of keywording is likely duplicated in the caption or title (although I often neglect those bits of metadata even in keepers). The problem is, though, I tend to have dozens of "decent" pictures which really have no hope of getting a descriptive comment, and so will only show up in searches if I use keywords. Worse, I often find I have a picture which just plain doesn't work as an artistic photograph, but which ends up being needed later on as a snapshot of that person (likely heavily cropped or edited).
For me, with my rather limited list of subjects, I find that a few sets of keyword buttons allows easy keywording of all my common subjects (family members). It typically takes no more than about two minutes of my workflow to zip through a session of 100-250 raw photos assigning subject keywords to them at the outset. I can't do the same with captions, which to me need to be legible, understandable English.
Still, though, for less frequent subjects (ex, in-laws and more remote family members), I just create groups which represent large numbers of people (ex, one keyword for my sister's family). The idea of keywording is to enhance search, and the idea of search is to get a manageable set of results which I can then eyeball. If there are a dozen pictures of a particular group and I'm looking for the one picture of one member of that group, I can do that with my eyes manually.
I don't understand how the keyword browser could become completely useless. Lightroom supports a keyword hierarchy, right? When you're not assigning keywords to a subjects, the whole "people" branch should collapse down to almost nothing.
In any case, I think this is a divergence point between "professional" photographers (specifically, those with a wide range of subjects / clients) and "amateur" photographers.
The arguement seems to be that captioning is more useful. However if you're putting the images in stock, the keywords become more important.
Generally I put the same words in both. If needed I can refine the search to Keywords only.
I still need to spend more time keywording though, as a lot of shoots have generic, rather than specific keywording. That time will of course be regained when searching for a specific image.
Re-question on Controlled Vocab link - I use it and quite like it. It's big, but it's quite impressive
I like LR's keyword implementation, for the most part, especially the fact that you can click-and-drag entire categories, with their child-keywords from one place to another, at any time, if you need to refine the "structure" of your system, and photos will get "re-labelled" dynamically. There are still a few quirks though (inability to do searches on synonym keywords in LR2-B, or for keywords that are hidden in collapsed stacks).
Perhaps I am biased by my background in linguistics, but I think that controlled vocabularies are a great solution for large, general-purpose libraries (whether it's books or stock images); they allow to preserve (i.e., they force) consistency throughout the library and among different people who may apply the classifications. At the same time, they are quite inadequate for personal libraries since they tend to over-specify the categories that are rarely or never used (in my case, "government" or "health" or "agriculture" would be unused, with ALL of their subdivisions), while at the same time leave under-specified those fields where many of us need more-than-general-detail because of personal or assignment/job-focused interest. For someone, the label "Kraków > Poland" will be sufficient; since I know that city so well, and have 100's of photos taken there, I'll need "Old Town," "Market Square," or "Old Jewish Quarter", etc. to make my photos easily accessible as I need them. I need lots of (sub)-categories for flowers, but at this point, just one for fish; I label photos by their color dominant (e.g., when they have a striking blue-yellow contrast), which is something many other people probably do not do... The beauty of LR in this respect is that you can add terms and place them in their appropriate spot in the classification structure, when they become needed.
I cross posted this on another of your posts:
Very informative and I look forward to the walk throughs. Mac games