I’ve written on the subject of digital photo processing workflows before, but I’ve changed my workflow a bit, and came across an excellent workflow article, so it’s time for an update.
Updates from Previous
My previous post had talked about an evolution of my workflow from the consumer side (just iPhoto) to something which worked well for me then (Aperture, exporting to iPhoto).
Essentially, what I ended up with there is still true today, although I have gotten less pedantic about exporting back into iPhoto (although I’ve been reminded of the importance of this step recently when Abby and Livvy wanted to find some photos to print for scrapbooks). I also don’t delete anything in Aperture. It seems to handle the massive library fairly well and I’d rather have an image around which didn’t turn out ideally than no image at all.
The main evolution of my workflow has been in the area I largely skipped last time: culling and keywording inside Aperture.
I’d invite you to visit Fraser Speirs’ Photo Editing Workflow and his later Project Organization post. He gives a lot more details than I’m going to give here, and what he is doing largely matches my own workflow inside Aperture (although he’s exporting to Flickr instead of iPhoto).
One place where we differ is our rating schemes. I rarely rate anything beyond 2 stars on the first pass; for me “no stars” maps to his “1 star” (technically OK), and “1 star” maps to his “2 stars” (needs processing). That leaves “2” and “3” as varying levels of ephemeral proud-love, before “4” and “5” essentially map back up to his ratings (“best of the shoot”, and “holy crap, I took that?”“, respectively).
Organizationally, I’m going to be trying a little more of his scheme. My scheme had been one top-level folder called “++ Recent Projects” (into which went … well, recent projects), another called “Archives” (all older projects), and albums as needed on my current projects. I kept things ordered in the two main list folders by naming all my projects like “2007-12-25 – Christmas Morning at Home”. The drawback to that approach versus Fraser’s is that it tends to show up on my relatively small screen as “2007-12-25 – Christm…” unless I give the project list far too much screen real estate, so I’ve tended to make the project list big and toggle it on/off as needed instead. Moving year and month to the folder level should buy a little space even after taking the additional indentation into account.
Similar to Fraser, I haven’t used “yellow folders” (folders inside projects) since my first fumblings through the Aperture interface when I found they didn’t act like I had wanted. However, I’m going to try using them for large sessions where Aperture is getting bogged down (~500+ pictures in a project and my G5 starts noticeably churning) as advised by the Inside Aperture blog.
Fraser hasn’t talked of his keyword system yet, but mine is simple enough to include here.
What it all boils down to is this: I keyword things which are easy to define and consistent, like people’s names, and which appear across large numbers of images, like often-visited locations perhaps, or different sports, and don’t keyword anything which can easily be filtered otherwise (date and event go into my folder structure; camera and other meta data are obviously already accounted for). Other incidentals about the picture, such as what was happening around the picture and a basic description of the picture itself, can go into the (still searchable) comments fields if they seem pertinent, or can be relegated to a visual scan.
You see, keywords to me are for searching. My most common tasks are “find that picture of Abby”, or “find a good recent picture with Livvy and Sophia together”. If I’m looking for birthday pictures, or Christmas pictures, I know to look at the photos from that date. Having a “Christmas” keyword wouldn’t help. I do, however, have “Portraits” as a keyword to tag more formal “portrait” images as opposed to informal candids which form the vast bulk of my library.
One feature I miss in Aperture is the ability to search for a lack of a keyword; I can for instance show all “Portraits” in the library, but not all “Candids”. I can show all “Abby” or all “Abby and Livvy”, but not “Abby but not Livvy”, or “Abby alone”. Perhaps in Aperture 2.0. There apparently is a way (or three) using the IPTC filters, but it’s not obvious and not ideal (no checkboxes, case sensitivity, substring matching…).
Who is in a photo is the fundamental search I tend to run, and that information is not there outside of a keywording system. So it all starts with this: each person gets a keyword. Extended family get keywords in groups (eg, “Grandparents: Dibble” would be my parents; “Chris & Cathy Family” would be my sister’s husband and brood), as they appear in a small enough set of pictures that I can find one of any specific person in the group from the screen pretty easily. If a particular person appears as an individual keyword and as a member of a group, pictures with them in it get both keywords. “Nature” as a keyword reflects something non-human and natural as a main character in the picture (eg, Abby holding a banana slug gets the “Nature” keyword along with the “Abby” keyword). That’s pretty much it.
What do I not keyword? Incidentals about the photo are for comments, not keywords. For instance, Abby in a red dress might get a comment of “Abby in a red dress”, but a keyword of “Abby” only. “Red” and “Dress” are not good keywords in my world. I know I lay in conflict with many other photo organizers here, but hear me out.
Keywording colors is silly. First, colors are a silly thing to keyword, as they jump right out of a large visual display anyway. If I had “Abby” and “Dress” as keywords I could bring up all pictures of Abby in a Dress and click through the red ones. Second, colors are subjective to some extent. While “fire engine red” pretty much always qualifies as “Red”, there are many other shades which border between colors, no matter how expansive your keywording scheme is. Keywording colors seems like trying to poorly imitate in code what our brains and eyes have evolved to do over millions of years. It’s a losing proposition.
Keywording other aspects of a photo can be fruitful, but, again, a concise and clear nomenclature is required. When your job is photographing weddings, you likely have some very specific things which crop up time and again (alter, wedding dress, bridesmaid, etc). But, when you’re photographing just “around town”, even if the “red dress” is a key part of the photograph, you’re not likely to get very many photos where “Dress” can be applied (and when searching you’ll need to consider “Skirt” and “Gown” and “Formalwear” as well as “Dress”).
I’m not a stock art photographer, nor am I interested in treating my photos as a personal stock art library. When I am looking for a picture, it is based on who was in it, what was happening, or where we were. I don’t care if it shows a Car or a Dress or a Zebra. This makes it somewhat hard those handful of times when I do want to find a particular photo of tigers at the SF Zoo, for instance, but a visual scan will suffice for me there too, so long as I can filter the view to animals at the zoo.
The key to keywording is that you have to balance the unwieldiness of the overall system (how many keywords do you have to sift through to find the one or two which apply here) and the likelihood of you absolutely needing a particular keyword. If you, unlike I, were a stock photographer, “Red” and “Dress” would probably be as important a set of keywords as “Abby” in my example above. But don’t force yourself into hours and hours of labor in a year just so you can save five minutes of eyeballing a stack of photos. It needs to be practical!
For individual projects I’ll often put together a whole new keywording system. For instance, for last year’s softball pictures (project was to put together a slideshow on DVD and support assembling individual scrapbooks for each of the girls, plus supply photos for the yearbook) I had one keyword for each of the girls on the team, another for “full team”, a “Pitching”, a “Catching”, a “Fielding”, a “Running”, and several others. When putting together the individual shots, I made sure to grab the best-rated pictures of each person, and to include every “position” they played that year, along with a picture of them with the coach. From the set of 12 individuals’ pictures, I picked the most representative pictures for the slideshow to fit the background music and themes and put those in another hand-picked album. The keywording scheme was critical in supporting this effort, allowing me to easily locate the “must have” pictures with these specific criteria across an entire season’s worth of photography.
In Aperture, I put together a few “Keyword Buttons” sets, one for People (all the kids, Jodi and myself are always on that one; in some sets of pictures I’ll adjust the last several buttons to include different people) and one or more for each project (for example, the Softball project had a Softball Subjects button set and a Softball Actions button set).
Which brings me (at long last) to post-production workflow.
When I’ve pulled out all the rejects and have differentiated 0, 1, and 2-star pics, I do a keywording pass on the photo set. If I’ve got a bit of time, I do all but rejected pictures; if not, I restrict the view to 1+ star pics. I start at the top picture, and click the button for each person who is in the picture (or each softball action, etc).
By “in” the picture, I mean that the person is recognizable and either a primary or secondary subject of the composition. The goal here is not cataloguing the owner of every body part intruding into the frame. Remember our end goal for these keywords is to find pictures “of” a particular person. Sometimes, as a matter of fact, the subject of a photo isn’t pictured in it. For instance, many pictures taken at events surrounding my father’s death are tagged with “Dad” although he appears in none of them.
When working a particular project, the definition of “in” a picture might vary. For instance, with the Softball Actions set, actions “in” the picture are those being carried out by one of the people “in” the picture. So, since I didn’t care about non-team-members, if the other team was pitching to our player batting, while out of focus in the background the coach is signing, the pic would get assigned to the batting player and the coach (always be aggressive in keywording; it is easier to mentally filter out a few extra pics in the smart album than to find the pics you hadn’t tagged in the full-library morass), and would get both Batting and Signing keywords.
The golden rule when applying keywords is this: when in doubt, add the keyword. It is better to end up with twice as many pictures in a smart album based on a keyword than to not have the one picture you were looking for in there and end up having to look through thousands of times as many photos to find it!
If I’m rushed on a particular day, I’ll skip tagging 0-star images. Instead, I’ll select them all (filter the view to include only unrated photos then command-A) then add the “Need Keywords” keyword to all. Then, periodically when I have time, I’ll bring up the “Need Keywords” smart album and apply keywords to them all.
All the pics which are 1 or more stars, then, get a pass at editing: sharpen the edges as needed, adjust the exposure, balance the whites, do a little red eye or spot healing on rare occasion. If I can’t get the picture to look the way I wanted it to look, it may well “lose” a star at this point (and get marked with “Needs Correction”).
Using the Results
After all this hard work, eventually we want to use the results.
One scenario is a specific project, such as the Softball Albums. There, I created a single “smart” album for each player, to pull all their photos in for review. With all the various “cuts” of the keywords possible, I chose this one because failing to properly fill the pictures for any girl would be far more noticable (and far more disappointing to one of our daughters’ team mates) than any other cut. In short: I paid the most attention to this cut because this was the critical cut of the project.
As the season progressed, I kept track of which shots were “missing” or “not great” for each player, and adjusted my attention during the games to try to capture those specific positions and shots (I have not found a magic way to be ready to take a picture of every one of the ten players on the field at a time, and am not good enough or fast enough to swing my camera up, adjust the zoom, compose the picture close to what I’ll want, and snap a picture in the time it takes a ball to travel from the crack of the bat to the mitt of the short stop). The last couple of games had some pretty specific assignments, and although I wasn’t able to get as many “great” shots as I’d have liked, I think the end results were at least satisfying.
Once the season was over, my work was straightforward, although still significant: assemble the photo albums for each girl, then filter approximately the same number of “keeper” shots from each girl’s album into a “slideshow” album. I knew approximately how many photos needed to be in here because I knew the song I’d picked for background (meaning, the length of time to fill) and the pace I wanted to keep on average for each photo (I wanted an aggressive, tight pacing of about 3 seconds per picture, beat-matched to the music). With about the right number of pictures in place, I then looked for some “semi-motion” stacks of keeper shots. By this, I mean 3-5 shots I’d taken in sequence (typically with the “continuous” shooting mode on my camera) which could be double-timed together to spice up the monotony. Typically, only one of the photos in a series showed up in the “keepers” albums, so I had to use a bit of memory and originals-diving to find the stacks in the main album (I can’t think of any way to have made this easier on me, unless I’d identified “series with a good shot” as a keyword from the start). The whole shebang got time-sorted, then dragged over into Final Cut (if I recall, there is a setting in FCP to say what the default duration of dropped images is, and I set that to the beat-matched duration). A little tweaking and re-matching of the beat (the beat was very close to a specific number of frames, but not quite; every ten or so pictures I’d have to have one frame less of video), and double-timing of the series, and I found I had a few more pictures to put in. So, back to the original albums, and I came up with a few more shots to put in sequence.
The end product turned out really nice, and I was able to put it together in only a few nights (most of which was spent matching the photos to the music in Final Cut, not shuffling through photos in Aperture).
In a somewhat related project, we wanted to make a yearbook dedication for our three girls, who have been in softball for several years. Finding “the best” pictures of them over the past four years was fairly straightforward, filtering the overall library for each of them in turn and the “Softball” keyword. We quickly picked a selection of pictures which all fit the theme of the dedication, made black-and-white versions of each, dropped them all into a Pages page for arrangement, and printed the result to a PDF file to send in. This whole process took about an hour.
A non-softball scenario was that one of our daughters wanted to put together a scrapbook, and to include pictures of her siblings. Finding pictures was a matter of setting up a smart album (across the whole library) with each person she wanted to include, filtered to 2 or more stars. I made a button set with just “Print Me!” in it, and taught her to click on the picture she wanted to include and then on the “Print Me!” This eliminated the likely source of library-modification had I asked her instead to drag them over to another album instead. When she’d gone through and found all the pictures she wanted, I went into a new smart album on just the “Print Me!” keyword, double-checked that she wanted each of them (a shift-click on the “Print Me!” button would take one out of the album), then sent the whole batch off to be printed.
As I said above, this last project made me nervous. Having anyone’s hands on my Aperture library, pushing buttons, makes me nervous beyond all measure. This is where I’d rather have had a “view-only” library over in iPhoto for her to look through. Scratch that: what I’d really like is a “kiosk” mode in Aperture where I could define the world of image manipulation allowed (in this case, just adding to a smart album and/or adding a keyword). Maybe in Aperture 2.0.
Wrapping it Up
It’s that time in the show when we come around and relate everything we just jabbered on about to the Real World. And that relationship is this: the end goal of all of this is not to catalogue photos, or to take better pictures, or to rationalize the purchase of a bigger, better, faster Mac next year. Oh, all of those are good ends in and of themselves, but they pale in comparison to the real deal. The real reason to do all this is to allow us, five, ten, thirty years down the road – God willing! – to look back at these quaint inscriptions of events and spark memories of what we’ve done and those we’ve loved.
The payoff is not today or tomorrow; we’ll feel the effects of our small efforts today for years to come, if only we do it right. The payoff is priceless.
That’s it, for now. Of course, this will all likely be obsolete six months down the road.
That’s how I closed the previous installment, and it’s likely to be just as true this time around. Still, it seems like my workflow has settled quite a bit in the past year, so maybe “obsolete in 12 months” would be a better prediction.