home
contents
copyright
donations
contact


12.3 Filtering

Filtering images—i.e., deciding which ones to post-process and which to skip (or even delete)—is probably the most time-consuming aspect of digital bird photography.  Because digital cameras allow you to take many more photos than you would have been likely to take with a film camera, the tendency among many photographers nowadays (including both amateurs and pros) is to take far more photos than necessary.  The good thing is that this affords more opportunities to get an exceptional photo—e.g., a chance capture of the bird very briefly assuming a novel pose.  The bad thing is that you then have to sift through massive numbers of image files later on, to find the ones worth keeping and/or publishing.  In many cases, that process of sifting through images can take far more time than it took to actually capture the images out in the field.  In this section we’ll discuss methods for efficiently filtering large numbers of image files.
    The first issue we’ll focus on is learning how to decide which images to keep and which to discard.  This is mainly a problem for beginners.  The key is to be very picky when choosing which photos to
publish (i.e., which ones to post on your web page).  A wise man once said: An amateur photographer is one who shows all of his photos—even the not-so-good-ones—to the world.  A professional photographer is simply an amateur who has learned to show the world only the few photos that turned out exceptionally well.1  This has some truth to it.  If after taking 10,000 bird photos you were to pick out the best three and show those to a random person, it’s reasonably likely they would be impressed with those three photos—especially if they got the impression that you had taken only three photos in total.  The point is that the vast majority of bird photos that most people take are dull and uninteresting to other people.  By taking enormous numbers of photos and keeping only the best 1%, you’ll begin to see how a successful photographer can build up a portfolio of stunningly impressive images.  The trap that many beginners fall into is that they get impatient and want to keep 75% of their photos.  Of course there’s nothing wrong with keeping all of your photos, but if you plan to share your art with the world, it’s better to hide the 99% that are run-of-the-mill images and reveal only the 1% that are remarkable.  That’s where filtering comes into play.
    Before you can perform efficient filtering of large numbers of photos, you need to find a convenient way to quickly view images.  If Adobe Camera Raw (ACR) loads quickly on your computer, then you can simply use ACR to preview images.  On some computers, however, the time it takes ACR to load your image after you double-click the filename in the file browser is simply too long.  On some operating systems, the file browser provides its own built-in preview functionality.  The figure below shows one mode of the Finder utility in Mac OS X version 10.6 (so-called Snow Leopard), in which previews of multiple files are shown in a 3-D lineup.  By resizing the Finder window you can enlarge the preview thumbnails so as to provide more detail.  In many cases, this preview will be large enough to allow you to decide very quickly whether a photo is even worth opening in ACR.



Fig. 12.3.1: Efficient filtering of vast numbers of image files requires some
way to quickly preview each image.  The Finder in Max OS X provides a
quick preview of each image; though the preview is small, it’s enough to
assess overall scene composition and exposure.  Images that pass that test
can then be opened in the Preview utility of OS X for a closer look.  Similar
utilities are available in Microsoft Windows and Linux.

    When the tiny thumbnail isn’t informative enough, you can sometimes get a quick preview of a file by opening it in some lightweight utility program rather than opening it in ACR.  In Mac OS X you can open image files (even RAW files, for supported cameras) in the built-in Preview program that comes with Mac OS X.  This program loads quickly, so that on some computers it is more efficient to open files in Preview first, rather than in ACR.  For those images that look promising in Preview, you can then close them and re-open them in ACR.  The Preview program in Mac OS X also allows some image manipulation, which is useful inasmuch as the manipulations that it performs are very fast.  Thus, if you open a photo in Preview and it appears blurry, you can quickly discover how sharpenable the image is by adjusting the Sharpness slider in Preview.  This technique will give you only a rough indication of the sharpenability of the photo, but with practice you can develop some intuition for which photos seem to be sharpenable enough in Preview to warrant opening them in ACR and proceeding with full post-processing. 
    To reiterate, the point of all this is to quickly identify those images that do not have the potential to become great images, so that you can quickly discard those files and move on to more promising ones.  It’s all about managing your time and avoiding wasted effort.  As you get practice with large-scale filtering, you’ll naturally fall into a rhythm that allows you to work most efficiently on your particular computer system.



Fig. 12.3.2: Assessing image sharpenability in the Preview utility of
Mac OS X (similar utilities are available for other operating systems).
This utility can open RAW files very quickly—sometimes more quickly
than Photoshop.  If the image looks good in Preview, it’s then worthwhile
opening it in ACR/Photoshop for full post-processing.  Note that if ACR
opens quickly on your computer, you can skip the Preview step.

    Note that when we say that filtering is about discarding images, we don’t necessarily mean deleting those images.  For images that are clearly of no use whatsoever—e.g., that are totally blurred, or massively underexposed or overexposed—you may indeed want to simply delete them during filtering.  On Mac OS X systems, deleting a file in the Finder can be done instantaneously by pressing the key combination Cmd-delete, so deleting files needn’t take any more time than simply skipping them.  For images that aren’t defective per se, but appear rather unexciting, you may instead want to skip them rather than actually deleting them from your hard drive, in case you later find a use for them.  I rarely delete files anymore, since the rate at which hard drives have gotten larger over the years has kept pace with my rate of taking photos (and the rate at which the file sizes have increased, due to the megapixel race).  However, the good thing about deleting files that are clearly useless is that later searches through your archive can proceed more quickly, since there will be fewer files to examine.  Note that most operating systems nowadays have a trash can mechanism, so that after deleting a bunch of images you can go back and make sure you really wanted to delete them before emptying the trash can (which permanently deletes the files).
    The way I like to filter a directory of images is to create three subdirectories, called YES, MAYBE, and NO.  As I preview the images, I note which ones I definitely want to publish, and I drag those into the YES folder (see the figure below).  For those that are clearly of no use to me, I drag them into the NO folder.  All others I drag into the MAYBE folder. 




Fig. 12.3.3: Filtering images via drag-and-drop.  I classify my images
into quality categories (best to worst) and drag them from the main
directory to an appropriate subdirectory.  When all images have
been dragged to a subdirectory, I then review the highest quality
category and possibly subdivide those as well.  The goal is to pull
out the
best of the best and publish only those.

    When deciding which files to drag into the YES versus MAYBE or NO folders, I apply what I call the BBLP rule, where BBLP stands for Beautiful Bird, Lousy Photo.  Whenever I see an image of mine that depicts a bird that is clearly a beautiful specimen, but that has been poorly captured in the image, I say that it is a beautiful bird, but a lousy photo.  Many beginners have difficulty skipping over photos that depict what they know are beautiful birds, but that have been captured by a truly horrendous photo (whether due to blurring, poor exposure, or any other technical or aesthetic shortcoming).  For some people this can be a severe mental block, because they fail to acknowledge that a photo can be lousy even if the bird that was photographed is a truly beautiful specimen.  They may even feel that by deleting the photo they’re somehow offending the bird.  This is where the mantra of BBLP comes into play.  Telling yourself that the bird is beautiful but the photo is lousy may allow you to more easily detach yourself emotionally from some of your more mediocre images, so that you can focus more fully on the more promising images in your collection.


Fig. 12.3.4: My classification schema for image filtering.
I quickly classify each image into the YES, MAYBE, or NO
categories.  The YES photos will get uploaded to my web site;
all the others will remain unused.

    After performing an initial pass to classify all files as YES, MAYBE, or NO, the main directory will then be devoid of image files.  At this point I then look to see how many files are in the YES subdirectory, and ask myself whether I’ve got too many or too few YESes.  If I think there are too many YES files, I’ll then re-examine each and every file in the YES subdirectory and painstakingly find the least promising among them, which I then drag from the YES subdirectory into the NO subdirectory (or perhaps into MAYBE instead).  Conversely, if I realize that the YES subdirectory contains fewer images than I’d like, I may decide to re-analyze the files in the MAYBE subdirectory and drag the very best of those into the YES subdirectory.  I’m also careful to make sure that no two images in my YES bin look too similar to each other; if I’ve got many excellent images of a bird in the same pose or very similar poses on the same perch, I’ll force myself to pick only one.



Fig. 12.3.5: After the initial filtering step, you can then revisit some
of the decisions you’ve made by dragging some of the best MAYBE
photos into YES, and/or dragging some of the more questionable
YES photos into MAYBE.  These decisions are often very difficult
to make, so it’s good to make a second pass through the entire set,
preferably after resting your eyes for several hours, to see if you’ve
changed your mind about any of the images.

   Once I’m satisfied that the images in the YES subdirectory are indeed those that I want to publish from this set, I move the entire contents of the YES subdirectory back up to the parent directory, and then consolidate the MAYBE and NO sets into a single subdirectory called UNUSED.  In this way, I still have access to the unused images from each set, in case they prove useful at a later date (such as for documentation purposes).  The YES photos I’ll typically upload to a single photo album or blog on my web site.
    Note that filtering can be applied either to your RAW archive or to your collection of post-processed JPG images.  Indeed, there is good reason to apply filtering to both.  During post-processing, I generally perform a simpler form of filtering on the RAW images: I simply skip through the files in each each directory until I find one that piques my interest.  I then post-process that image into a JPG file, which goes into my (separate) JPG archive.  Once I’ve gone through all the RAW files in a given directory, I then perform a separate pass of filtering on the post-processed JPG files, using the YES/MAYBE/NO protocol described above.  The YES images then get uploaded to a single page on my web site.  If a client then finds an image on my web site that she or he wants to re-publish in another medium, s/he can simply send me the URL (internet address) of that photo from my web site.  I then track that JPG back to its RAW file and then re-process the image to a higher-resolution JPG or TIFF which I can then deliver to my client.
    Filtering mountains of image data is a tiring process.  Keep in mind that many of the post-processing decisions you make (including your filtering decisions) can change as your eyes get tired, your attention span shortens, or the lighting in your work space changes.  During subsequent post-processing sessions I often double-check images that I’ve already post-processed, and often I’ll find that I feel differently today about an image that I post-processed (or filtered) yesterday.  This is a natural part of the artistic process.  Just remember that post-processing images can be at least as difficult as capturing them in the field, and deserves at least as much effort and attention if you want to achieve your full potential as a digital artist.



1 Liberally adapted from a quote by Ryszard Sytnik.