Thripp Photography

Thripp Photography

Main menu

Skip to primary content
Skip to secondary content
  • Home
  • About Richard
  • Daytona Homelessness
  • My Newsletter
  • Old Gallery
  • Photography Portfolio
  • Search
  • Search Tips
  • Site Index
  • Tweet This Blogs
  • Tweet This Custom URL Services
  • Tweet This Icons
  • Tweet This Polls
  • Tweet This Press
  • Tweet This Version History
  • Tweet This, a WordPress Plugin for Twitter [1.8.3]

Tag Archives: metadata

How to Break into Stock Photography

Posted on 2008-06-23 by richardxthripp

The five chapters in your adventure:

1. an introduction to stock photography
2. taking the photos
3. nitty-gritty editing
4. how to pitch a model release
5. building effective keywords

— 1: an introduction to stock photography —

Stock photography is not art photography. If you’re looking to express your creative spirit while making a comfortable living, this is not the place for you. You can do the latter with work, but not the former, because stock images are boring as salt.

Curiously, the best stock photos are interesting. Crafting a photo that is not boring yet appeals to advertisers is a lot harder than creating a whole bunch of boring photos and making it on volume. I don’t know how to do the interesting, successful ones, but they usually involve people shaking hands or flying kites at the beach. In this article, I’ll be introducing you to the technical details that will help you to create boring stock photos. Then you can move up to better ones later. If you don’t learn these basics, your great ones will look slightly imperfect and won’t sell (read: won’t be accepted by your microstock agency), which we can’t have.

Most people elbowing their way into the stock world start with the microstock networks, because they’re the only shot for an average Joe to make any sort of money. Ones like iStockphoto, Shutterstock, fotolia, and Big Stock Photo. These websites let up upload your photos, which they then sell to their customers, taking much of the profits but giving you a commission (something like twenty-five cents per sale). They’ll only take stuff they think will sell, and only if the image is “perfect”: grain-free and plastic looking, six or more megapixels, no artifacts (if you have a cheap camera, this is impossible), and other requirements. If you’ve tried going out on your own, you know how hard it is to get viewers, let alone customers (I have enough trouble giving my photos way), but you’ll get guaranteed traffic at these sites, just as you’ll get more visitors to your MySpace page than to your Geocities page (sorry for the outdated examples).

In microstock, you have to make it on volume, because you’re giving non-exclusive, royalty-free rights cheaply. The good thing is you have the potential to build a passive income, if people continue to buy your photos after you’ve posted them. If you’re fairly good, you can expect to make a few dollars a year per photo, which will snowball with your tireless work and hundreds of contributions.

— 2: taking the photos —

Stock agencies are really crazy about noise (a synonym for grain). They’ll reject photos for any sort of grain, even if it’s artistic. DSLRs produce less grain because of their larger sensors, but images, even in broad daylight, will be often be rejected if they’re done on digital compacts, because they have smaller sensors and produce lots of noise all the time.

The best you can do in-camera is to use the lowest ISO speed setting, as this has the lowest light sensitivity and produces the least grain. If your photos are blurry because your shutter speed is too slow, add more spotlights or get a tripod to keep the camera steady.

Enough about noise, onto your subjects. A good subject is people, making business deals, doing things, vacationing with family, surrounded by palm trees, at the beach, pretending to be happy, etc. Too many people focus just on still life, because they’re too shy to scout for models and ask them to sign the necessary forms. It isn’t that hard, though.

Take pictures of your family! You still have to get them to sign forms, but they may be a good start for you to build up your portfolio, and you have access to them all the time. For stock portraits, you want to use a fairly shallow depth-of-field: enough so that all the people are in focus, but not so much that the background is crisp. You can do this by putting your subjects an equal distance from your camera and using an aperture like F2.8, though this won’t work on digital compacts because their smaller sensors permanently put them in wide-depth-of-field. Don’t cut off arms or legs, because those might come in handy later.

While you normally want to keep your images open to manipulation, the best-selling stock images don’t do this at all. Take a look at the top fifty at Shutterstock, and you’ll see flower compositions like this, which will no doubt be dropped right in to advertisements or presentations. I don’t imagine any users will be whiting out the background or warping the perspective on one of the flowers. It also violates the rules by having an artistic rather than practical title, but images sometimes are the most salable. It’s hit and miss to figure out, so start off with the boring stuff and work your way up.

For normal stock images, you want your subject to be sharply in focus with a small aperture, on a solid white or black background, with studio lighting, and with loose cropping. Go rent a studio, or set up a white sheet, a white table, and some spotlights in your bedroom. Then, start taking uncreative yet useful pictures of generic objects like fruits (nice fruits), unidentifiable candies (turn those Skittles upside-down), staplers, thumbtacks, and other colorful objects. Make sure every step along the way you’re using the lowest ISO speed, a high F number (you’ll need bright lights), focusing on your subjects sharply, and leaving enough space around them for your users to crop or add text.

Soon, you can move up to more unusual objects. Then, go out into the world and take uniquely beautiful nature shots. If you have an unusual job, like working in a steel mill, or on a farm, or at the top of the Empire State building, you have an even better place to take valuable stock photos from, because you have a different angle to work with from the start. Stock photography does have a certain creative edge to it, of finding the most valuable image for your customers. Watch for scenes that represent concepts, like solitude, inspiration, or bravery, because if they hit, they’ll be your greatest photos ever.

— 3: nitty-gritty editing —

This is all geared toward Adobe Photoshop. The reason I talk so much about it is because it’s the editing software most people use, including myself. It’s $600, so most people pirate it (you can find it easily with an online search). All this stuff applies to other software too, though.

Many people say you shouldn’t edit your photos; you should let that up to your users, as they’ll want to manipulate the image the way they want, without your intervention. Phooey. You need to edit your photos, to remove grain, imperfections, add color and contrast, and possibly isolate your subject (make everything else white). Your users don’t want to spend two hours fixing your image; you have to do that yourself, even if it alienates a few of them. I could never sell this as a stock image, for example, but the edited version of Simplicity could make money (if it wasn’t free).

You have to do a darn good job at editing, though. I like Noiseware for noise removal; there’s a free version way down at the bottom here, which is still pretty good. You should do your noise removal first, as when you add contrast, the noise gets amplified, and it won’t remove smoothly after a lot of editing. I like to edit destructively, so my workflow is cloning and spot removal -> save a copy -> convert to 16-bit PhotoRGB -> noise removal -> curves and contrast improvements -> save another copy -> save a separate 8-bit JPEG. This means I need to estimate how aggressive I should be with noise removal, before adding the finishing touches. I do this intuitively from years of experience. You can use trial and error till you get it, or figure out how to use masks or layers or whatever the young ones do nowadays.

16-bit PhotoRGB is great because you’ll avoid rounding errors that cause color banding. It takes up lots of space, though. Expect to use 75MB with your two copies of each photo. It doesn’t matter though, because if you take 20,000 photos in a year, you’ll only be doing this for the 500 best ones, and that’s 38GB, which is cheaper than ever to store.

Back to the editing. You need to remove specks of dirt from your roses, or acne and blemishes from your models, or the black specks you know to be birds in your skies, even if you’re fundamentally against it. I do this on all my work, because my goal as a photographer is to present a realistic ideal of the world. This means my photos (should) look perfect, but believably so. You must do the same with your stock work. Use Photoshop’s spot healing brush for blemishes and dirt, but look closely so you’re doing a good job. Zoom in to 400% if you have to. Hit Ctrl + Shift + Z and try again if it turns out ugly, which it often will. If that fails, switch to the clone stamp to copy one part of the image to another. This is more tricky, but you can get it right with practice, by picking an area of similar shape and brightness, Alt + clicking, and then clicking over the bad part. After two years and 1000 photos, this’ll be second nature and you’ll be getting through a photo with hundreds of specks in fifteen minutes flat, and making it look believable.

After you get done in this stage, use the levels tool (Ctrl + L) to brighten and darken the image a lot. You can undo it right afterward. If you’ve done your job well, nothing will look suspicious. But in the beginning, you’ll get bad results all the time. Here’s an example. Say the floating antenna is all the rage now, so you dutifully use your spot healing brush to produce this:

aerial antenna with base edited out

It looks fine, right? You might not even zoom in past 25%, because you’re working on a huge ten megapixel file. Then you add your wonderful contrast, and something seems astray. You might chock it up to your imagination and post your photo anyway, but little did you know the problems you’d created:

aerial antenna with base edited out, turned ugly

Obviously, I skipped noise reduction here for effect. But nevertheless, this is what will happen if you don’t pay attention. Blue skies are tough to edit. Everyone who tells you you can reshape bodies and clear power lines out of skies easily has no idea what they’re talking about. It’s tough work, especially if you want to get it right.

So at this point, what do you do? Go back and try it again. For blue skies, use the cloning brush with a higher hardness setting; perhaps 90%. You can see the shortcomings of the healing brush right here; it makes everything into a blurry mess, not matching the surroundings. This image has been effectively ruined for stock use. In time, you’ll learn these limitations, and working around them will be a breeze. Even then, try to avoid spot-editing by getting it right in-camera. The floating antenna wasn’t such a hot idea after all.

Even if your image looks fine on your calibrated, CRT monitor, it might look awful on your customers’ overly bright, uncalibrated LCD monitor, because these defects will be revealed. This is the reason to use the levels tool to over and under-expose the image: you can quickly and easily check for the invisible problems you missed.

When you edit for contrast, do whatever it takes to make the image the most appealing to your customers. This will most likely involve contrast adjustments with Curves (Ctrl + M), perhaps with a reduction in saturation (Ctrl + U). If you’re editing in an RGB color space, curves pushes the color channels as far as luminosity, which is not ideal. I’ll often reduce the saturation by 30% before using curves to counter-act this. Activate your histogram (Window > Histogram), and watch all four windows (brightness and the RGB channels). If any of them trail off to the right, you need to scale back because that channel is clipping. When the color clips, there is no detail, because it’s as saturated as it can be. While it can be artsy, for stock work, you must avoid this. The histogram in Photoshop is just as important as the histogram in your camera.

Try Image > Adjustments > Auto Contrast too. It stretches the brightness histogram across the gamut automatically, and often works quite well. Do not be afraid of automatic tools. (We were afraid of autofocus, remember?)

I recommend against sharpening, because it adds too many artifacts. If you need to sharpen, you might not be dealing with such a good image to start with. If you want to submit it anyway, you may have more luck scaling it down from twelve megapixels to six, as you’ll be more likely to escape auditing. This is wrong, because you’re actually giving a lower quality image, but it’s how the review process works.

— 4: how to pitch a model release —

If you’ve come here by Google, you’ve probably just read two dozen other articles about marketing your stock photos. They all tell you how afraid you have to be of being sued, that you have to take every possible precaution, you have to pay your model a big sum of money, give her a detailed explanation of the rights she’s forfeiting, act professionally, have three witnesses sign the form, check your model’s driving license in case she (or he, heaven forbid) is underage, and then make a copy of that driving license. Maybe a notary also has to be involved. Oh, and you must have your model release form reviewed by at least three lawyers before using it.

It’s all a bunch of nonsense. Just get a blanket stock release form and have your model sign it. Unless you’re shooting nudity, just ask her, “you’re not under 18, are you?” Most of the time, she’ll tell you she’s 28, proudly. I don’t know why the particular question makes women want to tell you their age so often. If she’s lying, it doesn’t matter because no one will find out.

Remember that your model release form is not to protect you; it is to protect your licensees. If you’re licensing your photos through middlemen, i.e. stock or microstock agencies, they want everything to be covered because they’re afraid of everything. It’s not to protect you, because it’s the stock agency that’s going to take the blame if one of their clients is sued. When you register with them, they’ll tell you all the liability is on your shoulders (legal protection), but this probably won’t happen even if they’re sued, because it would be bad publicity to go after a photographer. Get your models to sign a blanket release either way; it’s not too hard to do. Tell them what I wrote in this paragraph (in your own words), because it’s quite convincing.

In many of the United States’ states, you have to provide some sort of compensation to legitimize the waiver—a model release is a bonafide contract. A dollar is good if you’re feeling wealthy. I like to give six 4*6 prints of some of my best work (I carry them with me). Then I say they’re worth $1.95 each; they would be if I didn’t give them away. This puts the compensation I provide above $10, which seems like a good place to be, and obviously the prints don’t cost me nearly that (nor does a copy of a music CD cost $20 to produce). Here’s an example of a completed form.

Other than that, you’re good to go. Tell your model, jokingly, that she should read what she’s signing, because she’s signing all her rights away. Most people don’t even care. The 10% that raise a ruckus aren’t worth your time. I try to get a witness for signings, but if there’s none around I’ll just scribble down “N/A.” Then again, I’m not playing the stock game, so the big microstocks may not put up with that. Check with the stock agency you’re looking to work for. Many of them will provide their own model release forms for you to use, which are more convenient, though not personalized.

— 5: building effective keywords —

Keywords are important, because they are your customers prime way of finding your photos in the website’s search engine. You want to target as many relevant and unique keywords as you can, so you’ll appear to the most customers. This varies by agency, but in the backwards land of keywords, quantity often reigns over quality. Say your submitting an image like this:

The Fountain

When I started out keywording my images on my deviantART gallery, the only thing I’d be able to think of for this photo is “water” and “fountain.” I’ve since given up keywording (so boring), but if you’re wanting to sell stock, it’s a necessary evil that you will become better at over a long, arduous journey. So what can I think of for this photo now?

water, fountain, wishing, well, wells, wish, droplets, drops, raindrops, raindrop, liquid, park, day, outdoors, splash, splashes, summer, fun, coquina, rock, rocks, pattern, texture, yellow, aqua, clean, clear, fresh, refresh, refreshing, refreshment, cool, cold, clarity, frozen, movement, fast, float, floating, speed, stream, streaming, light, bright, geyser, geiser, gush, gushing, spring, springs, burst, flow, life, wet, h2o, drink, taste, nature, shine, shiny, reflections, spatter, splatter, purity, pure

Yeah, that’s it, spam the heck out of them. Give them so many keywords, there’s no way your image won’t be seen by thousands of people, 1% of which will buy it. It’s the only way to do it.

If you’ve noticed, there is some stuff I left out. I could’ve used these:

tree, trees, forest, forests, sky, cloud, clouds, cloudy, sunshine, sun, bright, blue, skies, white, green, shade, dark, shady, grass, vivid

The reason to leave those keywords out is because they’re all describing background elements; the stuff that’s out-of-focus behind the fountain. It’s of no interest to searchers, because they can’t use the photo for those elements anyway (they’re blurry and obstructed). You never want to keyword like that, even if the off-topic stuff takes up a large section of the photo, because it’s irrelevant and of no use to your customers. Recognizing this and developing keywords, like all things in stock photography, takes time and practice. You can even pay $3 an image to have a person assign photos to your keywords (I also offer this service upon special request :sunglasses: ). However, the point of breaking into stock photography is to make money, not to waste money, so you need to do your keywords yourself, because it’s a skill that will serve you well; I’m using them even in the tags for this blog post (editing, guides, metadata, model releases, passive income, photoshop, stock photography).

Stock photography is hard work like all else, but if you enjoy it, I can see it becoming painless and rewarding. Plus, it’s nice not to have a traditional job for once. Have fun out there.

Posted in Photography Articles | Tagged editing, guides, metadata, model releases, passive income, photoshop, stock photography

How to give file names to your photos

Posted on 2008-06-08 by richardxthripp

This is a lengthy post (~4500 words). I cover file names in great detail, but go much further into the differences between a literal and abstract asset management system (descriptive file names vs. not), spend many paragraphs debunking time zones, daylight time, traditional date formatting, and use 500 words to debate underscores vs. hyphens vs. spaces to break up words in your web addresses. The implications go way beyond mere file names. Read on if you’re in for a adventure . . .

I don’t like that all the articles I read on organizing your photos recommend giving them descriptive file names. The problem with files and directories is that they’re just like their non-computerized equivalents: rigid and inflexible. Your photo cannot appear under “flowers” and “macros,” because a file can only belong to one folder. Similarly, it can only have one file name, and if you fill that with keywords so you can use the Windows search to find it, the name becomes long and unwieldy. Plus, if you take a lot of photos (I’m averaging 500 a month), it’s totally impractical.

Why is it impractical? Because you’re restricted in length and taxonomy, there are no connections between files besides rigid folders and rudimentary keyword searches, and you’re adding metadata in a bad place, because the file name should be the unique and persistent identifier for the image. If you want to change all your pictures of “cars” to “automobiles,” you’re in trouble. Every time your taxonomy scheme changes, you have to change dozens of file names. This is fine if you’re the average, uninformed user: you have one copy of each photo in “My Pictures” on your hard drive, and that’s all that exists. But even then, unless you’re use batch renaming software like 1-4a rename (the biggest kludge ever), it’s still tedious to change your terminology. If you’re an informed, smarter user, you’re backing your photos up to the closest thing we have to an archival media: recordable CDs or DVDs. And then, you have to deal with all sorts of tracking issues when you change your file names, because the names of the files on the discs you’ve already burned are etched in stone.

The ideal file name provides no information whatsoever about it’s content. I’m all for ideals, but this is just impractical. No info about the file is like DSCF0001.jpg. A serial number, sort of. But really, what we need to use is some constant, timeless fact about the photo. The date and time. If you’ve set your camera properly, it’s recorded in the Exif metadata of each file (or for RAW, somewhere like it), so we can just pull it out and use it as the file name. But first, the other options . . .

Maybe you think you can be smart—you’ll just combine it all to get the best of both worlds. Create folders by date, make the start of the file name the date. Then, give a whole bunch of info about the file after that. Perhaps you’ll even pioneer a controlled vocabulary system so you have a taxonomy you feel you can stick with for the next few decades. I love this page: File Naming Conventions for Digitally-stored Images. It’s a classic example of this newbie mindset: that you can control everything within the Draconian structure of the file name, enumerate all the important details, pick a syntax to stand the test of time, fit it all in 64 characters, and apply the same processes to the hundreds of photos you’ll be taking every month. With this gauntlet, you could become a librarian and fill eight-hour work days, because there’s no end to the battle. As soon as you take a photo, five minutes of administrative decisions for your “catalog” come into play. And you don’t have a “catalog,” because if you had one, you wouldn’t be using file names.

For pre-computer libraries, we had card catalogs. We can’t move books about on a shelf like we can records in a computer database, so it makes sense to have the metadata stored separate from the items being cataloged. Film photographers devised similar (though hopefully less complex) methods to track their negatives. With a card catalog, if you want to add a new subject, you create new records for all the items under that subject, even though they already have other records (cards). If an author has a pseudonym, he gets a separate card. Card catalogs waste space, are redundant, hard to use, and torturous to update. Using file names to organize your photos is no different.

One principle that I hold, is that it is easier to have one way to do things, than several divergent methods. The same applies to organizing your photos, and I suggest you simplify your life by doing the same everywhere.

What are these things? First, write the date as YYYY-MM-DD, always padded with zeroes if needed, and never omitting the century. The time is HH:MM:SS with a 24-hour clock, always padded with zeroes. When written together, separate the date and time with the letter “T”. So the time now is 2008-06-08T18:30:22. This is ISO 8601 date formatting, and it will serve you well for all your papers and computerized files. The reasons: first, it’s unambiguous across regions. In the U.S.A. we often do MM-DD-YY (completely senseless), and in the U.K. it’s DD-MM-YY (better, but still senseless). When you put the date in descending order, with the full year, everyone understands it. No one’s going to think it’s YYYY-DD-MM… because that would be stupid.

Also, we have this crazy predilection for dividing the day into two halves, one “A.M.,” and the second, “P.M.,” which stands for something in Latin. I don’t even want to know how this came about. It’s just pointless. Don’t even think of naming your files with that format. They won’t even sort properly. The correct form of the date (descending order, each field padded with zeroes), sorts chronologically in even the oldest, simplest computer systems, because they’re all based on lexicographical order.

Next, give up time zones. Even if we’re looking just at your file names, why would you use a time zone? What are you going to do when you begin your great photographic adventure, crossing time zones left and right? Are you going to change your camera’s clock each time? What happens when you get home, and you want to show a chronological slideshow of your journey to your family? But then, you have photos you snapped at 3:30, and then you have photos that you snapped at 3:00, but really they were snapped a half-hour later than the ones from 3:30, because you crossed into a new time zone. Do you want to have to work out that mess every time you travel the world?

“But Richard,” you might say. “Why can’t I just use my home time zone all the time, and put that in my file name?” Sure, you can do that. But then you’re holding an arbitrary time zone while in another, arbitrary time zone. What if you live in California, but you vacation to Florida, keeping your California time zone on your camera? Then a year later you move to Florida, and switch the time zone for all photos from then on. Then you go back to California to visit your family, but now, you’re keeping the Florida time zone. Then 50 years later your grandkids are trying to decipher the mess you’ve left them. Or ten years later, you want the exact time you shot a photo, but you can’t even remember.

Even worse than time zones, is Daylight Saving Time. Just looking back in Florida (my state). In 2006, it started in April and ended in October. In 2007, it was changed to start three weeks earlier and end one week later. In 1974, it started on January 6. And try to figure out the rules in any other state or city, and you’re in for headache. In Indiana, there is no daylight time till 2006, and now they’re on board with us. Some cities followed daylight time before, though. Daylight time isn’t enough; many counties and even half the state have switched back and forth between Eastern and Central time. There’s a whole article on Wikipedia about it: Time in Indiana. Can you tell me they’d need that if they followed rational time all the time (UTC)? We even have a year 2007 problem due to the U.S.A.’s recent changes to daylight time. I don’t remember where my government got the power to declare to me what time it is.

This is a mess, a terrible mess. It’s so bad, the Zoneinfo database defines rules by city, because it can’t even assume the rules will be consistent across a particular region. Reading the Wikipedia entry… “The database attempts to record historical time zones and all civil changes since 1970, the Unix time epoch.” It attempts? We aren’t even daring enough to claim that it has all the rules for the past 38 years? In 2050, maybe we will be attempting to remember all the rules going back to 2012? Do you really want to be poisoning your photos’ metadata with this nonsense?

Daylight time can even kill you. This report, titled FDA Preliminary Public Health Notification: Unpredictable Events in Medical Equipment due to New Daylight Saving Time Change, warns about the changes to daylight time that began in the U.S.A. in 2007:

If a medical device or medical device network is adversely affected by the new DST date changes, a patient treatment or diagnostic result could be:

* incorrectly prescribed
* provided at the wrong time
* missed
* given more than once
* given for longer or shorter durations than intended
* incorrectly recorded

A treatment could be missed or given more than once? And we’re doing all this to “save” energy? In reality, it leeches from everyone. Computers, DVRs, VCRs, and countless other devices need to be needlessly complex to accommodate these idiotic rules. People have to change clocks—dozens of them—twice a year (everything has a clock now). And I’m sure the missed appointments number into the thousands. It messes up all e-commerce systems too. All the sudden you have one hour that’s skipped in the spring (2 to 3 A.M.), or duplicated in the winter (2 to 1 A.M.). You don’t have accurate statistics for that day or hour, and any order tracking software that’s worth its salt has to include extra code to work with the phenomenon without exploding.

The entire concept of time-zoning is evil and pointless. Daylight Saving Time is even worse. We must forsake both in a unilateral shift of thought. You can be the catalyst. The beast must be slain. Of course, you can’t give it up for appointments, or at your job (unless you’re a free-thinker and self-employed), because everyone else is using a convoluted system. But like Ghandi said, “Be the change you want to see in the world.” So do it wherever you can.

When you switch to better date formats and smarter time, you’ll fix all sorts of issues on your computer. Now you can just prefix your documents with the date (YYYYMMDD), and they’ll sort in order because the date is in the right order. Same for the time, with the better 24-hour system. I gave up daylight time and local time on 2007-01-01, as my New Year’s resolution. Before that, I’d have lots of issues with file synchronization, because the modern NTFS file system is pretty smart about time zones, and uses UTC all the time just like you’ll be doing. Your time is just an offset from it, and when your computer’s clock switches for daylight time, all the files change (even if they shouldn’t and don’t need to). So then, if you’re backing up to other devices, especially ones using old file systems like USB flash drives, all the time stamps are wrong. You’re synchronization software (I use SyncBackSE) will re-copy all the files thinking they’ve changed, even tens of gigabytes of them. There’s some option to override that in the newer versions, and most other software has hopefully provided for the problem as well. But it doesn’t need to be a problem. If you’d give up the silly notions of changing time that you’ve been brain-washed into by the state, you won’t ever need to deal with this problem again.

You do have to go all the way and give up time zones while you’re at it. My clock says 20:00 right now, but the “real time” is 16:00 here (or 4 P.M. if you’d prefer). If I need to make the adjustment for friends or relatives that are still stuck in their old ways, it’s a snap to do in my head. Then, if my friends use my computer, they get to be confused and ask funny questions like “why would anyone do this?” Before, I’d just say “nevermind, I could never make you understand it.” Now I can, with this article. You can too. Don’t let anyone sway you in your decision to rebel against time zones. Time zones are for average people. They mess up online communication, they hinder commerce, they confuse travelers. “Noon” should be completely arbitrary. It isn’t even when the sun is at the top of the sky now, with the daylight time problem and our expansive time zones. Do you think that with a time zone that’s hundreds of miles wide, that “true” noon is going to be at the same time in all of them? To do that, we’d have to have time zones by the second, that change every few feet. We don’t even have a good system now, so whatever we move to has to be better.

What’s my final recommendation? Switch your camera and computer to Universal Coordinated Time (four hours ahead of North American Eastern Daylight Time). Abandon “logical” file names. There was nothing logical about them after all. Keep the Gregorian calendar. Use the date for your name; don’t go for completely arbitrary names (like DSCF0001). Format it as YYYYMMDD-HHMMSS@#. The number counts from 1 if there are multiple photos in one second (burst shooting). The @ is your initials. Not your name, just two or three letters in case you get your family on board for naming files like this, and you need basic tracking. You can name your photos with their titles, if it’s small scale like for an email or a website. Never do it in your main archive, because you’ll have a lot of photos there and you want consistency. If you’re any sort of photographer, then you will have a lot of photos there. 100 a day perhaps. 20 a day if you’re like me and cull a bit more. It’s still a lot, so you need a good system. Put the files into folders by month, like 200806.

The whole reason to name your files like this, is because it can be easily automated. You can have a computer program apply your file-name format to all your photos as you save them to your computer. I recommend and use Downloader Pro for this, because it’s flexible but not horribly complicated. My file-name string is {Y}{m}{D}-{H}{M}{S}rxt{l} . Then, after downloading and backing up my photos (which Downloader Pro can do too), I delete them from the memory card immediately. Oh, and when I plug in my memory card, the software starts up and copies the photos automatically. Very nice and swift. It’s $30 though.

If you want something free, Digital Image Mover looks good. If I used it, my string would be %Y%M%D-%H%m%Srxt . There’s no counter that only comes into play if there’s a name clash, so if you have two photos in the same second, it won’t work for you. You can use %c, but then you’ll have a number on every file. What I’d do is just rename the problem files on my own, because it doesn’t come up often (unless you burst shoot and don’t do your deletions in-camera). Be sure to check “Use EXIF ‘DateTimeOriginal’ field” and “Verify files after copying” under Options. You should always use Exif time instead of the file time-stamp, which can be seconds later (because it’s from when the camera wrote the file to the memory card, not exactly when it was snapped). Verifying checks the source against the new copy twice. It’s slower, but it’s insurance against corruption.

The third option is to just move your files over in Windows Explorer, and then use IrfanView to rename them. I used to do this. It works well, though you have to do it each time you import your photos. Go into File > Batch Conversion/Rename, and specify a file name like this: $E36867(%Y%m%d-%H%M%S)rxt . Replace “rxt” with a two or three-letter name for yourself. You’ll have to deal with file name collisions in the same second on your own.

Keep in mind that Exif metadata is only a property of JPEG files, not RAW, and you should be shooting in RAW if you have it. I know Downloader Pro gets the date in whatever proprietary format my Canon Rebel XTi formats it to, but I’m not sure about the other programs.

Anyway, it’s important you switch file name schemas now; change gears completely. Whatever you did in the past is past. Gone, zilch, nada. Don’t change old stuff, don’t update, just switch to a smart naming scheme now, because this is something you shouldn’t have to deal with anymore. Same if you’re on board to abandon traditional date formatting and time zones in your daily life. You shouldn’t have to wait until you catalog your photos to being archiving them to discs or sharing them with friends. But that’s what you have to do if you use “reasonable” file names.

Let me make a note here that you should never use spaces or capital letters in your file names. When you post them to the Internet, that causes trouble. URIs aren’t supposed to have spaces; read this for a good discussion of it: spaces in file names. In the UNIX world, which your web server will probably be using a derivative of, all file names are case sensitive. I know, I don’t like it either. But it’s better just to stick to lowercase so if anyone writes down a web address from your website, they don’t get mixed up. Sure, when you need to post your photos online, you could just change the file names. But it’s so much better to get it right to start with.

I remember posting an album to Photobucket to share with friends… and I found that since I’d used capital letters in the album name, it wouldn’t work when you typed it in lowercase. Not at all. Photobucket considers “Photos” and “photos” to be completely different. You can even have both albums and put different photos in them. This is nothing like Microsoft Windows (which is case insensitive except for display), and it’s confusing for everyone. So just stick with lowercase.

For separating words in file names, don’t use spaces ever. Hyphens or underscores are your best bets. I like hyphens, because if you have something like my_photos, and it’s underlined, it shows as my_photos. The underscore looks like a space. And it will be underlined, if it’s a part of a URI on your future website. And it will confuse your readers, because they’ll be sending the URI about by email when your site gets popular… and then someone will write it down on paper to share with a friend… and they’ll use a space instead of your beloved underscore… and then the world will end, because someone won’t be able to look at the photo or article that will change their life, all because you used underscores instead of hyphens in your URIs.

Historically, Google has treated underscores as part of a word, but hyphens as spaces. This meant that if you used underscores, you’d rank much lower in Google searches in general. This changed a year back; read Underscores are now word separators, proclaims Google. But still keep using hyphens, to prevent the issue with underlining. Also, they look nicer and are easier to type (no holding down Shift). You may be asking: why are you talking so much about the Internet, if this is an article about the file names of photos on my computer, which aren’t even on the Internet? The reason is that if you ever post your photos online, the file names will become an issue. It’s always easier to do things one way, as I talked about several paragraphs up. Your computer keyboard doesn’t do things one way, and that is why the Caps and Num lock keys cause so much trouble.

How should you implement your catalog? Definitely not by file name; that’s the lowest form of cataloging. It hardly even deserves to fall under the umbrella term of “digital asset management.” But of course you know that (or are fighting it) if you’ve read this far. Don’t do it directly with IPTC metadata either; that’s barely a step up from file names. Use an abstracted database. Browse your photos in that database. I like IMatch, because it’s scalable, terribly complicated yet still easy to use, and it decouples your photos from your files. Basically, the assumption that you should have the files on the computer to browse them is gone. It was never a good assumption anyway. You can browse your photos by thumbnail and metadata with just the database (~3% the size of the files), and they’re in “offline” mode; but you’ll be fine for copying or manipulating them once you bring them online (i.e. turning on your external hard drive, or inserting the CD the photos are on), and nothing blows up or disappears. You’ll wonder how you got by with the old way, where the photos had to be there to be browsed or cataloged, even if they took up 250GB (like mine are approaching). And of course, photos can be assigned tags, called “categories,” though of the flexible kind. A photo can belong to any number of hierarchal categories, or none at all, or child categories but not parent categories, or anything you can think of, without wasting space. The categories aren’t even like folders, because when you click a high-level one, and it shows its images plus all the images in sub-categories, which is much better than clicking through dozens of folders in Windows Explorer. But of course, you can change this if you want.

IDimager is my second choice, and seems to do all the above. ACDSee is third, though I haven’t used it. If you switch between software, you can export your work by embedding it into the IPTC metadata, and then importing that in another program. IMatch provides a bunch of other methods, and the others probably do, too. The problem with IPTC metadata is that it puts your files out of sync with the backup copies you’re making to CDs or DVDs, because they can’t be updated. I don’t like that. Think of the files as books on a shelf, static and permanent. Your database is your card catalog (on steroids). Using file names or inline metadata is like going to each book on the shelf, and putting a whole bunch of keywords on its spine label. Senseless and impractical. We wouldn’t even try that sort of thing in the stone age.

So, avoid syncing IPTC data with the database, unless you don’t mind your archival copies being out of date. I mind it. I don’t use IPTC, even as a backup; I just backup my database instead.

The three programs I mentioned are all about $60, quite a hefty investment. But less than your camera, and certainly worth it if you value your time. I see people spending more time naming their pictures, than actually taking them, and I don’t want you to fall into that trap. You can easily fall into that trap with a dedicated cataloging application. But accomplishing as much as you did with file names takes a small fraction of the work, and that’s the important thing.

Don’t touch Picasa with a ten foot pole. I just tried v2.7, and it still has the same horrid behavior it had when I attempted to use it in 2005. What did I do? I indexed a folder on my USB flash drive. Then I closed it, removed the drive, and started it again. What happens? All the photos disappear. Oh wow, Picasa. You’re so smart. Not. They reappear with my captions when I plug the drive back in (as long as it has the same drive letter). But Picasa shouldn’t pretend to be a photo management solution, with captioning and descriptions, if it’s all smoke and mirrors. And it is. Read this: Dear Google: Picasa Needs Improvements, particularly the “Database Export/Import” section. Picasa makes it really hard to move the keywords and descriptions you’ve assigned between computers. Some settings are strewn through the file system in “picasa.ini” files, others are in a proprietary database, which is dumb and sloppy. They should all be in the database, and the database should be easy to transfer, backup, and export to other formats, if Picasa wants to be a real cataloging application. But it’s not, and if you use it, don’t treat it as such. Oh, and the keywording interface is awful.

Abandon the old and start with your new abstract file-names immediately, even if you haven’t thought up how to catalog your work. With 2000 photos, it’s not too hard to find them by thumbnail, and you’ll have more time to shoot when you end the tedious process of assigning descriptive file names. If you use Picasa, don’t touch the functions for keywords or captions. I don’t even recommend the program for browsing; FastStone Image Viewer is simple, free, doesn’t pretend to solve digital asset management, loads thumbnails quickly, has a great slideshow mode, does batch converting and file renaming well, and can be run from a flash drive.

Always, when you’re cataloging your photos, catalog what you will search for. I do this with the photos in my gallery; you can see a list of tags below each one, and they link to similar photos. There aren’t many tags. Just simple stuff like flowers, raindrops, dark, bright, shallow dof, etc. See, if I wanted to be thorough, I could record where each photo was taken, add the hour, year, aperture, etc., each as a tag, use scientific classification for plants and animals, and other terrible stuff. Newbies to cataloging get caught in this trap. It ain’t worth it—you’ll never need it. And if you do need it once in a blue moon, it’s still a waste of time. The time to reward ratio just doesn’t match up at all. It’s all pain and no gain.

Don’t be too dedicated to tagging your photos, whatever method you choose. I follow all the stuff I recommend for file names here, which is done automatically with Downloader Pro, but I haven’t cataloged anything in six months, despite having my abstract database. It’s just not needed when I only use 4% of my photos for my website or exhibition; I can keep track of those fine, just browsing by thumbnails. I’d rather be out shooting.

Posted in Library Science, Photography Articles | Tagged dam, exif, files, internet, librarianship, metadata, theory, time, urls

Recent Posts

  • Tippyfi, My New Website on Personal Finance
  • I Won’t Give Up, Wedding March, Clocks, Amazing Grace (Piano)
  • Newly Released 2012 Photos
  • Photo: Flowering Shrub
  • Photo: Barren Branches

Archives

  • February 2018 (1)
  • April 2013 (1)
  • October 2012 (1)
  • March 2012 (16)
  • September 2011 (1)
  • May 2011 (1)
  • April 2011 (7)
  • March 2011 (21)
  • February 2011 (18)
  • January 2011 (5)
  • December 2010 (26)
  • November 2010 (25)
  • October 2010 (16)
  • September 2010 (49)
  • August 2010 (42)
  • July 2010 (2)
  • January 2010 (1)
  • December 2009 (4)
  • November 2009 (15)
  • October 2009 (13)
  • September 2009 (34)
  • August 2009 (7)
  • July 2009 (7)
  • March 2009 (1)
  • February 2009 (4)
  • January 2009 (5)
  • November 2008 (20)
  • October 2008 (17)
  • September 2008 (19)
  • August 2008 (14)
  • July 2008 (50)
  • June 2008 (41)
  • May 2008 (35)
  • April 2008 (33)
  • March 2008 (26)
  • February 2008 (24)
  • January 2008 (24)
  • December 2007 (11)

Categories

  • Classical Piano (8)
  • Help (2)
  • Library Science (5)
  • Other (23)
  • Personal Development (88)
  • Photography (405)
  • Photography Articles (25)
  • Photography Ramblings (16)
  • Portraits (43)
  • Scholarly Essays (14)
  • School (13)
  • Shop (88)
    • Special (16)
  • Stock Photos (339)
  • Technology (48)

Meta

  • Log in
  • Entries feed
  • Comments feed
  • WordPress.org
Proudly powered by WordPress