How to keep your bits from wandering off
Updated: 2/29/2011 names, RAIDs
A question came up at the recent Photograph Nebraska symposium I attended, and expanded to its full implications, it's a question that every photographer must answer: how do I minimize my risk of losing images?
The potential problems start in the field:
1. You lose the card.
2. You corrupt the card.
3. You write over the card.
Let's make sure we deal with #3 first, as this is completely under your control. First, you need to have File Numbering Sequence set to On in your camera. For some silly reason, NIkon has insisted on having this default to Off for many of their consumer cameras, and this leaves you vulnerable to downstream overwrites (because if you try to put another DSC_0001.JPG into a folder that already has one, you will overwrite the old image). It's long past time that the Japanese camera companies dealt with the 8.3 file name legacy, but apparently they think everything is hunky dory with the current system. It is not. File overwrites due to name duplication are as much the camera industry's fault as they are yours.
But there's more to keeping from writing over images than just making a camera setting. You need to have a disciplined routine for dealing with cards. You should have a clear way of identifying which cards have images on them and which are empty and clear for use. I use multiple tactics for this. When I transfer images to my computer, I format the card once I've verified that I've indeed gotten good copies. When I put a card into a camera, I press the Play button to see what's on it. I shouldn't see anything on the card. If I don't, I reformat in the camera once more and start shooting.
To help me keep track of which cards I've used and which are still empty in my card case, I use label in/out techniques: if I see the label, the card is empty. If I put a full card into the case, it goes label down. You could get more careful about this and put empty cards in one case and full cards in another, but you need some sort of system by which you can immediately get a strong indication of which cards are used and which aren't. When backed up by examining the card when you put it in the camera, this should create a relative foolproof system of not writing to cards that have images still on them.
There's not much you can do about card corruption except to watch for it. It happens from time to time (it's happened to me twice in 16 years of shooting digitally, knock on wood). By "watch for it" I mean that you need to be on the lookout for a couple of key signs: (a) horizontal lines or changes in color when you preview the image; and (b) missing images when you drop into chimp mode to look at your images. Either is a signal to stop using that card immediately! Why? Because once there's corruption on a card, any additional writes to the card may start overwriting previous image data. You may be able to recover information from that card with one of the many data rescue programs that are available, but not if you continue using the card and start writing over pieces of previously recorded information. Simply put, if you get any sense that a card may not be saving images correctly, stop using the card. Immediately. Put it aside for a long, leisurely examination on your computer with a image recovery tools.
Next we get to the top field issue: losing the card. While the Japanese manufacturers seem to think smaller is better (xD, mini-SD), in my experience the smaller the card is the more likely you'll misplace it or worse still, lose it. It's real easy to drop cards in the field if you're not 100% careful when you perform card swaps. I believe cards should be in cases before you get them out to use, and in them again immediately after you've taken them out of the camera. It's tougher to lose a case than it is a small card. Those of us who shoot in the wilds or at hectic events need to double our discipline in this. Trying to save a teeny bit of time by stuffing a card in your pocket is a fool's errand. When you bend over to do something and the card falls out of the pocket without you noticing, don't say I didn't warn you. Or perhaps you just forget which pocket you put it in or even that you put it in your pocket, and if you're lucky you discover it in the dryer after doing your laundry. For once-in-a-lifetime trips, events, or photo opportunities, you don't want any risk. Develop a card discipline. To paraphrase a slogan made famous by a TV station in New York City: it's 10am, do you know where your cards are?
There is a related issue to these things that doesn't get discussed enough: use more smaller cards rather than one large card. For all three field problems--overwrite, corruption, and loss--having all your eggs in one basket increases your risk of loss, something known as "single point of failure." I know it's convenient to get a 64GB card and shoot with it for your entire vacation and never have to worry about changing cards, but you're increasing your risk of image loss by doing so. One catastrophe now wipes out your entire vacation shoot. How small should your cards be? It depends a bit on the camera and whether you're shooting JPEG or RAW, but I'd say that a card that gives you 200-300 shots is a good balance between not having to change cards too often and not losing too many shots due to a single card loss or failure. If you're shooting very rapidly (sports, some wildlife, some events), maybe up that to 400 images per card. But 1000's? Nope, you're increasing your risk of image loss. Don't do it.
Once we're out of the field and back at our hotel room while traveling, we're at the second point where we risk losing images. Generally, we're transferring our images from card to computer at this point. Some day the cloud will get fast enough and connections ubiquitous enough that perhaps we'll be transferring from cards to cloud, but we're still a long way from that. Moreover, even if you can substitute cloud for computer the same problems come up:
1. What you transfer to shouldn't be a single point of failure.
2. You don't want to write over images.
3. You need to be able to find those images later.
Again, I'll tackle these backwards. You might not think "not finding your image" is a possible risk. Believe me, it is. When you're in a hurry, you don't always pay close enough attention to what you're doing. And who isn't in a hurry to get our images transferred? It's one of those drudge chores that you just want to get done quickly and out of the way. Worse still, too many people still don't understand where programs like Lightroom or Aperture are putting their images. If you use a program rather than just a brute force drag-and-drop transfer, you absolutely need to know what the program is going to do with the images, and you need to verify that it's doing what you think it is every time you make a transfer. If you use multiple catalogs (say personal, eventX, eventY, travelA, travelB) to keep away from the megacatalog problem, it's easy enough to not notice that your travelA images just went into the eventY catalog you were using last week. Depending upon how much you shoot, how you rename files, etc., it could take you awhile to discover where misplaced images like that went. So don't laugh. It's one of the reasons why I use rigid file naming and folder structure practices. If nothing else, I should be able to search for INT_PAT and 2010 and find the images I shot in Patagonia in 2010 no matter where they ended up (and no, I"m not perfect at this; when I get some time I'm going back through my files and fixing some names to better adhere to my practices).
There's that "don't write over images" thing again, too. Hopefully you've already turned File Numbering Sequence On, but if you haven't, this is your biggest point of risk, as it's easy to drop DSC_0001.JPG into a folder that already has a DSC_0001.JPG. Depending upon how you do that, you'll either lose the first image with the name or you might get a DSC_0001-2.JPG image, which isn't necessarily all that helpful, either. Still: File Numbering Sequence On. But how you name your files is important, too. Your renaming convention during transfer should be such that it can't easily create overwrites, too. Note that many of the renaming templates have sequence numbers in them that are four digits long. That's not enough for me. Now into multiple terabytes of images, I need at least six digits to guarantee that I don't accidentally use a name and sequence number that's occurred before. Deal with it now: use a fixed name, date, and a long sequence number. For example: INT_PAT_CHILE_1-2010_######.JPG is good, CHILE_####.JPG is not. Note that I use underscores instead of spaces. Some operating systems don't like underscores, so when you move the file to them at some later date, you can have problems. Another thing to watch out for is the total length of the name. I try to stay under 33 characters. Again, some systems have limits to the length of the name and you can end up moving a file to another system and having problems with your file names if you go too long.
I should mention that many photographers like the date embedded first in their file name, followed by time (e.g. 20110329_092700_restofname.JPG). This allows them to put find images in the order taken. If you've got a DAM product like Aperture or Lightroom, there are other ways to do that, though. Event and journalist photographers seem to prefer the DATE_TIME_NAME method.
Some photographers I know go further and protect (read access only) files they transfer. That makes it even tougher to overwrite a file. But naming discipline is usually more than enough to lower your risk.
Which brings us back to "single point of failure." Just like having all your images on one card is a single point of failure, having all your images on one computer hard disk is also a single point of failure. The laptop may be stolen from your hotel room. The laptop might get dropped and the drive damaged. You might drop your laptop in the river and the drive get damaged. US Customs may seize your computer when you return to the US and not give it back to you for a long, long time (yes, that is definitely a possibility these days). You might accidentally move a folder that has your images in it into the trash. An errant motherboard may send signals to the hard drive that has it write randomly over the surface (yes, I've seen this happen). In short, you need a backup. Good transfer programs can send copies of your images from your card to multiple drives simultaneously, and most pros I know use this facility. They attach a small portable drive to the computer and ingest to both the computer's main drive and the portable drive. Then they separate the portable drive into a different piece of luggage or carry it with them (my choice) in their photo backpack and separate from the laptop. I know one photographer that does double backups. But you can get triple redundancy by just not reusing your cards if you don't need to. Ingest the card to two drives, set the card aside in a "backup" envelope and don't reuse it unless you run out of other cards. Short answer: storage (cards or drives) is cheap these days. Far cheaper than having to recreate lost images.
Phew! We've made it back from the field with our images and now we're in our office and home and bring them over to our main work computer. We're still not out of the woods, though. Our images are still at risk, and several different risks:
1. Loss due to drive failure or theft.
2. Loss due to negligence.
3. Loss due to media choice or file format.
I seem to be working backwards today, so let's look at #3 first. One fear many have is that raw file formats are proprietary and that some day in the future they won't be able to convert their images. That's possible, but somewhat improbable. There are enough open source converters available that understand the various raw formats so that I don't worry too much about the proprietary nature of the file format. I also believe that people in the future will still be intelligent and can research and reverse engineer something if it's really important to get to the underlying data. We're doing that today with things that were done decades ago, for instance. It might get costly to do, but it will still be done by some specialist somewhere.
The bigger problem with our files is the media we store them on. Let me illustrate with an example: I've got quite a few of my very early works saved on floppy disks in Vector Graphic, Osborne, Apple II, or another format. Eek. Even though I could get to the meat of the works if I could just get the file on my current computer, the problem is the media. Listen up, folks: you MUST keep migrating your files every time there's a major media disruption. You should never wait on this. Once a new medium is clearly taking over from an older one, it is time to act. Move the files!
So how does this apply to images? Easy. Right now floppies have gone away and CDs are about to do the same. If you've got images on floppy drives or CD, you need to be in the process of transferring them, probably to hard drives. Transferring them to DVD isn't the solution, because it's clear that will go away in the not too distant future. Don't scoff. Media formats have a strange way of dying very rapidly. Ten years from now do you think your computer will have a CD drive? No, it won't. Apple is already moving away from that, and they're just the first of the bunch. So will there be third-party CD drives you can buy in 10 years? Probably, but there's no guarantee (try buying a 5.25" Micropolis format floppy drive that'll connect to your computer today). You may be bidding on eBay to get what you need. Bottom line: keep migrating your images to "current media." Someday that will be a cloud somewhere, I suspect. For now, the current media of choice is hard drives.
Hard drives themselves are a problem, too. They fail. Let me strengthen that: your hard drive will fail. If your images are only on one hard drive, they are at severe risk. Best practice: four locations: local hard drive, local backup hard drive (can be Time Machine), off-site hard drive, and cloud. These are your life-savers in order: if something happens to your work drive, you restore from you local backup. When your house burns down, you restore on your new computer with your off-site hard drive. When all hell breaks loose and you're not settled at all for some period of time, you work from the cloud. At a minimum, have a backup and make it offsite (cloud). But I'd say that best practices at reducing risk is what I just outlined: on-site backup, off-site backup, and cloud.
Note: a RAID system isn't really a backup. It's redundancy and possibly hard drive failure protection, depending upon how you have the RAID configured. You need to back up your RAID!
I skipped over negligence, but it is solved by back-up, too. I've done some stupid things on my work machine from time to time, accidentally trashing a file or folder I actually wanted by not paying enough attention to what I'd grabbed. Most of the time, I discover that fast enough that I can pull things back out of the recycle bin. But even small things: like editing the original file instead of making a copy, start to put my images at higher risk than if I just did the right thing in the first place.
You will make mistakes, you will have technical problems, you may have a physical loss (fire, theft, etc.). When any of those things happen, you risk losing images that can't be replaced. So, to summarize, whether you take my specific advice or not, at least do the following things:
- Minimize single points of failure.
- Create discipline and habits, which are better than randomness.
- Be aware of things that cause overwrite or change of original data.