dante stella stories photographs technical guestbook

"Tell me, Winston, how do you really feel about Big Brother?"

"I hate him."

"You hate him. Good. Then it is time for you to take the final step. You must love Big Brother. It is not enough to obey him; you must also love him."

 
Digital? Maybe later.
 

I wrote this in 2003.   Since then, I have spent quite a bit of time with a Kodak DCS 14n and a Nikon D2x - and despite that, I don't take back one damn thing I originally wrote in this article!

Digital, digital, digital. I keep going out to buy a digital camera. I am fixated with the idea of one for color work - just the odd roll of snapshots. The rest of what I do is with a 6x9 and in black and white. I don't entertain the idea that for black and white work, that an off the shelf camera is going to work, at least any time soon. So I go to the store, go through the cameras, and I never bring one home. Maybe I'm too deliberative about the things. Every time I think about it in the abstract, I think how neat it would be to have a digital system - where there would be no hanging out in the darkroom, no terrible labs, no negatives to file, no prints to sort. But in the here and now, sifting through the costs and benefits, digital leaves me blase. Although the technology is fascinating, at this point, it only seems to change the nature of the steps, not the amount of trouble or expense, it takes to make photographs.

1. My sense of the why or why not

The first time this editorial came out, I made the mistake of not really explaining my viewpoint. I don't think it's completely typical of advanced amateurs, but it will help you understand how I see this. I like black and white photography. I have spent about six years shooting 6x9 black and white and about 17 years shooting various forms of 35mm film. My favorite film is Verichome Pan (there is a freezer full of it here), and when it is too dark for that, T-Max 400.

Spending all day in front of a computer, I find it relaxing to make photographs without being completely dependent on using one. The exception is some post-processing on Photoshop (of which I was an early adopter in 1992 - with version 1.0. Somewhere I have that versionon an 800K Mac disk). I find a simple pleasure in getting 8 pictures from a roll of film, not worrying about what it looks like at the time. In fact, I like to develop the negatives and put them away until I forget enough about the scene that all I am viewing is the composition. Then I turn on the Durst enlarger (which has its own densitometer for getting at least in the ballpark) and then get to printing.

The beauty of this arrangement is that for all of its asceticism and technological backwardness, it works. Because the equipment was quite economical (bought from various people and places going out of business), I don't feel as if I have to shoot a hundred thousand pictures to justify it. In fact, if the equipment produces one good picture in its useful life, that's great. And when I am up to working on stuff for the web, everything ends up in Photoshop.

2. How I observe the "paradigm shift"

The usual paean is that the "digital revolution" is being led by photojournalists and professional photographers. A lot of people think that if digital is good enough for them, it should be good enough for anyone. This is as good a place as any to start, perhaps with the point that many media applications don't require as much technical perfection as they do equipment survivability, and increasingly, speed. There is also a factor involving reduced costs to employers and customers.

First, whether digital has benefited journalists as much as their employers is not clear, at least to someone on the outside. In the old days, a photojournalist could wander around with his Leicas and whatnot, none of which had any great volume and the sum of which had relatively little weight. In terms of the Contax equipment Capa took to Normandy, all of it together would weigh about the same as a D1x does now - the body, that is. Earlier this year, American Photo did a layout of the absolutely massive amount of equipment people were taking to Iraq - it was something like two D1xs, battery chargers, inverters, microdrives, laptop computers, and full chemical gear. Even looking at the antecedent, an F3 with an MD-4, it is a staggering amount of equipment -- one would think that technology, especially miniaturization, should make things easier. Today, as in many fielfs, the "independent contractor" worker now performs his "work for hire," and is left holding the bag for capital equipment.

Second, in many contexts, digital seems to have shifted the burden of post-processing to the photographer. With analog materials, this burden (getting to the proof stage) rested on laboratories, which were expected to be good at outputting. Lab services could be expensed by the media or passed through to the customer with portrait shooting (although whether labs are any good anymore is an open question). Digital has pushed this "workflow" back into the lap of the photographer. This hapless person's day rate has climbed not one bit since the 1980s (if even that late). Those who do not want to be a part of this are replaced by students, interns and people who are willing to work for bottom dollar. Even in the midst of this, though, there is cause for optimism:

-- In some contexts, such as in professional studios, digital post-processing can be easier. One example is if you previously did high-volume color printing yourself. Color printing can be miserable.

-- With digital, there is the potential that you can skip the entire post-processing sequence for a large number of frames by simply viewing their thumbnails and pitching them.

-- For high-volume commercial work (such as catalogs or web-only), digital would not have some important advantages - starting with eliminating the need for a color darkroom, something which is a health and environmental nightmare. And there would obviously be no scanning.

It was once pointed out that the photographer is regaining control (or responsibility) for all phases of the process, making things more personal. This is true and not true; some steps add value in terms of final product; some don't. I tend to wonder if I could train someone to develop Tri-X the way I like, what it would be worth not to be developing it myself. In the end, I always choose to retain control. All of this is a tradeoff; it all depends on what you perceive to be the best use of time or money. Black and white printing is not something which is easy to delegate, because it is so dependent on frame-by-frame judgments.

Third, I find it a little troubling that current digital file formats (whether they originate from film or digital equipment) make it difficult or impossible to control dissemination of copyrighted material. This means money out of the pocket for professionals and some loss of control over the finished product for hobbyists. It seems that this could turn into an egregious problem for portraitists, because once a digital photo is emailed, it can be proliferated on an exponential scale without any control over the future revenue stream. With old-school photography, you would have to scan, make internegatives, or engage in other questionable processes which tended to degrade image quality. In the here and now, an intelligent worker could potentially calculate a present value for all future rights and charge that, but for every smart one there are a hundred whores who will try to reach a race to the bottom in rates, based on a much shorter-term business model.

Fourth, in the amateur world, digital is capable of making it just as difficult if not more to get to the equivalent of the 4x6 proof print. Perhaps the solution is that you don't make 4x6 inch prints and instead view them on a computer (which is what happens to 70% or more of digital pictures, per some recent studies. But that is not like a photo album that you can flip through.

Fifth, when you sit down and break down costs, digital doesn't really make photography cost any less; it mainly redistributes the capital cost of equipment. How, for example, this works, depends on the output and the image type. Here is an example using the N80 vs. the D100 and assuming you do black and white (I actually use a Fuji 6x9 and 13 rolls of 120 film to do this, but N80 and D100 are much more apples-to-apples). I am not doing this in color, because color darkrooms have never been that pleasant to have in houses.

Item
Digital
Conventional
Input device Nikon D100, $1500. Nikon N80, $300
Lens Nikkor 24-85 f/3.5-4.5G DX $300 Nikkor 24-85 f/3.5-4.5G DX $300
Media for 100 maximum range images 512Mb Compact Flash Card, $100 4 rolls of film, $10
TTL Flash Nikon DX flash, $300 Nikon normal TTL flash, $150
Computer $2,000 $2,000
Negative scanner N/A $800
Printer Epson 2200, $700

Enlarger, timer, easel, lens, trays, tray magnifier, safelight, and tongs, $800.

In the end, in terms of capital costs, they're about even. You could conceivably skip the computer, negative scanner and/or darkroom equipment if you shot conventional and never did your own printing and had, say, PhotoCDs made. Or if you were lucky enough to be at school and have all of these things available for your use. But it is a significant capital investment on either side, assuming that you are starting from zero - which you largely end up doing when you switch from one to the other (from one-third to one-half of the $4,000+ cost of each system is specific to that system). Of course, you need to make adjustments as necessary; if you are printing primarily color from film, you would probably buy the printer instead of the darkroom equipment.

Sixth, where you do output images, and depending on how you handle it, digital can end up being just as expensive. Digital output materials seem to be priced at parity with black and white paper and chemicals, so you don't really save much money either way. If your system is correctly profiled, you might get away with fewer digital prints. Whether you will have to reprint inkjet prints in the future is an unknown (the tests I have seen on longevity all seem to assume very low light levels, certainly a lot lower than room light in most cases).

Finally, in terms of operator time, the cost there with either system (there is a cost) depends on how much time you want to devote to doing the basic work on images; with either, you can devote as much or as little time as you want to processing images. Depending on how much or how little you value your time, and what the anticipated output is, having an entire set of prints made for $7-12 by dropping them off somewhere may cancel the entire processing advantage of digital.

The ultimate economic analysis for the non-pro user will vary as much as the person; but one thing is clear: you will spend about $2,000 in a changeover, and your variable costs will probably remain constant. If you want to think about the economics, I guess you'd better be good at Excel.

3. The unrealized promises of digital - at least at this point

Promises of better performance have been pervasive in the photographic industry for decades. In the old days - and at least as late as my childhood, manufacturers have always tried to make rank amateur formats easier. The original Kodak could not be opened by the end user. The brownie format (120) came about to help eliminate the need to deal with plates. Then 126, 110, Disc, APS - well, you get the picture. 35mm film is rapidly losing adherents in new cameras, but is it because its sales spiked when they figured out how to make idiot-proof 35mm point and shoots? I tend to think that it only became as popular as it did because someone figured out how to make a mass-market product for it in the 1980s and 1990s. As someone who has watched (and heard) various promises about digital expanding capabilities, it has failed to live up to some of the talk.

Loss of certain types of lenses. Manufacturers began to make subframe (less than 24x36mm) DSLRs, touting their compatibility with existing lenses. While it is true that older lenses do connect, their characteristics (and utility) can change greatly when the viewing angle is reduced by the smaller digital sensor. Manufacturers have not adjusted their product lines accordingly.

--For example, the 105/2 AF-D DC Nikkor, a splendid portrait lens, becomes the equivalent of a 150mm f/2. To take a head shot, you have to back up five feet, which actually increases the depth of field (something fast portrait teles are designed to knock out). Wideangles lose their wide-angle nature - but none of their size. The only benefit is making tele lenses longer. This benefits sports photogs, yes, but they are a very small part of the DSLR market.

--28mm and 35mm perspective-control lenses become a lot less useful when they are scaled up to essentially 42mm and 52mm lenses, respectively. Although you can do some perspective control in photoshop, you cannot change the angle between the film and subject planes of focus - which is a good part of what PC lenses are about.

-- to get a 20mm angle of view with most DSLRs, you end up needing a 14mm lens. That's about an $1,000 price difference if you are a Nikon person.

This was supposed to have been addressed by scaling up the sensor size to 35mm, but no manufacturer has been completely successful in doing do. The Kodak 14N has been trashed for noise, and the EOS 1Ds has incredible color fringing at the sides of the frame (take a look at the Kodak DCS 14n vs. EOS 1Ds tests on the web to see just what microlenses do to frame edges).

Viewfinders. This is where digital SLRs are not much fun. You would think that at least as a consolation for the reduced angle of view that the viewfinder would have a larger magnification and 100% coverage - but no. Subframe DSLRs simply mask down the existing 35mm viewfinders of the camera platforms they are built on. This makes the AF sensor areas unusually big compared to the final frame (ever see a D100 finder?), making it hard to see whether you are focusing on an eye or an eyebrow, especially with the advent of focusing screens that don't really have matte surfaces on which to focus.

Smaller, faster cheaper lenses? Where are these? This was a selling point for sub-frame sensors, and it has not materialized. A 12-24mm Nikkor DX lens (which covers only digital sensors, not a 24x36mm frame) is one stop slower than an 20-35/2.8 (a big issue where most digital SLRs fall apart at EI400), not that much smaller (it is the same width at the front end), and not much cheaper (is it even?). And some Nikon DX lenses are now sporting absolutely massive 77mm filter threads (12-24 DX and 17-55 DX). Not that the camera bodies are getting any smaller. Even when freed from the constraint of covering a 43mm image circle, optics apparently cannot be successfully scaled down that much. So the image circle shrinks, but the lens does not.

Artifacts of digital. There are the chromatic aberration, moire, and noise issues inherent to moving from an extremely thin film medium to a checkerboard CCD or CMOS chip. One of these problems, moire, is almost completely absent from the analogue world. Another, aliasing, is correctable only by "low pass" filters which essentially cut resolution. The Foveon X3 chip was supposed to sidestep the moire problem, but it never ripened into a realistic solution; Foveon's chips are absolutely tiny, and your sole choice of camera is Sigma - with a huge 1.7x change in angle of view (not to mention a shortage of prime lenses). Either Foveon is too greedy, lacks intelligent sales people, or can't implement the technology successfully in a 10MP chip.

Thinking inside he "box." Digital has failed to escape the conventions of 35mm film bodies. As you probably figured out, the 35mm SLR film body (the "box") has a width defined by the size of a 35mm film catridge, a 24x36mm film gate, and a takeup spool about the same side as the cartridge. Digital SLRs have not broken from this, despite the smaller sensor size and lack of a transport system.

With Nikon, you have to guess that it is capital investment, although Nikon pundits will tell you that the D1x and the D100 are ground-up bodies. It sure doesn't explain Canon, which has beaucoup bucks from selling copiers.

The genius of the new Olympus is that it is an SLR which doesn't feel like it has to look like a 35mm SLR. But is a 4/3 chip better? Maybe from a cost standpoint (35mm-full-frame chips have close to a 100% rejection rate in manufacturing, which is what makes them so costly), but the smaller physical pixels (which will only get smaller when the pixel pitch increases) are fighting a battle against a lower signal-to-noise ratio (since it takes a certain number of photons to register a pixel). Maybe the solution is not a 24x36 sensor, but one that is even bigger?

In a greater sense, there is a question of whether good digital can escape the size of 35mm SLR cameras. Digital lenses have to be telecentric, meaning that they have to have a narrow angle from lens rear element to the sensor. SLR size factors work well, because they have retrofocus wide angles, and the telephotos have enough back focus so that light hits the sensor dead-on. If the s/n is optimal with an 18x24 sensor (or bigger), then the size of digitals (at least front to back) may never drop significantly from that of a 35mm SLR.

So despite the technological promises, digital has failed to advance the ball in any of the areas where it promised improvement over conventional equipment.

Too much light blinds us. In addition to being a ways from its goal, digital imposes its own unique handicap: it has much more limited dynamic range and far less exposure tolerance at the ends than film does. This makes shooting digital as demanding as shooting slide film.

Compare 8 or so stops of range with digital to 15 stops of straight line response with a T-Max 400 or T400CN. The difference with film is that everything gets recorded; what you lose is your choice. With digital, that choice is made at the time of exposure - and sometimes in situations where you are not best positioned to make the call. You can't look at a histogram every time you are shooting scenes with white objects. If you have a tripod, you can take two, one low exposure for highlights, one high exposure for shadows and composite, but this is pretty time consuming (and requires a very motionsless camera. RAW shooting and Kodak's ERI-JPEG technology may bea good start toward a solution, but the former needs tons of storage space and the latter is available only on $4,000 cameras. If you are into scenes with a lot of information in the highlights, all of this is a little biy troubling, because the teeny tiny LCD screen doesn't do a good job of showing highlight separation.

It will be interesting to see how this all plays out with Photoshop CS, which should (finally) allow users to bring in raw 16-bit b/w scans from CCD negative scanners and edit them in 16 bits and to do the same thing with digital camera files.

4. Encoding and storage

One aspect of the digital versus film battle you never see discussed is the issue of storage - both at acquisition and in ultimate storate. Film is its own storage medium, at least film negs are. Digital requires a different type of storage and several things about digital are just... different.

This begins with the shooting. Every film frame is 19 megabytes in 35mm; a roll of film is 684 megabytes. A roll of film costs from $1.50 to $8, depending on the type; a 512Mb compact flash card is about $100 (on Ebay; retail is more). If you are traveling, unless you want to take a $400 iPod or portable hard drive as a storage device, to shoot digital the same way you shoot film, your storage cost would be 30 times as high. Yes, one argues, but digital lets you discard on the spot. Whatever the surface appeal of this argument, unless you discard 29 out of every 30 pictures, or take fewer than 40 pictures per trip, or shoot all in JPEG, this reasoning simply doesn't work. Digital simply requires a bigger investment up front.

The encoding speed for film (which is nothing more than optical storage) is a lot faster than digital; you can open up with a $600 F4 and shoot 36 pictures in seven seconds, something which in digital can only be done at greatly lowered resolution and greatly increased cost. This is probably a non-issue for many people.

Next is the processing. That's probably a wash, although your results might vary with what you are doing. With digital, a computer is a necessary part of color correction and exposure fine-tuning, something which for most film-based prints is handled by a machine at a minilab. Unless you shoot about 300 rolls of film in any two year period (at $7/roll processing), you would be hard-pressed to justify the necessity of a $2,000 computer to process snapshots. Of course, if your ultimate output is electronic, you have to buy the computer anyway. So for "processing," it only really boils down to the value of your time and what efficiencies you can develop (with Photoshop, you can develop considerable ones

Then, the storage. With film, whatever you want for your computer can be scanned on a frame by frame basis. You only need to scan the pictures you want to keep. Marginal ones can be put in storage and scanned later. With pure digital systems, you need to keep ALL of the images you think you might EVER want to use. This leads to greatly expanded storage needs, to say nothing of the fact that you can't just store images on cheapo CDs. The tradeoff is that digital allows you random access (you're not flipping through negative books) and greatly reduced physical storage space.

5. Message versus medium, and the divide between efficiency and aesthetic choice

One of the greatest challenges film cameras face is the computer. Even before digital became viable, the small sizes of images displayed on a computer screen made the differences between expensive film cameras and idiot-proof point-and-shoot types largely vanish. Digital goes one step further, because even relatively poorly-performing consumer digital products (down to two megapixels or so) can make 72dpi on a screen with enough colors to make the eye think it is continuous tone. The flip side of the coin is that on a computer screen, no one can tell how the image did originate, making it a great equalizer in another way: the photographer's skill becomes important. Once it's on a computer, it's on a computer. Film does retain one key advantage: it retains more than enough data to "look good," and after one initial scan at 4000dpi or so (19-40Mb for a 35mm frame), it is a question of downsampling, not interpolation.

The comparison of the development of digital over analog to the development of photography over painting is interesting, but not from a technological standpoint. In the old days, painting was used almost the same way photographs are used today. As much as you might hate to think it, most paintings you see in musueums served no higher purpose than pictures that came out of Grandma's 110 Kodak. When photography developed in the late 19th century into something for the masses, painting became less utilitarian and more an issue of a deliberate choice to use a particular medium.

To attempt to take the argument further, to somehow suggest that traditional photography will be marginalized to an "artistic choice" is somewhat inflammatory, but it may be correct. It recognizes that there is a considerable divide between utilitarian photography and photograpy driven by aesthetics. To be fair, all photography is a tradeoff between convenience and aesthetics, and the dominant process represents the popular balance between quality and price. In the late 20th century, this was 35mm film. In the late 19th century, it was glass plates and colloid film. The balance does not prevent people with the time and money from doing it they way they want to.

To illustrate this point, 8x10 photography (which shares chemicals and basic materials with 35mm and medium format, but virtually none of the same shooting or finishing techniques) remains surprisingly viable, even 100+ years after the development of rollfilm. This is difficult to understand in a world that is dominated by measures of lens resolution (or megapixels) which says that for all intents and purposes, 120 is as good as 8x10, 35mm is as good as 120, and digital is as good as 35mm. The measures (a rat is a dog is a pig is a boy) would suggest that 8x10 should have gone by the wayside and that the equipment should not be sold for anything except decoration.

When you tally the comments made by professionals who prefer digital, it always seems to be for reasons related to efficiency (e.g., no processing, retakes, instant checking of results) - not for the reason that digital makes their photographs better in an aesthetic sense (unless you count grain reduction and sharpness as aesthetic - but those are more technical features than anything).

6. So why do manufacturers put all the money in digital?

You would too. One thing that people have emphasized is that manufacturers "lose money" on every digital camera sold. Yes and no. Manufacturers may not meet unit costs of production now (average of the sum of the fixed and variable for the quantity they build), but they are making incremental costs. Once you factor out the cost of the factory, tooling and other infrastructure, consider that there are very few moving parts in these cameras (reducing the need for skilled labor and lowering overall cost). The elimination of the drive system alone eliminates numerous motors and gears, all of which (on a high-end camera) would require manual assembly. As long as you can scrape by making the variable cost and paying interest on your debt, you can hang in there.

But why take the risks? What's the payoff? Because with film there were no worlds left to conquer. Digital opens up a new world of planned obsolescence that could not be stoked in the film word. Consider that SLR technology progressed for about 50 years after World War II. Manufacturers added TTL metering, autowinding, multicoating, DX coding, multipattern metering, auto-winding, and ultimately, auto-focus. Unlike in the prewar era, when the lenses were the limiting factor, postwar lenses were at least the equal of the film. This meany that image quality gains were largely driven by improvements in film. As you can probably imagine, this created huge backward compatibility and emphasized the role of film, not exactly something that the Canons, the Minoltas, and the Nikons of the world wanted to see.

Autofocus, the last innovation, as Canon and Minolta made clear, allowed a huge break and allowed manufacturers to build and sell whole new systems from the ground up. This began a new feature race. Although the accuracy of AF systems never really improved, their speed and convenience steadily increased (and could be controlled to maintain product differentiation and an upgrade path. When the Nikon F5 came out (as well as its Canon counterpart), there were simply no technological worlds left to conquer for fim: you could shoot eight frames per second, focus on objects travelling at 200 miles per hour, and meter scenes using 1000 Matrix segments. Moreover, there were literally millions of mechanical, manual focus cameras on the market, the market was saturated with AF SLRs, and there were no new features to sell. From a consumer standpoint, better pictures did not come out of successve generations of equipment, and pros were (and are) such a small part of the market that they probably could not have any significant effect on sales, even if manufacturers came up with something new to sell them.

Manufacturers could sell digital to pros (and always has) on the grounds that it is faster. As the market for photojournalism became more cutthroat, digital took over, and the more reliable, the better. Witness the primacy of the D1x, a camera which is obsolete by any practical measure, but nearly indestructible.

For consumers products, though, the business model was changing: to paper comsumables rather than film-type consumables. Canon figured out that if you sold cameras, printers, ink and paper, you could control all of the revenue from a product. Cameras are expensive, printers are cheap, and ink is outrageously-priced. To some extent, cameras moved to being a planned obsolescence item (like computers), subject to the limitation that to drive this, you need:

o Sensor improvements;

o Image processing improvements; or

o Mechanical failure.

This model assumes that one or more events will happen. Sensor improvements will probably be hard to sell as long as storage is an issue. Memory cards for digital cameras can cost as much as 100 rolls of film, and increases in resolution may actually exacerbate demand, driving prices up. Image processing improvements may not have much impact on everyman.

Image processsing upgrades might sell, but only to a point; the 35mm SLR market has shown that most people will settle for 2-3fps, which doesn't challenge most digital systems on the market today.

Mechanical failure both looks bad and is largely unattainable by consumers - a "wimpy" 30,000 exposure rating for a D100 is still more than most people shoot in their entire lives. So ultimately, manufacturers may be painting themselves into a corner with DSLRs. If they miscalculate, and sales flatten (or drop), it could spell the end for some manufacturers. Eventually, you do need to pay for that factory.

7. Upshot

As the Chinese curse goes, "May you live in interesting times." These are, indeed, interesting times - for everyon. And as always, there are always two dangers in predicting the future.

One danger is to predict that a particular technology will never develop. Every time someone does the former, he ends up disproven by history - witness going to the moon. It is not realistic to predict that digital will never surpass film in quality or convenience for most people or ultimately all people (given the time and money). If you have money to burn (about $14,000), you can replace medium format film with a Kodak DCS Pro back today. And as far as time goes, although the top-end price point for electronics never seems to change, something attractive (like a 14Mp DSLR) may eventually filter down into a popular price range.

But the other hand, it is an equal danger to predict that something is inevitable, as some digital proponents would predict that film will inevitably be marginalized - or that it is already "dead." As a student of history, I can say that a lot of "inevitable" events failed to materialize, for one reason or another. I'm sure that 100 years ago, someone said that photography would inevitably kill painting...

DAST