Cameras: the Next Ten Years
With a title like this, this article is, obviously, doomed for failure. The world around us, and technology in particular, evolve, but rarely in the direction predicted by self-proclaimed prophets.
Still, I'm going to take the risk. One, we may have a good laugh browsing through old issues of Quest five years from now; two, if you disagree with some or all of my predictions or remarks, my goal has been achieved, your curiosity stirred.
As a child I've read a book, printed in Communist-ruled Poland and placed ten or fifteen years in the future, where collective farm workers commuted in small air taxis, happily praising the legacy of great Uncle Stalin. Therefore as far off as I may end up in my predictions, I feel safe from ending up in the low end of the spectrum of failed future-tellers.
Film: thing of the past
I will not dwell on this subject, discussed at large in my first Digital Corner column (Quest #4). Voices which as recently as three years ago kept proclaiming that digital will never match the quality of film, are strangely quiet now.
The transition is happening faster than expected. It is driven by all three segments of the market: low-end (ease, convenience, gadgetry, and instant gratification), professional (image quality, predictable and instantly verifiable results), and advanced amateur enthusiasts (all of the above, plus no darkroom drudgery). It also enjoys strong support of camera makers, eager to see millions of users replace their film equipment with digital one (and upgrading the latter more frequently during the initial phase).
I would expect film to disappear from convenience stores and newsstands within the next five years, in ten years becoming a special-order item, very much like B&W negative today, or like vinyl audio records.
The Quest readership, lovers of well-crafted precision instruments of yesteryear, is not the best audience to which I may preach my expectations, but this will happen whether we like it or not. It is better to accept the news from an insider — I am an old-time film shooter and camera collector, and I own almost fifty film SLRs, most of them in working order — but this is the time to smell the coffee. I'm still using my film cameras from time to time, but not for real, result-oriented shooting.
After all, the silver-halide chemistry has been in use for more than 150 years. The brute-force, digital approach to information storage and processing has made great inroads in all other areas; the film bastion is perhaps the last one to fall.
It has been sixty-nine years since the first 35-mm SLR, the Kine Exakta, saw the daylight. There is no doubt about SLRs being most versatile tools in general photography, and the basic construction principles remained unchanged during all those years. An amazing longevity, but this may, I am afraid, end soon.
Fig.1: A Kine Exakta of 1938 (with the lens from circa 1949), the second version of the first model of the first 35-mm SLR.
This camera is still taking pictures, and will keep taking them long after my all-electronic wonders are unserviceable because of lack of spare parts or batteries — if only 35-mm film is still available at specialty suppliers.
Still, at this moment SLRs are, for most uses again, digital cameras of choice in the professional segment of the market, and largely so in the advanced amateur one. I expect this situation not to last forever: a new generation of EVFs will be coming soon.
As you may recall, I'm not a fan of electronic viewfinder (EVF) digital cameras. Their finders are non-responsive and jerky, resolution is ridiculously low, tonality stinks to heaven, and they eat batteries fast. Unfortunately (?), this may be the kind widely replacing SLRs in the next ten years or so.
If the chip processor speed is increased by just 50% a year (a modest assumption in embedded applications), in six years this means an increase by thirteen times; more than enough to provide smooth display and enough power to spare to do all real-time image processing (autoexposure, white balance, autofocus, etc.) A ten-fold increase in the number of finder pixels will bring tripling of the resolution, enough for most applications; one may also expect better tonality. Last but not least, new display types offer much promise in terms of energy efficiency: organic light-emitting diodes (OLEDs), already used in a few Kodak models, should reduce display power use by a whole order of magnitude when the technology matures; they are also cheaper to make than LCDs.
Therefore EVF-based cameras may soon provide, arguably, the best of both worlds: real-time viewing of a quality comparable with today's SLRs, with less mechanical complexity and at a lower cost. In addition, this will allow for lots of extra information, part of it genuinely useful, to be overlaid with the image preview, including not just exposure data, but also brightness histogram and/or over/underexposed area indication, sorely missing in today's digital SLRs.
(A notable exception here is the discontinued E-10/E-20 SLR line from Olympus, using a semi-transparent prism instead of mirror, and allowing for real-time image preview and processing. Ironically, while Olympus abandoned the concept, I've heard voices in the Nikon community, asking for a similar model to address the built-in shortfalls of conventional digital SLRs.)
This will not lead to the demise of the SLR, at least not immediately. EVF cameras may, however, soon dominate the middle segment of the market (advanced amateur), with SLRs withdrawing to the higher ground and becoming more expensive, due to the reduced scale factor advantage.
I wish I were wrong in this prediction, but I fear I'm not. We are witnessing the swan song of the SLR, digital or not.
Extrapolation: more of the same
The megapixel race seems to be over. The time is now to address other issues of image quality.
What should, and hopefully will, come next, is the tonality, more exactly the image bit depth.
Current digital camera sensors capture from ten to fourteen bits of each of the color image components (R, G, and B); this corresponds to from one thousand to sixteen thousand brightness levels per component. Unless the image is saved in the raw format, this is converted to just eight bits, or 256 levels, in the resulting JPEG image file (true, the levels in a JPEG use a logarithmic scale, which means that the brightness range is not lost, but the transitions are not as smooth). With better sensors, more processing power, and cheaper storage the saved color depth should increase, perhaps to sixteen bits per color. As a result, we will see better tonality and smoother transitions; the highlights and shadows also will be better protected from loss of detail.
To take full advantage of this, the industry will have to abandon the JPEG image file standard. I'm puzzled why this didn't happen years ago: the new JPEG 2000 format has been around for five years; it is supported by most graphic applications, and provides not only less quality degradation, but also better compression than the "old" JPEG. Perhaps the mass market just does not care, but what about the other segments?
Another area which will see much progress is the sensor sensitivity, usually expressed as the film-equivalent ISO setting. The sensors (CCD or CMOS) allow to adjust this within a given range; higher sensitivity usually comes at the expense of increased image noise (very much like film grain). At present, most of the decent digital cameras offer usable sensitivities up to ISO 400; a few (notably Canon SLRs) even to ISO 1600 or 3200.
While for most users, shooting under daylight or with flash, ISO 400 is all they will ever need, higher settings offer more freedom in available-light photography — which usually means less blur caused by camera shake. I would expect ISO 1600 or more at acceptable noise levels to become common within the next three years, at least in mid-to-upper range cameras.
Higher sensitivity will bring the demise of image stabilization. Let's face it: this is a Ruby Goldberg solution anyway: a complex circuitry to detect camera shake, and then using mechanical means to move a corrective optical element within the lens (or, in case of Minolta SLRs, the image sensor itself) to compensate for that; all of this just to allow for doubling or quadrupling the handholdable exposure time. The same effect can be more easily (and much more cheaply) achieved by doubling or quadrupling the sensor gain, as discussed above, and using higher shutter speeds. Therefore I consider image stabilization to be a transitional technology, becoming irrelevant as soon as alternative means of achieving the same goal become available.
(In principle, sensors with electronic gating, which limits their response to a given time window, could allow for fully electronic shake compensation, without any moving parts, but at the expense of radical re-design of the circuitry, totally impractical at this time. I don't expect this to happen.)
Incremental improvements in power consumption, display brightness, autofocus reliability, and some other aspects are rather trivial in comparison, and we will see some evolutionary progress in these areas every year.
New stuff: the useful and the gadgetry
Lots of improvements and innovations will happen, I would expect, in the area of digital in-camera image processing. As the market is becoming more saturated, the manufacturers will become eager to introduce features distinguishing their models from the competition. Some of these features may be really useful, some will be mere advertising ploys.
I would expect the new developments to depend mostly on camera's image processing. You may leave some optical flaws of the lens uncorrected (to reduce cost or to allow for better correction of others) and to compensate for them in the firmware, before the image is written to file. Olympus was the first company to use this approach; in their E-1 and E-300 the camera compensates for image brightness fall-off, based on the characteristics of the lens in use (the lens is computerized itself and passes the relevant data to the camera). The distortion, also based on the lens data, can be similarly corrected, but, at this moment, in postprocessing only.
I expect this approach to be accepted by other makers soon. We can see some degree of chromatic aberration to be corrected this way, maybe some other lens flaws.
Bordering on science fiction: using some very smart Fourier transform math it could be, in principle, possible to retrieve some detail lost by a not-so-sharp lens. This would be a real revolution in photo technology, allowing camera makers to cut expenses significantly. Note that this is not the same as the crude, mechanical image sharpening done in today's cameras, which increases the apparent contour sharpness (making the tonal transition steeper) but does not retrieve any detail: imagine a post stamp blown up to a poster size, with all contours retraced with sharper lines, but no dew detail added.
On a more realistic note, I'm expecting to see in-camera perspective correction (compensating for the camera pointing upwards or downwards). At present you can do it by hand in almost any postprocessing program; earlier we used to play the trick tilting the enlarger head (or, in a more practical way, the mask board).
This could be done either manually, with the user controlling the degree of adjustment, or automatically, with the camera knowing the zoom angle and the lens axis orientation. In principle, this is achievable today: when the firmware picks up the individual R, G, and B signal components, interpolates them and merges into RGB pixels, it does not have to put the combined result into the same place in the processed image. With faster processors and better live preview (see my ranting on EVFs above), you could even control the effect while taking the picture!
Let me skip all the "improvements" blurring the line between cameras and all-purpose electronic gadgets. A market for those will exist, and you would be able to use the same device to make a phone call, listen to music, do your pedicure, and take pictures of Junior to be send to the in-laws, and many people will happily jump onto the bandwagon. This is normal: some people do not want to be photographers, they just want to take acceptable pictures.
What remains? The photographer
What will not change in the coming decade, however, is the need for a knowledgeable, skillful individual behind the camera. Most of the hopeless pictures taken every day all over the world will not become better just because of the advances in camera technology. The best exposure automation can still be fooled, or it may give results different than intended, the autofocus may still have a different idea on what to keep sharp, and many camera users will still not comprehend the difference between seeing the subject and composing the image in the finder, regardless of the latter's type.
The technology progress does not help much: there seems to be a growing general attitude: "I paid X hundred dollars (pounds, guldens) for this thing, and now I have to learn how to take pictures?" Advertising claims by manufacturers do not help, either, often reinforcing that posture.
The readership of books on photography does not seem to keep up with the demand on new gear; the books themselves also follow a declining trend in terms of quality, depth, and technical expertise. Recently I find much pleasure reading books on photography published fifty years ago or earlier: get a book on Leica photography from the late Twenties and you will see what I mean: more meaningful information than in most of those glossy, nicely printed and lavishly illustrated books of today. An introductory chapter on lenses from a general photography book from the Fifties can tell you more on how the image is created than a recent book on photographic lenses from a major U.S. publisher (and fewer factual errors, too). No names, please.
Fig.2: A German stereo camera from the very beginning of the 20th century, taking stereo pairs on 9×13 cm glass plates.
(Reproduced from Ratgeber im Photographieren by L. David, 78th Ed, Wilhelm Knap, Halle, 1914.)
Oh, well, already Plinius the Elder was complaining about everything going downhill. Maybe I'm just getting grumpy. Get a camera (film or digital) and go out to take some pictures before the end of the world as we know it.
This article was originally published in Quest, the quarterly newsletter of the Olympus Circle; September, 2005.
|Home: wrotniak.net | Search this site | Change font size|
|Posted 2005/12/17||Copyright © 2005, 2006 by J. Andrzej Wrotniak|