Cameras are changing fast - and we're only at the beginning of the "digital revolution".  This article is about why I think cameras are going to start changing faster than ever...

The perfect camera

What is a camera anyway?  I think photography can be broken down into 3 stages:

(1) The (mind’s) eye
(2) The camera
(3) The “darkroom”

The mind’s eye is unlimited. The camera and “darkroom” introduce constraints, and the perfect camera could be defined as one which doesn’t introduce any constraints. The ultimate car is one which costs nothing, uses no fuel, lasts forever and has performance limited only by the driver’s ability. The perfect camera is equally easy to imagine. Some people would say their ideal camera would have the image quality of a 5x4 field camera but be the size of a mobile phone and cost £100. These people are too easily pleased! The perfect camera can capture images of infinite resolution with faultless colour reproduction in any lighting conditions but costs and weighs nothing.

Do the photographic Oakley shades used by Tom Cruise in Mission Impossible II represent the camera of the future?

New cameras for a new Millennium

The sixteenth century gave us the camera. The nineteenth century gave us film, and photography as we know it. The twentieth century gave us colour. The new millennium inherited from its predecessor some photographic equipment which hadn’t moved on much in the last few decades, and in some instances (APS film?) had moved backwards! The constraints on photographers who used film in 1999 (i.e almost everybody) were very similar to those constraints in 1935 when Kodachrome was first introduced. Camera design was constrained by the range of film available, and the best films hadn’t improved much in that period.

Caught between a shaky rock and a noisy hard place

The photographer in 1935 had few worries taking photos in bright daylight. But as the day drew to a close, the light started to dim and the photographic potential got more interesting, he suddenly ran into difficulty.  There wasn’t enough light to take photos at short enough shutter speeds to hand-hold the camera.  Solutions – a tripod, faster film or faster lenses.  None of these are great solutions.  Faster film means more grain.  Faster lenses get exponentially heavier and more expensive, depth of field and sharpness reduce at faster apertures and there is a physical limit on how fast lenses can get. Tripods only work perfectly when the subject is stationary (and no subject is perfectly stationary). And they’re a big pain in the arse (sometimes literally!). By 1999 faster film had become available and had become somewhat less grainy but as anyone who has ever used 200ASA Sensia will testify, 100ASA is still the realistic limit for top quality colour slides.

Two new technologies to the rescue

The photographer in 2003 has two new solutions to the low-light problem; image stabiliser lenses and digital cameras. Image stabiliser lenses can, the reports say, yield acceptably sharp results at much slower shutter speeds than ordinary lenses. Like tripods, however, they can’t help when the subject is moving. Digital cameras (but only a select few of the many currently available) can yield acceptably noise-free images at much faster film speeds than real film. Combine the two technologies and the camera has suddenly, after many decades of being constrained by the same old problem, become genuinely closer to being perfect.

Perfection in the 21st Century

Bizarre fetishes and global holocaust aside, it is unlikely that there will be any film-based cameras in production at the end of this century. Evolution is inevitable and the next evolutionary step for cameras is digital. But in these early years of this millennium the only interesting digital cameras are saddled with technology from the last. There are only a few digital cameras which can be used at higher film speeds than real film with acceptable noise levels, and in each case all their key components are taken from film-based predecessors.

The next stage in the evolution of the professional quality camera is the digital camera which is designed from scratch to be digital. The new Canon EOS 10D apparently has a camera body designed to be digital from scratch, but shares its lenses as well as most elements of its design with film-based predecessors.

 

Abandoning constraints in the digital age

Because of time constraints and market forces, all the best current digital cameras are SLRs. Until very recently the only people who could afford these cameras were professionals. These professionals are accustomed to using SLRs, and in most cases have a collection of existing film-based camera bodies and lenses. Also, designing new lenses takes time, so in order to get something onto the market which people would buy, camera manufacturers made digital SLRs which are compatible with current lens ranges. But time constraints and market forces aside, are there any convincing reasons for making digital cameras which use lenses which were designed for 35mm film cameras? And are there any convincing reasons for retaining the SLR design?

The goal of WYSIWYG

The advantage of the SLR over other film camera designs is that, to a certain extent, what you see through the viewfinder is what you get when the film is developed. However, what you see is not EXACTLY what you get, because the viewfinder takes no account of:

(1) The film type;
(2) The aperture and shutter speed; and
(3) Exposure.

The viewfinder image will still be in colour if you’re using black and white or infra-red film. The viewfinder image will not be grainy even if you are using very grainy film. Night lights may look white through the viewfinder but come out green on the film.

The viewfinder image shows what the lens sees at maximum aperture, not the aperture which is going to be used for the photo. Cameras with a “depth of field preview” can display the actual depth of field through the viewfinder but the image gets so dark it is difficult to judge how the final picture will look.

The viewfinder does not take exposure into account. Things may look fine through the viewfinder but the final result may be washed out through overexposure or too dark through underexposure.

One further problem is that with a very few exceptions, SLR viewfinders don’t show 100% of the image area. So unless you own a very expensive professional SLR you might carefully compose a shot so that an offending piece of fence is just out of view only to see the very same fence make an unwelcome appearance on your slide.

An electronic viewfinder could, potentially, solve all these issues and take the viewfinder one step closer to being truly “what you see is what you get”. The technology has some way to go, though. I have seen electronic viewfinders on current “consumer” digital cameras and they are not remotely acceptable. But the eventual dominance of the electronic viewfinder is almost as inevitable as that of digital cameras themselves.

As well as true “WYSIWYG” there are other advantages of moving away from SLR technology. Lenses could be made smaller and lighter. Lens design could take advantage of the digital nature of the camera. Image stabilisation technology in the lens could be coupled with similar technology in the camera itself, leading to a further reduction in the number of scenarios which require a tripod.

The goal of perfect exposure

One of the key skills in photography today is exposure technique. But the need for such a skill is a consequence of imperfect technology. If film had infinite dynamic range (or exposure latitude) the photographer wouldn’t have to worry about exposure. He could just concentrate on composition and shoot at whatever aperture he desired for ideal depth of field or optical quality. Decisions as to how light/dark/contrasty the image should look could be postponed to the “darkroom” stage. Modern colour print film has a wide dynamic range, enabling this technique to be adopted to a certain extent. But even colour print film does not have as wide a dynamic range as the human eye. A digital sensor could potentially equal or exceed the dynamic range of the eye itself.

Current digital sensors work by exposing a number of light-sensitive areas of a silicon chip to light for a period of time and measuring electrical characteristics at the end of the period. Those electrical measurements for each “pixel” correspond to how bright the image is at that point in the image. The problem with this is that overexposed areas get “washed out” (or “blown out” as the Americans call it) easily where the measured values hit their maximum. Fuji have recently introduced a chip which has two sensors per pixel of different levels of sensitivity, so that if the more sensitive one reaches a maximum value, a meaningful result can be taken from the less sensitive one. Another way to get meaningful values from “overexposed” sensors would be for a timer to be integrated into the silicon chip for each pixel. Each pixel’s timer could start at the beginning of the exposure, and stop once the sensor reached saturation, if that happened. The “highlight” information could then be reconstructed by the camera using the time values, and the rest of the image could be dealt with in the conventional way. Whether future digital sensors use this or other techniques, I believe practically “infinite” dynamic range ought to be a possibility for a digital sensor. It’s just a matter of making the light-sensitive silicon chip sufficiently clever.

Making the expensive parts cheaper

Photographic equipment, having entered the realm of the microprocessor, is now “governed” by Moore’s Law. Unlike real “laws” (think Newton and Einstein), Moore’s Law should really be called Moore’s “rule of thumb”, but it has proved remarkably accurate. Translated into the photographic world, it means that every couple of years the resolution of digital “film” will double. And it seems to apply exactly as expected – witness April 2000’s Canon EOS D30 (3 megapixels, £3000) and January 2002’s EOS D60 (6 megapixels, £2000). 12 megapixels for £1500 in January 2004? That may seem amazing but would only be a simple application of the “law”.

However, whilst the digital “film” improves with the same speed as the microprocessors in our PCs, the rest of the camera lags behind. The optical equivalent to Moore’s law (Zeiss’ law?) goes something like this – every 5 or so years optical quality will improve very slightly and lenses may also get a tiny bit lighter although prices will by-and-large stay the same.

But with digital cameras, especially with electronic viewfinders, perhaps it is possible to drag “optical” quality into line with Moore’s law by leveraging as much of the image quality burden as possible onto the electronics? This is the same technology as was used in the Hubble space telescope to use powerful electronics and software to compensate for the telescope’s optical deficiencies.

Vignetting? No problem – the digital “film” has infinite dynamic range and the vignetting can be compensated for in software. Distortion? Never mind – each lens can be accurately measured in the factory and all its known distortions can be compensated for in software, in “real time” so you don’t even see distortion through the viewfinder. Extreme wide-angles? No problem – make them the cheaper fisheye design and correct the distortion in software. Digital technology can thus improve upon the ultimate quality of top-end cameras, and (perhaps more importantly in economic terms) make ordinary consumer cameras lighter and cheaper by putting the strain on the electronics rather than the optics.

With film, the photographer has always been trapped between trade-offs. To get large depth of field, we had to use small apertures, and hence long shutter speeds. Large depth of field AND short shutter speed? No way, with film. But perhaps you could do it with digital? Infinite dynamic range would mean you could choose the optimum aperture AND shutter speed independently. And once you have shed the constraints of an optical viewfinder you open up the possibilities of weird technology like Wavefront Coding, which achieves extensive depth of field without small apertures.

Some 2003 camera types and their future replacements

Disposable camera / cheap “point and shoot”/ APS

Replaced by built-in camera on your “mobile phone”. This will not just be a phone and a camera, but a personal organiser, personal stereo, GPS system and “rape alarm”. The ability to instantly send a good quality colour photograph/video footage of an assailant together with precise location information to a police control centre will prove a valuable deterrent to crime, making these “mobile phones” the next best defensive weapon to a gun, provided the batteries are charged…

Quality 35mm point and shoot / “consumer” SLR

Replaced by increasingly higher quality variants of current digicams.  I expect them to be smaller than film cameras - along the lines of Ultra-compact digital cameras such as the Pentax Optio S or Minolta Dimage IX.

Serious amateur/ professional 35mm SLR

Initially replaced by high-end digicams and digital SLRs using lens ranges from film-based SLRs, but eventually superseded by interchangeable lens system cameras with electronic viewfinders. Not being SLRs, the body and lenses would have dimensions closer to those of the Contax G3 than to current 35mm SLRs. Sensor size would tend to be around APS size, which would also help contribute to reduced size and weight.

Medium and large format, including panoramics

Replaced by specialist interchangeable digital backs and lenses. Sensor size would vary from 35mm size up to perhaps 6x9.  Canon's D1s has already shown that medium format quality is possible from a 35mm sized sensor, so my prediction is for "medium format" type cameras to be replaced with something more like 35mm size.
 

Conclusion

Never be tempted to think "digital cameras have got about as good as they're going to get".  There's a big difference between good enough and as good as it gets.  Maybe in 2003 digital cameras are good enough.  But expect better to come.  Much better.

Acknowledgments

I have borrowed the wonderful camera sketch at the top of this article from the Canon Camera Museum - it was an early design for the Canon T90!

  -----   (c) Philip Harle 26 May 2003   -----