Kodak DCS 460 I

This is the first part of an article about the Kodak DCS 460, an early digital camera that came out in 1995. The article slots in between my history of Kodak's DCS cameras, and my article about the Kodak DCS 520, which was part of Kodak's next generation of digital SLRs. Between the DCS 460 and the DCS 520, which came out three years later, in 1998, there is a massive gulf. The DCS 520 resembles a modern camera, with an integrated body and an LCD screen. It would not turn heads today. In contrast, the DCS 460 looks like a film camera that has been attached to a huge motor drive. It does not obviously look like a digital camera; there is no screen on the back, and the digital module looks as if it comes off (which it does). It is strange and primitive today in a way that the DCS 520 is not.

In the first part of the article I will cover the camera's historical background and I will talk a bit about the camera itself. In the second part I will talk about the camera's potential applications, and the image quality of its ancient six megapixel sensor. In the third part of the article I will talk about the process of interfacing it with a period computer (an Apple Macintosh Powerbook 3400), using Kodak's vintage DCS Acquire software to capture and process an image. In the fourth part I will attempt to burn a hole through a teddy bear using a magnifying glass, and I will not tell you what is in the fifth part except to say that it involves deep hurting.

This is not the first article on the internet about the DCS 460. A man called Gisle Hannemyr owned one for a few years, and wrote about his experiences here. His article is much less waffly than mine. Jarle Aasland at Nikonweb has an interesting article on the DCS 460's lower-res brother, the DCS 420, at this page here. I do not know why the two best and most comprehensive articles on Kodak's earliest digital cameras are from Scandinavia. I myself am not Scandinavian, I am English. The thought of running around naked in the snow is alien to me. The mysterious "leofoo" at Photography in Malaysia goes into a lot of technical nitty-gritty here. Kodak's website still has a support page for the camera, with a downloadable manual and firmware updates. For whatever reason, Professional Marketing Services, Inc has a Kodak DCS 460 brochure available for download here, along with brochures for most of Kodak's other DCS models.

It is interesting to ponder the path that my camera, serial number 460-2950, has taken. The Nikon N90s body was made in Japan, and was then shipped to Eastman Kodak in Rochester, New York, the USA, where the digital back was attached to the camera. Kodak's assembly line was, in those days, almost like a craft shop, which each camera put together by hand, by skilled technicians. At some point my camera made its way across the Atlantic to the UK, where it was presumably used in a studio by a professional photographer. It then outlived its usefulness and was put up on eBay. I cannot tell when the camera was made. Its serial number is relatively high, and so I imagine it came off the assembly line some time in 1996 or 1997. The first DCS 460s were made in 1995, and the line was finally cleared out in June 2000.

I will now repeat the one fact that everybody knows about the DCS 460; it originally sold for $28,000, according to this contemporary article from the New York Times. I have seen other figures - $35,000, $15,000 - but the New York Times is a proper newspaper and would not make a mistake. Some people were prepared to pay that sum, but I surmise it only had a limited market outside of government or scientific institutions. When Kodak got rid of the last DCS 460s, in 2000, they were sold off for $2,500. In 2000 that was about $500 - $1000 less than the contemporary, three megapixel Fuji S1 Pro and Canon D30, which were more capable cameras than the DCS 460 in every aspect except for resolution.

Nowadays it is possible to pick up a second-hand digital SLR that is inarguably superior to the DCS 460 for less than £200, such as for example a Canon 10D or a Fuji S2 Pro or a Nikon D70. The DCS 460 is a historical novelty with very little practical use, although as Mr Hannemyr points out above its output is still sharp and smooth in good light. The DCS 460's native resolution of 3060x2040 is only just short of a 300dpi A4 print, and the camera's lack of an antialiasing filter means that its images are naturally very crisp, without obvious sharpening halos, and scale up very well. I hesitate to make pronouncements about digital print sizes. It is a controversial topic, and I rarely make prints.

I begin. But before I do, I would like to apologise for the fact that this article contains so many numbers, particularly towards the beginning. Digital cameras do not as a rule have names, they have numbers, arbitrary meaningless numbers that baffle non-specialists. I suspect that this bafflement is intentional. In the distant past those cameras that had names were aimed at the consumer market; there was the Spotmatic and the Hawkeye and the Brownie and the Retinette and the Canonet and the Pen, and so forth, all aimed at people such as you and I. At the same time, the more professionally-orientated cameras tended to have numbers, or short alphanumeric sequences, such as the III, or the A, or the IIa, or the P, or the F, and so forth. At some point in the late 1970s numbers and letters came to seem more professional and grown-up than names, and by the end of the 1980s cameras no longer had names at all. If you wanted to buy a good 35mm SLR in 1986, you had a choice between a Canon T80 or a Nikon F501, or a Minolta 7000, or a Pentax P50, an Olympus OM-4, and no doubt several others that you would need a book to tell apart. Perhaps the industry had simply run out of good names. The Minolta 7000 was sold as the Dynax and Maxxum in foreign markets, and nowadays Canon's consumer digital SLRs are sold as the Kiss or Digital Rebel in Japan and the USA, and with names like that, I prefer numbers.

Kodak's professional digital SLRs were never likely to have names. There was no chance of that. They were aimed at the high-end professional market, and they were based around cameras that did not have names themselves. Kodak's digital SLRs were esoteric technical gadgets created by boffins who did not smile or love or know the touch of a woman. If DCS 420 and NC2000e mean nothing to you, and you cannot remember which camera is which, simply skip the next few paragraphs and look at the pictures. Let the words wash over you, let them bypass your conscious mind and penetrate straight to your immortal soul.

All of the pictures in this article were taken with a DCS 460, mostly using a Nikon 50mm f/2.0 AI lens and an infrared blocking filter, a few with a Sigma 15-30mm. The pictures of the DCS 460 itself were taken with a second camera that I shall not identify by number, because I call her Salotine MacClacky.

In Context
I begin, again. The DCS 460 was launched in 1995 as the high-resolution twin of Kodak's popular DCS 420. It was conceptually and physically identical, except that it had a much larger sensor, with a higher resolution. The DCS 420 - and the closely-related Kodak NC2000e, a special model prepared for the Associated Press news agency - is an interesting topic in its own right. It was essentially the first digital SLR that photojournalists adopted in large numbers. On a technical level it was an upgraded successor of the Kodak DCS 200, with two important modifications; firstly, it added a removable storage slot that could accept PCMCIA cards, rather than being restricted to a fixed internal hard drive, and secondly it used one single battery rather than having separate sets for the camera and the digital module. Unfortunately Kodak made the odd decision to bury the battery pack inside the camera body, where it could not be removed without unscrewing the bottom of the camera. This was a problem when shooting lengthy sporting events, and Eamon Hickey covers this exceptionally well here. I will not duplicate his words. As far as I can tell the DCS 460's battery pack is made of standard rechargeable AAs, but they are ganged together and cannot be replaced without a certain amount of electrical know-how. The pack is charged while still inside the camera.

I have not scientifically tested the DCS 460's battery life; with the batteries I have, it seems to take about 100 shots before the indicator starts blinking. The digital back draws current all the time, even when the camera is turned off, and so the batteries tend to drain away unless they are disconnected from the camera. This seems to have been a problem with Kodak's cameras all the way up until the DCS 14n of 2003. In the days before NiMH batteries it meant that owners of DCS cameras had to undergo a bothersome routine of periodically charging and conditioning their batteries lest they discharge so much that the Kodak charger would no longer charge them up again. That's enough about batteries.

The DCS 460 was sold in parallel with the Kodak EOS DCS 1, which used the same sensor and electronics, but was based on a Canon EOS 1n film body. They were essentially the same cameras and most of the information in this article applies equally well to the EOS DCS 1. Glamour titan Ken Marcus used to use a DCS 1, although in general it seems to be even more obscure than the DCS 460. Kodak continued to use the 1n body for their subsequent Canon-based digital SLRs, the Kodak DCS 520 and DCS 560, and I have written a bit about the 1n in that context here.

The DCS 460, on the other hand, was based on the Nikon N90s, a 35mm film SLR camera that was, at the time, Nikon's leading semi-professional autofocus SLR. I am not sure why Kodak did not use the Nikon F4, which was Kodak's top professional model in 1995. All of Kodak's other early digital SLRs were based around high-end bodies, and there had already been a special digital adaptation of the F4 for NASA. Perhaps the F4 did not lend itself to mass-produced digital conversion, or perhaps Kodak decided against using a body that was on the verge of being replaced. The Nikon F5 was launched one year later, in 1996, and formed the basis for Kodak's subsequent Nikon-bodied DCS 600 and 700 series.

On an aesthetic level the N90s looks very dated by modern standards. The shiny black plastic and angular lines place it in the 1980s, although in fact it was launched in 1992. Nonetheless the basic design was derived from the N8008 of 1988, the year of Loadsamoney, and of Rick Astley.

The late 1980s haven't aged well. I can remember, when I was a kid, thinking that Loadsamoney was the funniest thing ever. Now he seems as unfunny as Dick Emery did to me when I was a kid. And no doubt adults in 1988 thought that Loadsamoney was as tiresome and bland as Charlie Drake. Comedy doesn't age well. People change, and tastes change, and standards of decency change; the essence of comedy is that it violently subverts our expectations, but in an age when men go hatless in public there is no baseline of normality to subvert. I confidently predict that comedy will die out in the next few years, to be replaced by a humanistic form of entertainment based around the mutual self-reinforcement of the spirit.

Still, I have no idea who designed the Kodak digital section, but it was clearly created so as to blend in with the N90s, and it is surprisingly successful in that respect. It looks like an official Nikon accessory. It's interesting to compare it with the EOS DCS 1, which blends well with Canon molten plastic design; clearly someone at Kodak cared enough about their digital cameras to make them look good.

The Camera
The N90s uses Nikon's screw-driven autofocus system, and is compatible with a wide range of AI and post-AI Nikon lenses, with a number of caveats. Notably, the G-series lenses, which do not have an aperture ring, will only work in program and shutter mode, because the N90s cannot otherwise set the aperture. Nikon's designed-for-digital DX lenses will mount and work, but they do not project a large enough image circle to fill the DCS 460's sensor, and so they will produce vignetting. I surmise that they will vignette less than they would on a full-frame SLR - the DCS 460's sensor is relatively large - but I have no way to test this. I only have one Nikon lens. Before using the DCS 460 I was unfamiliar with Nikon's cameras. The N90s is logically laid out, and I am impressed with the way that it integrates with pre-autofocus Nikon lenses (the 50mm in the photograph automatically stops down). Canon's modern bodies do not integrate with pre-autofocus Canon lenses at all, indeed the lenses are not even optically compatible. Despite being a consumer model, the N90s feels durable. The controls have a professional-style "hold down a button and simultaneously turn a dial" arrangement, similar to the EOS 1n, but generally simpler. The ISO setting, for example, has a dedicated button, whereas the EOS 1n used a two-button combination. I will talk more about the ISO later on.

The film version of the N90s had an integral motor drive that wound the film at up to 4.3 frames per second. The DCS 460 does not have a continuous shooting mode. It ignores the camera body's continuous shooting lever, and if you hold down the shutter button and keep it held down the camera will only take one shot. Judging by my own informal tests, which I performed by repeatedly tapping the shutter button whilst looking at a stopwatch, the DCS 460 will take a shot, pause for two seconds, take a second shot, and then pause for a further eighteen seconds until it takes the next shot. This is with a 1gb Sandisk "industrial grade" PCMCIA card. I cannot judge its performance with a Compact Flash card, because I have yet to find a combination of card and adapter that work consistently and reliably.

Did I mention that the DCS 460 does not have a screen? The earliest compact digital cameras and SLRs did not have screens, and neither does the DCS 460. Instead it has a small LCD panel that shows the remaining frames on the memory card, using numbers and a pie chart. It also shows the battery status, the SCSI ID number, and some error codes. The back has three buttons. One deletes the last image, one sets the SCSI ID, and a third, when held down, turns on the camera's built-in microphone. You can record an unlimited number of 25-second audio snippets, which are recorded as 8-bit .WAV files. The lack of a screen keeps the camera's power consumption down, and there is one less thing to go wrong, but there are no other positives about not having a screen. Film photographers got by without screens because film has a relatively wide tolerance for poor exposures, especially negative film, and doubly-especially negative black and white film. In contrast, the digital sensor in the DCS 460 does not tolerate poor exposures very well. If the image is underexposed, it will have lots of shadow noise when pushed up more than a stop. The camera deals surprisingly well with overexposure - it is at least on a par with modern cameras in this respect - but without a histogram it is hard to "expose to the right", as per the fashion. In general I find that the camera body gets the exposure right when indoors, and slightly underexposes when outdoors. If you really set out to create consistently well-exposed photographs with the DCS 460 you will at least learn something about exposure, I will say that. Kodak's next generation of SLRs had LCD screens, which were grainy and dull, but invaluable because they also displayed a clearly-legible and easily-interpreted histogram.

The Sensor
On a historical level, the DCS 460 is a freak, an aberration. Its specification was far in advance of competing portable digital cameras in 1995. The camera was based around a new imaging sensor, the Kodak M6, which presumably grew from Kodak's work for the US spy satellite programme. I cannot think of any other reason why Kodak would go to the time and trouble to develop a six megapixel sensor in the mid 1990s. There was no "megapixel war" back then. In 1995 the most advanced consumer digital cameras had a resolution of 1024x768 (0.7 Mp) at a push; the only competing digital SLRs - essentially the Nikon E2, the Agfa Actioncam, and the Kodak DCS 420 - were all roughly 1.2 Mp, give or take a few hundred pixels. The only contemporary digital camera that I know of with a higher resolution, that took pictures with a single exposure rather than scanning the image line by line, was the Dicomed Bigshot, a 16 megapixel medium format back with a full-frame 6x6cm sensor. It was even more expensive than the DCS 460, and that was without taking into account the cost of the Hasselblad 553 ELX camera that it was designed to work with. Its specification is still interesting today although I have no idea how its image quality compares with modern cameras, and there is very little about it on the internet. The only large image I can find - a terrifying shot of a woman - suggests that it had a problem with shadow noise but was otherwise very impressive.

Kodak's M6 sensor measured 27.6 x 18.4mm, slightly smaller than the APS-H sensors used in Canon's modern 1D range, slightly larger than the sensors used in most modern consumer digital cameras. In its day it was a marvel of technology, and won at least one award (the Technical Image Press Association's "Best Digital Imaging Product in Europe, 1995-96"); in the context of digital SLRs, its resolution was not surpassed until as late as 2002, when Canon introduced a CMOS sensor that had a few more pixels in each dimension in their D60. This is not to say that technology stood still between 1995 and 2002. The imaging system in Canon's D60 could be pushed from ISO 100 to ISO 1000, and to my eyes is less noisy at ISO 400 than the DCS 460 at ISO 80. The D60's sensor could also take four-minute exposures with no more noise than a split-second snapshot, whereas the M6 has trouble with bright white "hot pixels" when the shutter is opened for more than half a second or so. Furthermore, the M6's colour channels respond differently to noise - green is very clean, red is noisier, and blue is noisier still - with the result that images taken in poor light and then pushed tend to have odd colour splotches where the blue or red channels did not capture a signal.

Kodak developed the M6 over time. As far as I can tell, they still sell a version of it, the monochrome-only Kodak KAF-6303e, although it only seems to be used in astronomy cameras. In a conventional photographic context the final iteration of the M6 appeared in the Kodak DCS 760 of 2002. The DCS 760 was essentially a distant descendant of the DCS 460. It was a formidable camera in its own right, built like a tank and weighing only slightly less, but it came out a year or so too late, in the midst of a glut of cameras that were just as formidable, but cheaper and smaller and generally more flexible. The version of the M6 that appeared in the DCS 760 could be pushed up to ISO 400, but the results compared poorly with those of contemporary cameras, and the sensor and Kodak's DCS series and Kodak's SLR range as a whole had clearly reached the end of the line. It seems to me that the M6 was ultimately a dead-end for Kodak. By 2002 the company must have come to the conclusion that that sensor could no longer be developed economically, and that the returns from selling professional digital SLRs did not justify the expense of research and development on a successor. Kodak's next and final generation of professional digital SLRs used a different sensor from an outside manufacturer, with very mixed results.

Despite its high resolution, the M6 had a lower pixel density than the smaller sensor used in Kodak's press models. In theory this should have meant that the sensor would be more sensitive to light, but in practice the opposite is true. The DCS 460 at ISO 80 seems to have more noise than the samples I have seen from the DCS 420, which ranged from ISO 100 to ISO 400. It is certainly more noisy than ISO 200 from my Kodak DCS 520, albeit that the DCS 520 was part of Kodak's next generation of cameras. Perhaps the M6's pixels were no larger than those in a conventional sensor, but were simply spread out over a larger area, or perhaps the larger size of the sensor introduced more noise into the imaging pathway, or perhaps the camera's electronics had to work harder to extract the image data from the sensor. Perhaps the camera was unable to perform the kind of noise reduction that goes on inside modern SLRs.

In the early days of digital SLRs there was a lot of worry on the internet about dusty sensors. The imaging sensors in digital cameras have a static charge when they are in use, which attracts dust and hairs and so forth. The dust and hairs tend to linger because the sensor is fixed in place, unlike film which was wound on from frame to frame. Nowadays most cameras shake off the dust by vibrating the sensor, and a whole industry has emerged that sells air blowers, anti-static cleaning wipes, brushes and so forth. All of this was unheard-of in 1995, and still very novel a few years later. When Nikon released their D1 in 1999, the company's official advice was that photographers should only have the sensor cleaned by an authorised Nikon service centre. Kodak were far less squeamish. Their official cleaning procedure, as outlined in the DCS 400-series manual, is the most drastic I have seen outside the world of medium format backs; the company recommends removing the Nikon N90s camera body from the sensor, and gives instructions on how to do so. This leaves the sensor completely open to the elements, which makes cleaning a doddle, although there is the risk of scratching or cracking the filter layer on the sensor's surface.

The Nikon N90s can be entirely removed from the digital unit and turned back into a film camera. I have not tried this myself. The N90s unit still has its motor wind and film spools and so forth, although these are completely superfluous in a digital context. Once removed from the digital back, the N90s needs a new battery holder, some film, plus a small pin that fits into a slot just above the film rail and tells the camera that film is present. The camera also needs a new back door. In theory a standard N90s back would do the trick, but given the fact that 35mm equipment is now available cheaply second-hand, a Nikon MF-26 databack is a more interesting choice. This adds a few things that the DCS 460 (and the basic N90s) cannot do, such as multi-exposure bracketing, an intervalometer, and timed exposures of longer than thirty seconds (up to 99 hours, 59 minutes, and 59 seconds!). Kodak's subsequent SLRs, starting with the DCS 520, added bracketing and an intervalometer, but none of them could take such lengthy exposures, and indeed the DCS 460's sensor would have made this pointless even if it was possible.

There was also a vertical battery grip, which would solve the problem of sourcing a battery holder. In practice, a second-hand Nikon N90s with battery grip and databack sells on eBay for less than the cost of a DCS 460 plus the grip and databack, so it's not economical to buy up old DCS 400 cameras and convert them to film. And for not much more, you could buy a used Nikon F4. In fact for not much more than that you could buy a used Nikon F5. Film cameras have crashed in value over the last few years.

In part two of this article I will lay down some word on the DCS 460's potential applications, with yet more pictures.