A question I am frequently asked – ‘what flat bed scanner do you recommend I should buy?’
This is a nightmare of a question and one which is not easily answered without first responding with the question ‘what do you know about image post processing and pre-press?’, followed, usually in my case, by the whole book preamble on the history of photography. It serves little purpose. Most individuals in my experience of this subject, are impulse buyers who have swallowed the manufacturer carrot and purchase on the basis of product appearance and little understanding of the mechanics or specification. By the time I have gotten to Niepce, listeners are already bored stiff. You know the glazed eye look…
This aside, there are novices with a genuine interest and limited budgets to spend. So here goes.
Several years before I purchased my first flat bed scanner and at a time when I was building digital stock for the future digital archive, I used a local lab which had invested (early to mid 1990s) in some expensive kit. They had Agfa digital film recorders, a raft of Apple Mac computers, a Scitex flat bed and top-end Umax scanners. With all this stuff, the lab could meet the professional quality demands of its advertising agency client requirements – and believe me, they had some big players in the game.
I used their Scitex and Umax scanning services mainly for digitalising medium and large format film. Costs ran at around £10 – £20 per image, depending on set-up time. This was expensive and I only used the service when I knew I could sell the image and have the client pay for everything on top of the repro fee. Still, it was cheaper than another local lab which offered laser drum scans.
In time, I acquired my own flat bed scanner; a Microtek, which I was running on an Apple Mac Performa-Pro (1993). The only time you could tell you had done things right was when the magazine came out; my Performa screen had pitiful resolution and this combined with some dreadful scanning software was a non starter when it came to visualising what was right or wrong. It was very much seat of the pants scanning; you just crossed your fingers and hoped. And most of the time, reproduction was acceptable, only occasionally could I tell it had been a really long night on the pixels.
Soon I graduated to an iMac G3. The screen resolution was higher and a lot clearer. I bought one of the then new Heidelberg Linoscan flat bed scanners with a tranny hood, having first investigated the software it was bundled with and sweet-talked the distributor into demo scanning a stack of 120 format Kodachrome’s – yes, in the 1990s, my preferred reversal stock was available in medium format.
The demo scans were not perfect but the difference in quality between them and what I had been paying an arm and a leg for off the Scitex or Umax, was marginal. The test would be in the final repro quality and after a few weeks I had sold enough and seen the results to have a pretty good idea of what the machine could do.
Life went on. I scanned and scanned and spent a lot of time making up special holders for old glass plates and larger format transparencies to avoid the problem of Newton’s Rings. I hardly thought about what went on inside the box until one day, the lamp used for reflective scans gave out.
To fit the new lamp, follow the maker’s instructions by undoing two retaining machine screws and lift off the plastic glass platen frame. It was then that it hit me.
Light travels in straight lines right?
Right. Except when it’s inside a flat bed scanner.
Then it gets bent all over the place by thin mirror strips before finally being focused through a tiny objective onto a CCD.
Let’s just have a short think here about what we are dealing with. Light, imaging and scanning. A sort of electronic enlarger. The principle is almost the same. Light is passed through film and its rays are collected somewhere in the box to make a digital image.
Now, if I recall, the old film enlargers I used in my darkrooms in decades past, all had their insides well painted with a matt black, non light reflecting paint. Indeed, in high quality machines, light baffling was so sophisticated, one never had any concern for leakages. More attention was paid to the darkroom door.
But what have we here in the average flat bed scanner? Well, light is transmitted first through two large sheets of platen glass (one in the tranny hood, one on the bed) between which are sandwiched the negs or trannies being scanned. These are glass sheets the size of foolscap paper (or larger on A3 models.) Hold one sheet up to a lamp and anyone can see what it does to light transmitted through it – scatter, scatter,scatter. And lo, the whole of the interior box of the machine is a nice shiny grey plastic. All the bits of metal and other plastic used to house the lights are all highly reflective. Even the black plastic bed on which the thin mirror strips are mounted for collecting the reflected rays and bouncing them to another mirror strip before transmission to the lens, is all shiny and new.
What were the dumbells who designed this machine thinking of? Rhetorical. This Heidelberg turned out to be a rebadged Umax – made in Japan. I suspect many other similar products are put together the same way, with little thought given by designers or engineers to the real purpose of the device.
The question trickling through my mind at the time was, what happens if I paint all of the accessible shiny parts with matt black paint? Would this have any effect on improving image quality? Would it push up contrast; would it increase the dynamic density range?
There was nothing to lose. Out came the matt black Humbrol, masking tape and a paint brush. Soon, all of the parts I could reach were matted. On some of the edges of black plastic mouldings where it looked as if light rays would almost certainly be reflected back but which were difficult to reach with a loaded brush, I used double side tape and strips of black velvet from the gates of old film canisters. When I figured I had done enough, I cleaned the insides of both platen glasses and reassembled the machine.
Immediately, there was a noticeable improvement in the final scans, particularly in the area of shadow detail where the Heidelberg had not previously been able to extract much detail. Contrast went up and with it, apparent sharpness. So far so good. Scan quality improved as I learned to tweak the finer points of the bundled LinoColor software; clients were happy.
A note about the software.
LinocolorElite was put together by a company whose origins were deeply rooted in the British printing industry – remember Linotype machines? The German company Heidelberg bought Linocolor, marketed a raft of hi-end and low end rebadged Umax scanners with bundled Linocolor software and then jumped out of the business. Used Heidelberg machines can still be found but the better specified machines from the 1400 model up usually only run on old SCSI connections.
But I digress. The point about this is that most current flat beds have not significantly improved in mechanical design at the same rate at which resolution capacities have increased and while reproduction quality generally looks acceptable, I don’t see the giant leap. Visual appearance depends on printing methods, paper, inks and all the rest of that side of the business, but what I want in an ideal world is the near perfect scan and for this, one usually has to pay a premium.
Just recently, I noticed a slight fall off in scanned image quality while digitalising a batch of old b+w prints. One side of the image was o.k. the other side slightly fuzzy, as if someone had sprayed the screen image with diluted milk. I cleaned the monitor. It made no difference. An inspection of the flat bed platen glass provided no clue as to the cause.
In a fit of pique, I stripped the machine again. The thought of micro bugs similar to those sometimes seen in a camera viewfinder crawling all over the three line CCD low pass filter and leaving their minute traces of dirt, was uppermost.
I could neither find nor see anything. The interior was spotless, except when I shone a high powered torch onto one of the thin collecting mirrors; a barley visible thin patch of surface blooming like a small cloud such as would be visible on a steamed up lens, was the obvious culprit.
I suppose that even if one worked these machines in a scientifically audited clean room, foreign matter and atmospheric pollution would, in time, affect component performance. In the real world home or office environment, the risk of attack over a short period of time is much higher. I still do not know what caused the blooming, but for good measure I cleaned all the mirror strips and the collecting lens with impregnated tissues supplied by Carl Zeiss. And while I had the top off the mirror box, I added more velvet strips to parts which looked vulnerable to light scatter and blacked out the inside of the fluorescent light tube housing with more Humbrol.
The mirror cleaning cured the original problem, and now there was also another hike in both contrast and apparent sharpness levels. An acquaintance in the USA with access to special measuring equipment reported a recent set of scans from 5X4 inch Fujichrome originals having only a 1.2% colour balance error; better than the 2-3% normally experienced from scans produced on high end machines and far better than the 10 -20% errors often experienced on off-the-shelf consumer products. And this from the raw un-post processed scans made using my own ICC profile for the film.
Now I have a new project in hand.
I purchased a second Heidleberg for spares.
I am now going to strip both it and the old Microtek and matt spray paint the parts which can be and remodel the mirror housing, lens and CCD assemblies with matt paint and whatever else it takes to stop the light scatter. Some solution of soft pliable baffles will also have to be found for the undersides of the platen glass, which I am certain is just one of a number of contributing causes of scanned image chromatic aberrations. The aim is to reach a scan quality level which approaches that obtained for medium format film on a Howtek drum scanner.