I recently bought a DCS 420 - for £25, which is fair enough - and it's a formidable thing. The top professional digital SLRs of today boast 100% viewfinder coverage but the DCS 420 trumps them utterly, because it has 260%
viewfinder coverage. Nothing from Canon can match that. Also, the sensor has over one million pixels
! Mine is based on an original Nikon N90 rather than the N90s on my DCS 460. Perhaps the later ones were better.
It doesn't have the DCS 460's shot-to-shot delay but in most respects it's much less practical. Without an infrared filter, the DCS 460's images are very occasionally normal-looking whereas the DCS 420 suffers badly from what appear to be internal reflections in the mirror box area, which tints the top and bottom of the image purple.
This internal reflection problem was discussed elsewhere on this site with some suggestions that the photographer paint or mask off the edges of the sensor mount. This sounds reasonable although I suspect the process would be futile unless there was a way of masking everything with infrared-absorbing paint. My personal theory is that the problem might be the sensor cover glass rather than the surrounding mount, or the tint is not because of directional reflections but a property of the way that light enters the sensor. My second theory is that it's actually amplifier noise rather than internal reflections (but having said that, a five-second exposure indoors doesn't have the same problem, whereas a 1/60th shot outdoors in the sun does
). I have a hot mirror filter, and the problem seems to be restricted to infrared rather than visible light. If I mount the hot mirror filter, the internal reflections are greatly reduced. BUT it could also be that a slight loss in contrast from internal reflections is much less noticeable than a slight purple tint from the infrared. That article about the NC2000 mentions this problem, and the use of gradients in Photoshop to combat it, so perhaps the NC2000 had the same issue.
I'm digressing anyway, the point of this post was to present an unscientific comparison between the output of Kodak's DCS Acquire module running with Photoshop 5.0 LE on a Mac Powerbook 3400 and a more modern version of Photoshop using Adobe Camera Raw on a PC. They were shot with a DCS 420 and a Sigma 28mm Mini-Wide with a Tiffen hot mirror filter and a polarising filter mounted underneath the hot mirror filter.
Kodak on the left or top, Photoshop on the right or bottom. I used "daylight" colour balance with DCS Acquire (which produced purple, dark, low-contrast images) and then used "auto colour" and "auto contrast" in Photoshop on the PC. For the ACR versions - which were naturally very green, probably because of the IR filter, I used "auto" colour balance and then "auto colour" and "auto contrast". It's not very scientific - I should post a series of images showing the various manipulations I undertook - but my goal was to try and produce something that looks at least a bit normal, and I judged that the effort involved in a scientific study would be misplaced when there is much in the world I have not yet done:
My overall impression is that Photoshop makes things green, and that when corrected the images retain greenness; DCS Acquire makes things purple. Oddly, the same purple look seems to have been carried forwards to as far as the Kodak DCS 14n, perhaps later. When shot indoors, away from the sun, the images can be made more normal.
The DCS Acquire module cuts off the top two and bottom two rows of pixels. Photoshop retains these, but the top two rows are a bit yellow because there is noticeably less signal in the blue channel. The DCS 460 has the same feature. As with the DCS 460, the different ISO values are software hacks, in the sense that e.g. ISO 400 on the DCS 420 produces a very dark image with Adobe Camera Raw, presumably because ISO 400 underexposes the image and instructs DCS Acquire to boost the image during raw development rather than boosting the camera's amplifier at image capture. As with the DCS 460 there is nothing to stop the camera from shooting at ISO 1600, although the results are poor (I wonder if the DCS 410, which was supposedly ISO 100 only, was the same?).
Apart from the colour, and without wishing to duplicate the webmaster's very own Kodak DCS 420 page, it has much more of a problem with moire than my D2000, and appears to be one of a handful of cameras that actually produce more usable images when the lens is not
I surmise that the DCS 200 was shockingly bad. It would appear that the only people who ever used DCS 420s (rather than NC2000s) were scientists who did so under controlled conditions. There are a couple of shots taken by the US Geological Survey here
which suggest that no-one was capable of getting a normal-looking image out of a DCS 420 except under the most controlled of controlled conditions. The thought of a portrait photographer circa 1996 forking out £8,000 for one of the cameras in the hope that he would churn out lovely-looking portraits of kids sitting on Santa's laps makes me feel awkward.