just doing a comparison between the two and was wandering if you guys
can shed some light into something:
Nikon
======
Light source
R, G, B and Infrared (IR) LEDs
Image sensor
3,964-pixel, two-line linear CCD image sensor
Color separation
Performed by RGB LEDs
Minolta
=======
Image sensor 3-line color CCD, 5340 pixels per line, primary-color
filter
Scan method Moving film, fixed sensor, single-pass scan
Light source White LED
The light source, what''s the purpose of the Infrared one in the Nikon?
The infrared source permits the scanner to detect the presence of dirt
and defects on the film since the emulsion of most colour films is
relatively transparent to infrared. The Minolta also has this
capability with an IR channel too.
Since the image sensor for the Nikon is 2-line linear does that mean it
scans 2 lines at a time effectively making the scan time faster?
Yes.
The minolta uses a white LED light source whereas the Nikon uses 3
(RGB) and the IR. How does this effect the image scanned?
Two effects both of which are quite subtle.
Firstly, in the Nikon, each row of pixels in all colours is captured
with the CCD in exactly the same position relative to the film - the CCD
is stationary and the colours of each pixel are captured by sequentially
operating the LEDs. Consequently, there is no relative misalignment of
the three colours or the IR channel, right down to sub-pixel level.
In the Minolta approach, the three CCD lines capture individual colours
for three *different* rows of the image. Thus the full colour
information of each row in the image can only be reconstituted by moving
the CCD relative to the film. This means that the colours may not
perfectly align with each other. Although any misalignment is likely to
be well below a pixel size, it could cause colour fringing on sharp
transitions along the long side of the frame (ie. vertical edges on a
landscape).
More significant, however, is the effect on ICE and how the scanner
copes with dirt and defects on the film. For example, dirt or dust may
well be significantly off the focal plane. Since the CCD must move
relative to the film to capture all three RGB components of the image
then the image of this dirt will appear in a different position for each
colour - unless the light source remains stationary (ie. a full frame
illumination, as in a traditional enlarger, rather than collimated or
diffuse strip which moves across the film with the scan head). Which of
these positions is actually closest to the infrared image, which will be
different again, is insignificant, since the mask used by the ICE
process to conceal the dirt must be large enough to span the image of
the dirt on all three channels.
So the Nikon approach enables a finer mask to be used to conceal any
defects or dirt on the film - fewer pixels around each defect are
"guessed" at. However, don't forget that the pixels are effectively
smaller on the Minolta, so this is not as significant as it would
otherwise appear - it is *very* subtle.
The second issue is colour purity and saturation. To appreciate this
you must first recognise that scanning the image and subsequently
displaying it, whether on a monitor or print, is an additional process
compared to viewing it directly as a slide or projected image. In the
traditional view of a projected slide, your eyes respond to the spectral
characteristics of the dyes on the film emulsion, in the reproduced
scanned image your eyes respond to the spectral characteristics of the
phosphors on your monitor or the inks on your printer. All of these
components, your eyes, the film dyes, the monitor phosphors and printer
inks have relatively wide spectrums which often bleed from one primary
colour to another. For example, the red dye on E6 emulsions usually has
a significant density in some areas of the green and blue parts of the
spectrum and so on. Of course the film has been designed to "balance
out" much of this colour impurity by selecting the appropriate densities
for the dyes, so that the final image looks similar to the original
scene. The same is true of the monitor phosphors and the printer inks.
However, when you then introduce the additional process of scanning the
image you need to determine how best to capture each of the individual
colours in the image. In the Nikon approach, they use coloured LEDs
which have a very narrow wavelength range, so that the density of each
of the film dyes is measured at those wavelengths - and only those
specific wavelengths. This permits a very high purity of colour to be
achieved at the scanning stage, since the bleed of one colour into other
areas of the spectrum is eliminated. The resulting data can then be fed
directly to the monitor or printer to stimulate the production of the
appropriate density of phosphor or ink with its specific spectral
spread. In short, this approach enables a near perfect representation
of each colour to be achieved within the limits of the colour management
of your monitor and printer etc. without any additional processing being
necessary. This is also the approach that most high end scanners use,
with the LEDS being replaced by even more spectrally pure lasers for
both scanning and exposing the print paper itself.
By comparison, the Minolta approach (along with most other scanners) is
to use a process which is more akin to a digital camera, with coloured
filters on the CCD itself. These filters, however, are simply dyed etch
resist and have a very wide spectral response - which also bleed into
adjacent areas of the spectrum. The CCD data produced for each colour
is the average response across the spectrum that the filter transmits -
multiplied by the density of the dye at each wavelength. The result of
all of this is that the raw colours are somewhat muted, being further
spread into each other by the scanning process. You might have noticed
that if you photograph a colour print then the colours of the photograph
are never quite as pure or saturated as the original - and this is the
same process. If you repeated it many times then the final image of the
colour photograph would have no saturation at all and would be a dull
off-neutral monochrome image. The additional stage of the scan is
exactly the same as one of those reproduction stages - so the raw scan
is somewhat muted colours compared to the original film. The scanner
firmware compensates for this by increasing the saturation of the
colours - achieved by matrix manipulation of the colour data according
to the known or ideal spectral characteristics of the CCD filters. The
end result may be an image which has a colour saturation, but not a
colour purity, as good as that from the Nikon approach. However, such
mathematical manipulation of the data results in increased noise. This
is because the matrix manipulation involves the subtraction of two or
more colours at different weightings - which always reduces the signal
to noise.
These differences are very subtle and may even be insignificant
depending on the other imbalances in your system. In any case, they are
certainly insignificant when compared to the other major difference
between the scanners - the resolution. The Nikon is a 4000ppi scanner,
whilst the Minolta is a 5400ppi device.
