Clinton M James said:
Hey Folks,
I was wondering if somebody could fill me in on the scanner with the highest
DPI available. I have seen many references to 4800, but not over that.
Current top of the resolution stakes in consumer scanners is the Minolta
5400SE, just recently revamped, at 5400ppi (Pixels Per Inch). The new
version of this scanner has only recently been released and very view
reviews are available, however it is known to use a different light
source from the original and have cut back on some of features, so many
consider the original a better buy if you can get it.
There is a very low cost film scanner available from Plustek that claims
7200ppi. You might, as I did, consider that at well below $200 this
would probably have a poor performance and fail to deliver thie claimed
resolution. Well, I can't say either way, but I have not seen a single
complaint from a single user on this forum - which is very surprising!
Even the best scanners get complaints and queries posted here from
people who don't think they work as well as they should. So either the
Plustek is not selling at all, is only being bought by people with low
expectations or it is actually delivering exactly what it says on the
box and all their customers are happy bunnies. ;-) You'll have to make
your own mind up, but for almost a quarter of the price of more widely
respected products, you might consider it worth a gamble - especially if
you continue to believe that resolution is your goal.
I want to scan some film negatives in and want the best resolution possible.
Scanning negatives requires a lot more than spatial resolution (ppi).
Negatives, in particular, place great demands on scanner accuracy -
particularly signal to noise ratio. This is because negatives compress
the contrast of the image as it is recorded, so when you come to scan
the negative and produce the final image, you have to stretch the
contrast of the film image back to recover the original scene contrast.
There are lots of other processes that go on at the same time, including
removal of the orange mask, and usually this is handled automagically in
the scanner driver for you, to a greater or lesser degree of
acceptability. Neverthless, all of this processing increases the
visibility of scanner limitations and defects much more than slides,
which record the image as a reasonable representation of the original
scene and thus require little manipulation.
So, for your requirements, resolution may not be the dominant parameter
that you initially considered it to be - signal to noise ratio and
channel uniformity are just as, if not more, important.
I have used 800 ASA and even 1000 ASA film in the past and I know these are
equivalent to much more than 10MP in a digital camera.
Know, suspect or have been told that?
I doubt that many 800ASA or 1000ASA 35mm results come close to those of
a professional 10Mpix DSLR - even if they are scanned at 5400ppi!
One of the reasons for this is that the pixel count in the image is not
a particularly good metric for performance - George Orwell didn't design
digital imaging systems so "all pixels are NOT equal" - and resolution
is not the only relevant performance metric. ;-)
For example, a 10Mp dSLR will have a frame size of around 2600x3900
pixels compared to a 4000ppi scan from 35mm film with 3800x5700 pixels
and you might think that this would give the better image. However the
dSLR will have a luminance MTF around Nyquist of about 60%, depending on
the lens used, whilst the film itself will only be around 25-30% MTF at
this same resolution. MTF is just a technical measure of how the medium
reproduces the contrast presented to it at any particular spatial
frequency - a similar concept to the frequency response of your audio
system. Then their is the issue of noise. The noise density on the
dSLR has a flat response throughout its spatial frequency range, while
on film it increases dramatically due to film grain - and this is more
significant with fast films of 800 and 1000ASA than it is with 50-100ASA
film. The noise determines how well that reproduced contrast can be
distinguished from mush. Then you have an extra set of optics in the
film process, further impacting the MTF. In short, the dSLR just has
'better' pixels. So a comparison such as 10Mpix from a dSLR with 22Mpix
or more from a scan of 35mm film can be very misleading indeed, even
though the pixels in the scan are tricolor, as opposed to Bayer matrix
monocolor pixels in the dSLR. In almost all comparisons of this type,
the dSLR will win hands down - there are exceptions, but not from 800
and 1000ASA film stock.
And I am almost lead to believe 4800 DPI is just over 5MP.
That seems to contradict your comparisons with 35mm film and 10Mpix
cameras. In fact, a 4800ppi scanner will produce full 35mm frame scans
with approximately 30Mpix, but since all pixels are not equal it is just
a salesman's urination contest.
Therefore, if i am going to buy a scanner, I want the best because I don't
wish to stuff aorund buying a better one years later and then re-scanning my
memories again and again.
The scanners you might want to consider are the Minolta mentioned above,
at 5400ppi with 16bits per channel, the Nikon V (LS-50), at 4000ppi with
14bits per channel, or the Nikon 5000 (LS-5000), at 4000ppi with 16-bits
per channel. Any of these scanners has more than enough performance to
get everything from 800 and 1000 ASA films and they will get almost
everything from much slower film too.
The Nikon's have some advantages over alternatives because of the light
source that they use - each colour is captured on the same CCD line with
the image being illuminated in sequence by three coloured LEDs. This
gives exceptional colour purity and separation. Other scanners use
tricolour CCDs with white illumination, where each colour of the image
is captured with a different CCD line under a different colour filter.
The filters are less than perfect, bleeding into each other, requiring
some matrix manipulation to separate out, thus reducing the signal to
noise ratio. Of course the filters built into the film emulsion exhibit
the same type of spectral impurities, but at least the Nikon approach
doesn't make it any worse.
So, if you get nothing else from this message, there is a lot more to
getting good scan from film than resolution.
I basically want to scan them as best I can possibly. I will not even be
saving them in jpeg because I know jpeg is lossy for quality of image.
Good intentions but, considering that jpegs compress so well with
perfectly acceptable performance at high resolution, you might consider
saving a jpeg of the final edit alongside the original in loss-less
tiff.
The loss in performance on a scanned jpeg is a lot less than some of the
major issues that you have ignored in your quest for the best. ;-)
