I was browsing through eBay and looking at scanners I can't afford and came across an Imacon, which led me to the Imacon website. They have a 8000 dpi scanner but it only scans 4X5 at ~2000 dpi. If it can scan a 6X9 negative/transparency at 8000 dpi, why not a 4X5? This makes no sense to me.
It is more expensive to make a scanner which can scan the full 4 inch width at the higher resolution. Imacon must have decided that 2000 ppi was adequate for the larger format. They presumably use a different arrangement of sensors for the larger and smaller formats.
According to digital sampling theory, the maximum resolution in line pairs per unit length obtainable from a scanner is half the scanning resolution. That would be 1000 line pairs per inch, which would be the same as 1000/24.5 or about 40 lp/mm. No scanner will actually achieve the theoretical maximum, but a high quality scanner like an Imacon will come close. 40 lp/mm would allow up to 8 X enlargement. Such a print could be viewed relatively close-up without any obvious degradation of quality. From normal viewing distances for large prints, one could even make larger enlargements.
I believe that Imacon does make a 4 x 5 scanner with higher scanning resolution than 2000 ppi, but going as high as 8000 ppi would seem unwarranted. Also, a file produced at that resolution would be enormous. If the source were scanned only at 24 bit color depth, the resulting file would be about 3.4 Gb. Compression could reduce that some, but processing such a file in a photoeditor and saving the resulting scans would require much more computing resources than would generally be available.
I don't know how old this scanner is, but another thing to keep in mind is that there have been significant improvements in scanning technology in the past several years.
Correction: You divide by 25.4, not 24.5. That was a typo. I did use 25.4 in the actual computation, and the resulting numbers are correct.
8000 DPI refers to the sample rate for 35mm only. The CCD is the same for all scanning, just the optics move (like a zoom lens) depending on what format one is scanning. For each larger format the resolution is reduced.
You must also beware of marketingspeak. Many times, the resolution that vendors most prominently mention is *interpolated* resolution. This is an apparent resolution created from the actual scanned pixels - the *optical* resolution. This isn't to say that interpolated resolution is a farse; all digital cameras use must use interpolation to "fill in" parts of an image because of the nature of the semiconductors that register the light. But optical resolution is the best direct sampling the device will give.
Also, don't get fooled by resolution numbers. A scanner's quality will also depend on the maximum density it can record (the lower the number, the thinner a negative you must use), the color depth, if the film lies on glass that can distort to some degree the image quality, and the maximum file size it can produce. There are also issues of quality control, customer service if necessary, ease of use, and even how much space it will take on a desk, all of which could influence whether to pick one scanner over another.
Correction: 6X7 is 3200 dpi and maximum file size is 1.2 Gb. I have a feeling that max resolution is limited by an arbitrary max file size.
I think you're on the right track, Edward. Scanner resolution would be controlled by the optics, the size and type of sensor, and the internal RAM of the scanner used to assemble the image from the sensor. Many multi-format scanner specs will include the maximum scanned image size expressed in pixels. Working back from that, the maximum resolution in DPI (or, more properly, PPI) for each film size results.
<HTML>
There is another important aspect of digital image quality assessment : the noise problem. And this noise factor probably limits the interest of scanning any film at a 8000 dpi "resolution".
Even the best scanner fitted with the best optics has to live with the fact that grain noise when digitizing a silver-halide image increases when the analyzing slit decreases in size. This was well-known to micro-densitometer aficionados in the good old days of analog silver-halide image measurements, under the name of Selwyn Law. The definition of RMS granularity is directly derived from Selwyn Law, when applicable i.e. : B&W films for sure, color slides certainly, but color negative films with great caution.
I do not see any reason why a film scanner could escape Selwyn Law. After all, it is just the modern, computer-controlled and analog-to-digital converting version of the good ol' analog micro-densitometer (my favourite was the Joyce-Loebl model with its superb optical reference density wedge : this piece of glass would certainly cost the same price as half a dozen of office flatbed scanners ;-);-).
If we assume that Selwyn law applies, when you want to increase resolution in a digital scan, you increase noise as well. Eventually in order to get a decent image you'll have to average several adjacent pixels to average noise, hence losing resolution.
So as a conclusion, prefer to digitze a large format film with a low resolution scanner followed by a modest enlargement factor that digitizing a 35 mm slide at 8000 dpi ;-);-)
BTW and going off-topic : My feeling is that silicon is less noisy than film. If the final destination of your image is to be digitized, I'm convinced that a much smaller silicon surface with the same total number of pixels in the final file will give you a similar image quality than a digitized film of twice the surface. e.g. a 18x24mm silicon detector will probably give you the same image quality or even better than a scanned 24x36mm slide. For Selwyn reasons. But this is irrelevant here where everybody uses 4"x5" and above as preferred format ;-);-)
</HTML>
Posting Permissions
Reply With Quote
Bookmarks