scanning large format

[joe a kras]

I post this here as I shoot primarily large format and do scans on an epson 3200. My question is when I have scans done on drum scanners for 8X10 or 4X5 they are usually done at 300 meg. When I do them on the epson I have a choice of 6400 or 3200 or 1600. The file size is enormous at 6400 and over 500 meg at 3200 on 4X5 and more manageable at 1600. I do not see a difference between 1600 and the others besides file size. However the drum scans are so much better. I would like to know if there is some type of primer on all this information that I can read and understand to explain all this. By the way I usually print out what I refer to as proof prints on the epson 2200 and if the image is worthy of a print then send it out for drum scanning and professional printing.

[Erik Sherman]

I don't know about a primer, but consider the difference between the 3200 and a drum scan. With the former, the 4x5 negative lies in a holder atop the glass, which adds increased chances for image degradation. The lens moves forward and back past the negative to do the scan, which causes some amount of vibration in the lens. Furthermore, the optical resolution is 3200 dpi. While software interpolation can increase apparent resolution, this is the bottom line limitation on image quality.

A drum scanner, on the other hand, mounts the negative in a curved path on a drum which is what moves, so it eliminates the vibration of moving the lens. The optical resolution is more than twice that of the 3200, and remember that while you might get a more manageable file size from the service bureau, that is probably brought down from a larger base amount of image information, so you're going to lose less than with the 3200.

One more difference is that a drum scanner uses a wet mount, not a dry one. That means there's fluid between the negative and the drum. According to at least one person (check http://www.luminous-landscape.com/reviews/scanners/drum_scans.shtml) who has used a drum scanner, you can actually see a difference in quality looking at the film itself after it's mounted.

The real difference, essentially, is not the output resolution, but the actual quality of the input resolution. That's where the drum scanner wins out.

[Ken Lee]

Drum scanners are much more expensive - they are not consumer products. They often require a skilled operator and regular calibration, etc. They don't use CCDs, but more expensive sensors, which can be analog rather than digital. They give a greater range of density, and better color fidelity. It's like comparing a portable stero you buy in a drug store, with a fancy one you get at a special hi-fi store.

That being said, not everyone who owns an Epson 3200, or other consumer scanners, knows how to get the best scan out of them.

Once the scan resolution is high enough to record the grain of the film (or the dye clouds), higher resolutions simply give you "more grain".

We all have to draw the $ line somewhere.

[Leonard Evens]

I use an Epson 3200 to scan 4 x 5 film. There is no point in scanning at 6400 ppi. The maxium optical scanning frequency horizontally is 3200 ppi. The step size allows for 6400 ppi vertically, but if you choose that, the horizontal scanning will have to interpolate. You have recognized that although the scanner can scan at 3200 ppi, it doesn't deliver that in actual photographic resolution of fine detail. According to digital sampling theory, the maximum resolution in line pairs per unit length you can obtain is half the scanning frequency. In this case, 3200 ppi is roughly 126 pixels per mm, so the most you can hope for is about 63 lp/mm. In actuality, various tests have shown that you obtain at best 30 lp/mm with the 3200. That should suffice for an enlargement up to 5 or 6 X or possibly larger if the viewers don't get too close to the print.

I do the following. I scan at 3200 ppi and then I rescale in a photoeditor to about 2000 ppi, which produces a file I can work with on my computer. The reason for rescaling in the photoeditor is that I suspect the method the scanning software uses for scanning at a lower scanning resolution may not be particularly sophisticated. The photoeditor allows for the use of fairly sophisticated rescaling techniques.

It is hard to judge just why you are getting better scans from a drum scanner without knowing the details of how those scans are done. Probably the drum scanner does a better job and comes closer to is maximum lp/mm, but that may not be the whole story. One thing to consider is that you may not be using the scanning software to obtain the best possible scan. For example, Roger Clark, who is quite experienced with drum scanners has obtained excellent results with an Epson 4870, quite comparable in relevant ways to drum scans. See www.clarkvision.com, which contains a lot of useful information about scanning and other digital issues. You will also find many links there to other sites discussing scanning. He doesn't appear to have posted any examples of his Epson scans, but at

http://groups.google.com/groups?q=4870+Clark+group:comp.periphs.scanners&hl=en&lr=&ie=UTF-8&group=comp.periphs.scanners&selm=40AAC34D.70108%40qwest.net&rnum=1

you will find a newsgroup article in which he says he has obtained results comparable to drum scans. The Epson 3200 is can't deliver quite what the 4870 can, but it still does a pretty good job.

[tim atherton]

"I use an Epson 3200 to scan 4 x 5 film. There is no point in scanning at 6400 ppi. The maxium optical scanning frequency horizontally is 3200 ppi."

and because Epson kinda cheats about how they define it, the true resolution is actually much closer to 1600ppi

[Guy Tal]

Joe,
This may be what you're after: http://www.scantips.com/.

Guy
Scenic Wild Photography

[Julio Fernandez]

Ken raises a valid and very important point as to the differences in the mounting methods, Drum scanners use wet-mounting and I also agree that you can indeed see the quality difference by viewing a wet mounted slide. Wet mounted slides look a great deal more brilliant and seem to have a greater dynamic range than the same slide viewed dry. The optics are evidently quite different and favour the wet mount. Additionally, wet mounting (optically) dissolves grain and diminishes if not eliminates surface defects such as light scratches etc. This advantage is inherent to drum scanners which require wet mounting, however it can also be had with flatbed scanners. Wet mounting with flatbeds elevates their performance to the same degree that wet mounting enhances drum scanners. There are wet mounting turn key kits available for the most popular scanners and formats, if you like to learn more, drop me a line.

[Leonard Evens]

Tim,

Epson does exactly what every other scanner manufacturer does. They report the scanning frequency. That is the number of samples collected per inch in each direction. And they report it truthfully. As I noted previously, the maximum photographic resolution in line pairs per unit length is half the scanning frequency. No scanner achieves the maximum, but some come closer than others. How close depends on a variety of factors. I noted that the Epson 3200 achieves about half of that maximum or about 30 lp/mm. I agree that scanner manufacturers should tell you how close they do come, but the standard in the industry seems to be not to do that. So the question is how the Epson scanners compare to other scanners with the same scanning resolution. My guess is that they are about comparable to others in the same price range.

[Ken Lee]

Leonard - What you say makes excellent sense (as always). I finally get it. Thanks !

[Donal Taylor]

Going from memory, the tests we did of the Epson 3200 came out around 1.5-1.7 sensors per line which, if I remember correctly works out at between 1800 and just under 2000ppi

By comparison, I think those we did with the Microtek 1800f came out at somewhere between 1600 and 1800ppi (though the scanners are at different price points and resolution is obviously only one factor - and the Microtek isn't scanning through the glass flatbed. I think if we compared the new Microtek that directly competes with the Epson 3200 etc we would probably find the same issues - although the build of the Microteks is usually better).

One of the things with Epson is their use of a staggered sensor array which, certainly in the past other manufacturers used, but initially described in more "realistic terms" as say 800ppi, but when Epson came out with it for their original budget Epson Perfection 1640 scanner, they latched onto the staggered array as a beat the competition marketing tool and described it as 1600ppi - which in some was it is, in some ways it isn't, and I think is what was probably being referred to as Epson sort of cheating. But from that point on, everyone who was using or started using a staggered array also then started basically doubling their figures - it's not exactly deceitful, but neither is it accurate. This is still how they approach it as far as I can tell. And while the a figure of 3200 may accurately reflect the scanning frequency with a staggered array, it doesn't accurately reflect the actual resolution. So we could charitably say it's being "economical with the truth" as Bernard Ingham told the House of Commons.

When you get up to the slightly higher quality "pro-graphics" level scanners like the Microtek 1800f of the Epson Expression 1680, as I remember they don't tend to use staggered arrays and so the stated resolution is much closer to the actual resolution (and if there are any that do, I don't think they don't use it as a marketing tool - they aren't convincing the "amateur" market they are getting a quart in a pint pot).