Museum Quality Drum Scanning and Printing
The Seybold report scanner test (Volume 30, Number one) looked at a bunch of different scanners, from an ICG drum scanner to an EverSmart Supreme. Part of the tests involved scanning a chromium-plated glass-platen variable frequency resolution target at various angles. The surprising thing to me is that that they found that interpolation leads to higher resolution.
Here's what they say:
"We tested with and without interpolation, which was only necessary in the pixel direction. The reason is that a scanner steps at the same rate for both orientations."
Here's a shortened version of the chart of the results, in line-pairs per millimeter:
Scanner_____Horizontal________Horizontal interp.______Vertical______Vertical Interp.
Agfa t5000 +____45_________________45______________50_____________85
The drum scanners weren't tested for this, as the test slide is rigid.
So apparently what's happening is that choosing an interpolated resolution in the pixel direction causes the scanner to take smaller steps, but what confuses me is that for some of the scanners interpolation causes an increase in both horizontal and vertical resolution. Any ideas?
Choosing an interpolated resolution higher than the stated optiical resolution causes the scanner to take smaller steps in one direction, and that figure is then mixed with the optical resolution in the other direction to determine interpolated resolution.l
However, it is real optical resolution in one direction and this may give greater overall "real" resolution, depending on the subject.
So apparently what's happening is that choosing an interpolated resolution in the pixel direction causes the scanner to take smaller steps, but what confuses me is that for some of the scanners interpolation causes an increase in both horizontal and vertical resolution. Any ideas?[/QUOTE]
Ok, so in theory a Cezanne can resolve a maximum of around 120 lp/mm, which would be equivalent to about 6240 spi, right?
In theory that could be correct, assuming the lens of the Cezanne can send that information to the CCD.
On the Scan H-End forum some months ago there was some discussion of this very point. I quote from one of the messages there, by a fellow named Mr. Bill, who used to work for Scitex and knows more about scanners than anyone I know.
"I think Jake may have hit the nail on the head. The EverSmart Pro (and
Pro II) are 3175x8200ppi machines. This means that even an the maximum
scanning resolution, all the pixels in the long-bed direction are
captured optically and the interpolation is going on in the Y direction
and is only A 2.58X interpolation.
Interpolation gets a bad name from all those consumer flatbed scanners
that allow you to interpolate to 9,600ppi with a 1,200ppi optical engine
an 8X interpolation. Scitex only allowed much more conservative
interpolation, quite usable. You should never have any fear of pushing
any EverSmart scanner beyond it's minimum optical resolution.
Same would apply to the Cezanne and EverSmart Supreme which use the same Kodak CCD, but take smaller steps and scan in smaller rows so are capable of greater optical and interpolated resolution than the EverSmart Pro.
For what it's worth, I've had a number of 4x5 chromes professionally scanned on a Creo Eversmart Supreme, and since getting my Howtek 4500 I've re-scanned them. And the color accuracy and shadow detail are noticibly better on the Howtek thus far. I realize that there may be many non-hardware variables for color, but I just wanted to reinforce that spi resolution is certainly not the only issue and probably not the most important.
This thread has sure gotten off topic. It was, "large print digital back VS 4x5 color film."
Maybe time for the moderators to kill it and allow folks to start another more focused thread.
I would like to capture a moon rise in the background with the lighthouse:
in the foreground. Many problems include the park closes early so only winter months are possible, there are few days the moon is in right position per month, lack of coastal fog etc. are making it a challenge. A couple weekends ago I was totally asleep at the switch and missed a golden oppurtunity with everything in place, except me.
Good luck with the pursuit!
I took your question, what am I missing? as sarcasm and responded accordingly. Perhaps I misunderstood your intention.
Yes, the figure of about 2000-2200 effective pixels, or a bit more than 40 lpm, is what my tests have shown for the V750. And while that is close to the maximum detail in a 4X5 negative I think you would agree that resolution is only one of the components of image clarity, micro and macro contrast being the others, and in the case of a scan, shadow depth as well. So while the resolution of the V750 comes close to pulling all of the detail out of the typical 4X5 piece of film I certainly agree with you that image clarity will not be nearly so good as wiht a drum scanner or high end flatbed.
Another point. Earlier I stated that there was no need to scan beyond the information level of the film, which is determined by the camera/lens/film system primarily, and of course also by subject and conditions. I should amend that to say there is no reason to print at a resolution beyond the system because there may be several reasons to over-sample and then down size. In fact, I regulary scan at twice the optical resolution of the scanner and then downsize to reduce noise.
So for the record I don't think there is anything extravagant at scanning 4X5 film at 4000 spi, even though the film may not contain that much information. The problem, and this takes us back to the original purpose of this thread, is that a large percentage of the 320 mp count that you get is useless pixels. In my expereince the best 4X5 film shot in optimum conditions will not have more than about 125-50 mp of useful infirmation, regardless of the resolution in spi at which it is scanned. That is still a lot more information than in a P45 back so if there is a shoot out between the P45 and a 4X5 scan with your Premier my money will be on you. In fact, I think even a Howtek 4500 or my EverSmart Pro easly wins in this comparison with P45 versus 4X5 film.
One way that people can determine if the scanner is pulling all of the information out of the film is to scan at various resolutions and then look at the film through a microscope at about 20X. In most cases I think you will find that the limit of a piece of 4X5 film is around 50 lpm, or 2500 spi assuming that is real resolution.
I suspect you are incorrect in the max of 150 mp of info, however I have no data to support such a claim. It doesn't look like that to me on my scans - but that's not particularly scientific. I think what would be useful - perhaps for the entire group here - is to develop some testing methods to figure this stuff out. I think there remain some questions as to what is possible with scanners at the low, mid and high end. I think incorrect info abounds.... much of it no doubt put out by people marketing digital cameras, or the Reichman's of the world.
If a 20x microscope is the way to figure this out, by all means let's find one and put it to the test. I think we ought to develop a list of questions and hunt them down.... Of course, I have to get back to work now rather than start on this, but I think it would be great...
Museum Quality Drum Scanning and Printing