I'd like to hear some people's input on this one. Last year, I decided to get a drum scan of a 35mm slide that I was selling a print of to a customer. I decided to try a true drum scan at a new photo finishing place I found that caters to some famous photographers. When I got the CD of the scan and brought it home to post process, I was heavily dismayed by the grain and quality of the color. I brought the file into the person who scanned in the photo and he said that what I was seeing was typical of all film scans (Velvia 50 in my case). I understand this person has been professionally scanning for years. The owner talked with us and said I should consider an evaluation from a print rather than looking at a file on the monitor. We made a print from their $100k Heidelberg PMT drum scan versus my Nikon 5ED slide scanner. I liked the print more from my scan, much to the dismay of the owner. They are very professional and didn't charge me for the print and reimbursed me for the scan.
I was totally shocked by the whole experience. All I've heard about was how good drum scans were. Based upon a little more info I found out, it appears the PMT's may be so sensitive that they exaggerate the grain. I also thought that their dynamic range was supposed to be superior, but I didn't see it. If you add all of the color dots polluting the image, the dynamic range, color accuracy and color gamut go out the window in my mind.
Based upon some other posts I've read, I called West Coast Imaging and talked to their guy who does the scanning and told him my situation. He stated that I could expect to see the same grain artifacts I had seen in the other firm's scan. As a result, I didn't even bother having them do a test scan. Keep in mind, I went to a VERY reputable place to have my images scanned, and the person scanning it does this all of the time - so it shouldn't be from lack of experience or proper equipment.
For reference, I have included the photograph in question. The blue box represents the area that is the area of comparison of the drum scanned and home scanned images. The comparison images are posted below at actual pixel size.
The upper left hand corner is the original drum scan. The upper right hand corner is the original Nikon scan. The bottom left hand corner is the drum scan after post processing with Neat Image and other adjustments (it was a hard balance to process this image with Neat Image as color gamut and sharpness are degraded because the original file was so bad). The bottom right hand corner is the Nikon processed with Neat Image (not much required) and other adjustments. Equal amounts of sharpening were applied to both scans. As you can see from the original files, the Nikon has much better detail, color accuracy, color gamut, dynamic range and almost no grain.
Please let me be clear that I have not done ANYTHING to bias the results of these tests. I had intended to get all future images drum scanned for printing. Instead, I ended up learning to use my own scanner better.
Sorry for the long post, but does anyone have some experience or insight to why I should reconsider drum scans? I'd like to make sure I'm getting the best results for my prints. Thanks very much!
I had a similar reaction recently when I had 18 favorite images drum scanned (my first personal experience with such scans). I too was upset by the grain, which the Nikon scanners reduce with Digital GEM software. First thing I did was buy Digital GEM as a Photoshop plug-in to use in processing the drum scans. It is actually an advantage to apply Digital GEM after the fact, as you can leave the grain (and hence the detail) in the main subject areas (as in the case of wildlife) but remove it from areas like sky where it is more visible.
In your example, however, the drum scan, in addition to being grainy, has less detail than the Nikon scan. That was not my experience. The Nikon scanners do a remarkably good job if you adjust exposure accurately, but the drum scans I had made might be a few percent better in terms of resolution and color depth. There's also an apparent magenta bias in your drum scan. That could be the result of a different color space than Adobe RGB. In one or two cases, I actually preferred the color in my own scans, and couldn't reproduce the same color starting with the drum scans. But even there, the drum scan color was not objectionable, as it is in your example. So in short, whatever the reputation and experience of the vendor, I don't think you got good scans.
I used to run an Isomet drum scanner. Here is my take on what I see. At first glance it appears that the Nikon scan looks sharper and with less pronounced grain. Well, the drum scan has actually more detail but less contrast than the Nikon scan. The grain si sharper due to the higher resolution of the drum scan. The drum scan looks less contrasty, which affects the perception of image sharpness, because the drum scan has a greater bit depth and dynamic range. The color does look off to me, too, though. One would have to look at the original with a very high power loupe to be certain the color is truly different than the original but my guess is that the color is off. The most likely reason, in my experience, is that the scan was made with either the wrong "look-up" table for the film emulsion or the "look-up" table has been corrupted. I would try a different company using a different drum scanner to compare the results.
Thanks Cliff and Tom. I think it does come down to what actually edits and prints well versus what theoretically is 'better'. While drum scans may be better in some aspects, I think the quality transparency scanners actually produce visually better results. I think there's also a benefit to scanning oneself as there is complete control of the process. Let's not even talk about money - a 560MB scan of a 6x7 tranny at 16 bit runs about $200. 10 scans pays for the Nikon 9000 ED and glass carrier. In addition, everyone was getting paid for my prints except for me - not really the point of selling as my ego doesn't need to let the images go for next to nothing.
Doyle, I'm not sure what to say as your response is ambiguous (I am aware there are differences in the method digital capture, # of bits, color spaces, etc). Do you care to elaborate?
Just wanted to mention that I get my 6x7 chromes drum-scanned a Calpyso Inc (Santa Cruz, CA) and their quality has been really good so far. No complaints at all. They charge much lower than what you mention above. A 600 MB 16-bit scan (or a 300 MB 8-bit scan) for instance is $79. I typically wait for discounts, which are offered twice a year on average and can manage to get the same scans done for $50 each. For my low volume this works out ok.
BTW, I haven't seen the output from a Nikon 9000 ED and so can't comment on it.
Steve - I can't say what went wrong by looking at your samples, but this is simply a bad drum scan. Don't base your opinion from this one bad example! If you're paying $200 for a 560 MB scan from 6x7, I'll add that you're paying WAY too much :) If you care to try again at some point, I'll be glad to show you what I can do.
My initial response was actually similar to Tom's. When looking at the 4 examples without first reading which is which, I thought the bottom left one showed the most detail. The 2 on the right obviously have higher contrast and less grain making them appear sharper. This is the advantage of such software tools as Digital GEM.
That said I also agree that the deciding factor should be the appearance of prints. Higher optical resolution gives you a great starting point, but ultimately the overall appearance of sharpness and detail will be the deciding factor.
What you're seeing here is the Nikon scanner's built-in software doing much of the work for you, though you could achieve the same results from the drum scan - it has all the same detail (perhaps a hair more) but lacks the cleaning/optimizing.
Having worked with Nikon scans in 35mm and Medium Format (the 8000 and 9000 models) my personal impression is that results are generally as good as drum scans. In larger formats the available options are more limited and drum scans are hard to beat by desktop machines (other than the Imacon scanners which are in a very different price category).
I am not someone familiar with operating drum scanners, or the Nikon 8000, or using Velvia though I've had hundreds of scans from various machines over the last dozen years. And from my 35mm, MF, or LF have never received a drum scan back that looks like the one you posted. What I see here in the dark areas are a bunch of various hued pixels. So I would suspect that is noise not grain as noise favors no specific hues. I see grain in smooth areas like skies when scanning film above about 3000 pixels per inch. And in looks more like subtle variations of the same hue.
The advantages of well skilled drum scanning are not merely resolution but rather the clean capture of hue, shades, tones, levels of saturation, and color balance. ...David
I have compared 35mm scans done on both a drum scanner (locally at Denver Digital Imaging) and my Nikon Coolscan V. My impression was that I got better detail out of the drum scan, but just barely. For most prints I am perfectly happy with the Nikon, but for large prints, I go with a drum scan, especially if I have a client who is paying for the print.
The other thing I like about the drum scan is the image area. I was surprised at how much of the image was covered up by the slide mount.
In short, I've been really happy with drum scans, but the difference in quality is small enough that I only use drum scans when really necessary.
I've had some Imacon scans done and have been quite pleased with them. Everything I've read indicates drum scanning is quite an art form and very operator dependent, so just having one sample drum scan may not be indicative of what is possible.
Thanks for the additional comments everyone. I think one of the biggest benefits of home scanning is learning much more about scanning and post processing. I think when I get the right print, I'll try a drum scan again and be able to intelligently review the files produced. Danny - I'll keep you in mind when I think about getting another drum scan done.
Guy, the bottom left was processed fairly heavily with Neat Image which I know has a negative impact on sharpness. That does go to show how grain/noise negatively impact apparent sharpness. Considering what you and Tom said, I can see some sharpness there at the upper left that is not present in the Nikon scans, but not necessary good for they way it was processed.
David, noise was my first guess when I got the file, but they stated it was grain and not digital noise. In truth, I had another scan done that same time with another photo with the same results, but I'd guess they scanned them back to back. Imacon scans (which they had stated is not a true drum scan) were better in my opinion when I used to go to the Slide Printer here in Denver.
Brent, I do know what you mean about the loss of image area in 35mm, which was another factor to going with drum scans. Most of my serious photography is now MF, so I do get the full frame again which is nice.
True drum scanners produce extremely little to zero digital noise. Desktop scanners use ccds whereas true drum scanners (like the Isomet and Heidelberg PMT) use a photomultiplier tube. The photomultiplier tubes can create scans with higher resolution, higher bit depth, and a longer dynamic range than can ccds. What looks like noise in a proper drum scan is, indeed, the film grain.
The Imacon scanner is not really a true drum scanner. There is no clear Lucite drum over which the neg/tranny is stretched immersed in scanning oil. The Imacon calls itself a drum scanner because it uses a curved film holder instead of the Lucite drum. The curved film holder does have the advantage of not needing scanning oil and does create a more consistent sensor to film distance that cannot be matched by a typical desktop scanner. Also, the Imacon does not use a photomultiplier tube. It uses a very high density, high resolution ccd and high signal to noise sensor instead. The Imacon will produce digital noise but not to the degree of a typical desktop scanner.
The other thing I like about the drum scan is the image area. I was surprised at how much of the image was covered up by the slide mount.
No need for less than full-frame 35mm scans if you use Wess full-frame slide mounts for your Nikon scans. They are impractical for storage, being too thick for easy use with slide pages, but excellent for scanning (doubt they would work with an automated feeder, however).
Discussion
15 repliesI was totally shocked by the whole experience. All I've heard about was how good drum scans were. Based upon a little more info I found out, it appears the PMT's may be so sensitive that they exaggerate the grain. I also thought that their dynamic range was supposed to be superior, but I didn't see it. If you add all of the color dots polluting the image, the dynamic range, color accuracy and color gamut go out the window in my mind.
Based upon some other posts I've read, I called West Coast Imaging and talked to their guy who does the scanning and told him my situation. He stated that I could expect to see the same grain artifacts I had seen in the other firm's scan. As a result, I didn't even bother having them do a test scan. Keep in mind, I went to a VERY reputable place to have my images scanned, and the person scanning it does this all of the time - so it shouldn't be from lack of experience or proper equipment.
For reference, I have included the photograph in question. The blue box represents the area that is the area of comparison of the drum scanned and home scanned images. The comparison images are posted below at actual pixel size.
The upper left hand corner is the original drum scan. The upper right hand corner is the original Nikon scan. The bottom left hand corner is the drum scan after post processing with Neat Image and other adjustments (it was a hard balance to process this image with Neat Image as color gamut and sharpness are degraded because the original file was so bad). The bottom right hand corner is the Nikon processed with Neat Image (not much required) and other adjustments. Equal amounts of sharpening were applied to both scans. As you can see from the original files, the Nikon has much better detail, color accuracy, color gamut, dynamic range and almost no grain.
Please let me be clear that I have not done ANYTHING to bias the results of these tests. I had intended to get all future images drum scanned for printing. Instead, I ended up learning to use my own scanner better.
Sorry for the long post, but does anyone have some experience or insight to why I should reconsider drum scans? I'd like to make sure I'm getting the best results for my prints. Thanks very much!
In your example, however, the drum scan, in addition to being grainy, has less detail than the Nikon scan. That was not my experience. The Nikon scanners do a remarkably good job if you adjust exposure accurately, but the drum scans I had made might be a few percent better in terms of resolution and color depth. There's also an apparent magenta bias in your drum scan. That could be the result of a different color space than Adobe RGB. In one or two cases, I actually preferred the color in my own scans, and couldn't reproduce the same color starting with the drum scans. But even there, the drum scan color was not objectionable, as it is in your example. So in short, whatever the reputation and experience of the vendor, I don't think you got good scans.
Doyle
Doyle, I'm not sure what to say as your response is ambiguous (I am aware there are differences in the method digital capture, # of bits, color spaces, etc). Do you care to elaborate?
Just wanted to mention that I get my 6x7 chromes drum-scanned a Calpyso Inc (Santa Cruz, CA) and their quality has been really good so far. No complaints at all. They charge much lower than what you mention above. A 600 MB 16-bit scan (or a 300 MB 8-bit scan) for instance is $79. I typically wait for discounts, which are offered twice a year on average and can manage to get the same scans done for $50 each. For my low volume this works out ok.
BTW, I haven't seen the output from a Nikon 9000 ED and so can't comment on it.
Regards.
-Anil
Thanks,
Danny
Danny's the man! He did some scans of 645's for me. The price was right, and the quality is superb!
Just do it, Steve! You'll be glad you did!
-PB
That said I also agree that the deciding factor should be the appearance of prints. Higher optical resolution gives you a great starting point, but ultimately the overall appearance of sharpness and detail will be the deciding factor.
What you're seeing here is the Nikon scanner's built-in software doing much of the work for you, though you could achieve the same results from the drum scan - it has all the same detail (perhaps a hair more) but lacks the cleaning/optimizing.
Having worked with Nikon scans in 35mm and Medium Format (the 8000 and 9000 models) my personal impression is that results are generally as good as drum scans. In larger formats the available options are more limited and drum scans are hard to beat by desktop machines (other than the Imacon scanners which are in a very different price category).
Guy
The advantages of well skilled drum scanning are not merely resolution but rather the clean capture of hue, shades, tones, levels of saturation, and color balance. ...David
The other thing I like about the drum scan is the image area. I was surprised at how much of the image was covered up by the slide mount.
In short, I've been really happy with drum scans, but the difference in quality is small enough that I only use drum scans when really necessary.
Guy, the bottom left was processed fairly heavily with Neat Image which I know has a negative impact on sharpness. That does go to show how grain/noise negatively impact apparent sharpness. Considering what you and Tom said, I can see some sharpness there at the upper left that is not present in the Nikon scans, but not necessary good for they way it was processed.
David, noise was my first guess when I got the file, but they stated it was grain and not digital noise. In truth, I had another scan done that same time with another photo with the same results, but I'd guess they scanned them back to back. Imacon scans (which they had stated is not a true drum scan) were better in my opinion when I used to go to the Slide Printer here in Denver.
Brent, I do know what you mean about the loss of image area in 35mm, which was another factor to going with drum scans. Most of my serious photography is now MF, so I do get the full frame again which is nice.
The Imacon scanner is not really a true drum scanner. There is no clear Lucite drum over which the neg/tranny is stretched immersed in scanning oil. The Imacon calls itself a drum scanner because it uses a curved film holder instead of the Lucite drum. The curved film holder does have the advantage of not needing scanning oil and does create a more consistent sensor to film distance that cannot be matched by a typical desktop scanner. Also, the Imacon does not use a photomultiplier tube. It uses a very high density, high resolution ccd and high signal to noise sensor instead. The Imacon will produce digital noise but not to the degree of a typical desktop scanner.
No need for less than full-frame 35mm scans if you use Wess full-frame slide mounts for your Nikon scans. They are impractical for storage, being too thick for easy use with slide pages, but excellent for scanning (doubt they would work with an automated feeder, however).