Best Scanner for Medium/Large Format Film?

[noah]

<p>Stuart--Of course I don't mind. However to be fair you should also tweak the Howtek scan...both my my examples are straight out of the scanning software with no sharpening or tonal adjustments. Both can be improved upon, and for my printed tests I optimized both scans for the way I print.</p>

<p>In any event there's still quite a difference between your tweaked Epson scan and the untweaked Howtek scan.</p>

<p>M-dawg--I don't do scientific lens tests. I just make prints and decide if the quality is good enough or not. I'm sure on a given film area the Mamiya 7 lenses are sharper than the Sironar S lenses. But in practice, for a similarly sized output, the 4x5 wins every time. There's no substitute for film area.</p>

<p>Having said that, the Mamiya 7 lenses are superb and it's probably the closest you can get to large format quality in a 120 system. But again, even a cheap old 4x5 lens can beat a great 6x7 lens due to the large film area (assuming both formats are scanned on the same scanner, of course!).</p>

<p> </p>

[david carver]

<p>"Having said that, the Mamiya 7 lenses are superb and it's probably the closest you can get to large format quality in a 120 system. But again, even a cheap old 4x5 lens can beat a great 6x7 lens due to the large film area (assuming both formats are scanned on the same scanner, of course!)"</p>

<p>I agree Noah but the difference in ease of use between the Mamiya and the 4x5 is like night and day. I tried 4x5 and it was very painful for me. It was not fun. My Mamiya 7 is almost as easy as my 5D II. For B&W the "7" is so much better than the 5D, but that is off the subject.<br>

BTW, I use a V700 and am very happy with the results. The largest prints I make are the max on the Epson R2880. I figure that I can have a drum scan made if I need larger.</p>

 

[f_ph1]

<p>Re: Noah's comparison...</p>

<p>The scan from the Howtek looks crisper, no doubt. But I cannot possibly be the only one that sees that, although there is a difference,<br>

it is not very significant, at least in these examples. That seems to be an awful lot of money to spend for such a small increment in scan quality.<br>

Given a Howtek, I would sell it, buy an Epson, and maybe a Linhof or some nonsense... and still have quite a bit of dough left for an extended trip to<br>

Montreal.</p>

<p>I started life as a visual artist (painting, drawing, etc.). Perhaps this accounts for my lack of impressionability for such differences in what amounts to<br>

one dimension of visual expression.</p>

<div></div>

[noah]

<p>David--I find 4x5 can sometimes be easier to shoot simply because I have more control over the photograph. I can keep vertical lines parallel and use movements to improve my composition. However, they're really different tools for different purposes and in fact I find that they complement each other quite well. The Mamiya 7 is one of my favorite camera systems. It probably offers just about the highest image quality you can get from a handheld portable film camera. As you say, it's just as mobile and fast to shoot with as a DSLR and in some cases it's not any bigger.</p>

<p>The Phase One P45+/P65+ backs may offer higher quality, but at a high initial cost. </p>

<p>F Ph--We can agree to disagree on this one. In print (and on my monitor) it's a pretty large difference. The Epson scan looks like it has been interpolated, it's a bit jaggy. But it's straight out of the scanner. I don't know if it's due to differences in x/y resolution or if the machine is using hardware interpolation. But it's very visible, even in the tweaked version posted by stuart. </p>

<p>The drum scans are just more robust in terms of processing. They can be sharpened, up-resed, whatever you need to do. They behave more like digital camera files in a way. I do most of my scans at 4000dpi. That means for my 40x50 prints, the scans from 6x7 actually need to be upresed a bit. They hold up very well. The epson scans are already jaggy so I wouldn't think of scaling them up.</p>

<p>Keep in mind though that my normal print sizes are 32x40in. and 40x50in. from 6x7cm and 4x5in. negs. Especially the prints from 6x7 represent a big enlargement factor. Not everyone makes prints that large. And everyone has a different opinion of what is good enough. The OP wanted to know what gear would give him better quality than an Epson...and as I said in my previous post, for larger formats the only real option is a drum scanner. That doesn't discount the quality of an Epson in any way. </p>

<p>I moved to the drum scanner because I thought there would be an incremental increase in quality as well as other advantages (better software, worflow, etc.). I was surprised at how much of an increase in detail there was, especially in the large prints from the MF negs. </p>

<p>I got my Howtek system, complete, for around $4k. Sure, it's more than a $700 Epson, but not nearly as much as you imply. New Linhofs are going for around $7k. And, while the quality of a Linhof is first-rate and I'd absolutely love to have a Master Technika 3000, the photos wouldn't be any better than my $500 wista VX. If you are going for the highest quality possible from film in an analog-to-digital workflow, I'd argue that the scanner is one of your most important investments. The beauty of film, however, is that you can shoot it and scan on something like an Epson, which as I said is more than good enough for MOST photographic needs. Then if you occasionally need more quality you can outsource a scan or two...</p>

<p>Personally, I'll probably average 200-300 high-res agency and archive scans a year. In my city a good drum scan from 6x7 costs $50 and from 4x5 it's $100. So it won't take long for the scanner to save me money.</p>

[StuartMoxham]

<p>Yep Epson scans can start to fall apart if pushed around too much I've done that enough times with my V500 scans. I only tweeked the Epson scan is looked much worse than the Howtek to which it was being compared and I figured it could be tweaked a bit closer.</p>

[f_ph1]

<p>The closer to "right" an image is obtained in the camera, the less the need for tweaking. I suppose we will have to agree to disagree on the rest.<br>

The difference, although clear, is still not what I would call substantial. </p>

[StuartMoxham]

<p>Epson scans are soft compared to most film scanners. Thats not to say they can't do a great job but the scans need work to get the best from them. Personaly I am very happy with my Epson scanner but I would like a Nikon Coolscan 9000 problem is it's just too expensive for me. Heres three scans from a V500.</p><div></div>

[f_ph1]

<p>That is my point...</p>

<p>The Nikon is better, as I recall. But not enough to justify another 2000+ dollars. For that price, I want an increase in quality like, say the increase one obtains when comparing 35mm to medium format.</p>

[mwmcbroom]

<p>I just finished reading through this thread with some interest. You might describe me as a person who has a rather large archive of both 35mm and medium format slides and negatives who wants to digitize them as a means of preserving them from further decay. Many of my earliest shots are already showing color shifting, which for the most part can be corrected with the image processing software I use, so I'm thankful for that, at least. But I'm also a person with somewhat shallow pockets, which has forced me to scrounge around a bit for solutions.</p>

<p>I guess I'm in the minority amongst the people in this thread at least in that I think 35mm is a very viable film format. It seems to me that the biggest limitation to the resolution of a 35mm slide or neg is the scanner that is being used to digitize it. It would be nice to see scanner technology continue to improve to the point that drum scanner and/or Nikon 9000 resolutions become commonplace and affordable, but it appears that because of the reduction in use of film, this will likely never come to pass. A shame, since there are lots of folks like me who have large archives and who need to digitize them.</p>

<p>So what's a guy like me to do? Well, I started out using an Epson 3170 (3200 ppi max resolution). It did an okay job with 35mm and a good job with medium format, I've felt. But I wanted something better and hopefully something that would scan large format. But I didn't have a lot of money to spend, so both the Epson V700 and V750 were out. Then I came across the HP G4050. It scanned all formats, and best of all was pretty cheap. So I bought one, and found out quickly that, even though it claimed something like 6400 ppi resolution, it actually was worse than my Epson 3170. Much worse. I learned later that its actual optical resolution was something like 1000 ppi. Then at the recommendation of a friend who had been using an Epson 4990 for quite some time, and was very happy with its performance with large format, I bought one second hand. I found that the 4990's resolution was better than the 3170's, but only marginally so.</p>

<p>So I was frustrated. I mean, the 4990 does a perfectly acceptable job with medium format, I think, for enlargements up to poster size, but it was losing lots of detail with 35mm, which was obvious when I was looking at my slides thrugh a loupe. Then it occurred to me that I might get what I want if I tried using a slide duplicator with my DSLR for my slides. And that led to another string of events. First off, I bought one of the "digital" side duplicators that you see often advertised on the bay, and was using it with a zoom lens on my EOS DSLR. A big improvement, but I could still see that I wasn't getting all the detail that was there. The problem was two-fold: the quality of the zoom lens, and the fact that the duplicator used an inner element to correct for closer focusing. So I ended up cobbling up a duplicator of my own where there was nothing between the slide and the sensor other than a macro lens. I used the "digital" duplicator's tube to which I'd attached a slide/film stage off an old zoom-slide duplicator, which was threaded onto the front of a Micro Nikkor 55mm f/3.5. I then stacked a couple of extension tubes to the rear of the lens to get the right amount of magnification at the right focusing distance. And attached all this to my EOS with a Nikon-EOS lens adapter. At last I was getting dupes as good as I could get with the equipment I had available to me. But my DSLR is "only" a 10.1 megapixel camera, which translates into about 2700 ppi equivalent, and I still think there is more resolution to be gained from my 35mm images.</p>

<p>But instead of spending such a large amount on a Nikon 9000 or a used drum scanner or whatever, I'd much rather spend that amount on a better DSLR with a much higher pixel count, like a 5D Mk II. According to my back-of-the-envelope calculations, a 5D Mk II should render images at almost exactly 4000 ppi, same as a Nikon 9000. And price wise, they're about the same. Me, I'd much rather have a camera that can do what a dedicated scanner can do, because I still have the camera that I can use for things like . . . photography!</p>

<p>As for medium format, I've been giving that some thought, and I don't see why I can't rig up a similar setup for mf slides and negs. I'll no-doubt have to fabricate things, but I have a lathe and a mill and lots of hand tools, so I reckon I'll be able to come up with something that'll work.</p>

<p>One last point and then I'm done. I don't recall reading anyone commenting about the inflated numbers used by scanner makers about their claimed resolution. I have determined that my "4800 ppi" Epson 4990 is scanning at an actual true optical resolution of about 2000 ppi. On a good day. The Epson V750 has been shown to scan at about 2300 to 2400 ppi. 6400? Fuggedaboudit. Oh, and the plustek 7600 models? About 3600 or so. Pretty good when considering the price. But if one wants the max 3600 ppi resolution and all the other bells and whistles loaded up (e.g., dust reduction, etc.), you're looking at something like one HOUR per scan. Hrmm . . . So, one thing I'd really like to see come to pass is for the scanner makers to quit artificially inflating their resolution numbers. All it does by selecting large resolutions is bulk up the files, but with no further image detail. And it's for this reason that, when I scan medium format images with my Epson 4990, I select 2400 ppi. The image files are smaller but there's no loss in image information.</p>

[rob_piontek]

<p>I've also been experimenting with using my DSLR to digitize my negatives. One possible way to increase the resolution is to shoot multiple frames at greater than 1:1 for 35mm, and stitch (a program called hugin is free). I'm also wondering if this method could potentially get more shadow detail than a scanner, as if you backlight your images with flash you can put quite a bit of light through the negative. The big problem is the time. Even though stitching is pretty easy, it's still kind of a pain.</p>

[AlanKlein]

<p>How's the Epson V600 for 35mm and MF? I realize it doesn't scan LF but I thought you would have some idea how it compares to the others mentioned in this thread for these two formats.</p>

[timlayton]

<p>Thanks everyone for all of the information. I have learned a lot from everyone here on this topic both technical and just common sense points so I thought I would give you an update from my end.<br>

1.) I develop my own b/w so nothing changes here for me.<br>

2.) When I have my color negatives or positives developed at my local pro lab I have them develop only.<br>

3.) I view the negatives and positives on my light table and then select which ones I want to pursue for output. The rest go in the archive separated from the "keepers" For me, some of my output is for the web and the other is for various size prints. Depending on my output requirements I will take a different path. <br>

4.) For web output I feel my Epson V750 is "good enough" and plan to just stick with it. I just got a third party film holder from betterscanning.com and my scans look better than the scans I was getting with the original Epson holders so this is very helpful to me. I am using the AI Silverfast software with my Epson. <br>

The comment from Michael about the V750 peaking at 2400 dpi is something I would like to know more about. If anyone has some technical knowledge about this I would appreciate the help in understand the details. On my V750 at 2400 dpi I get a 16-bit TIFF file that has a pixel ratio of approx. 6600 x 5250 for my 6x7 film and the file is about 190 Mb. Assuming this is a desirable image I calculate that I could get a 22" x 17" print at 300 dpi. I just got back some 30" prints from this scenario that looked good to me and my client was thrilled. <br>

5.) For print output I put these into two categories: prints that I do myself and those I send out to a pro lab. For the prints I do myself I use a Canon Pro 9500 Mark II for matte and a Canon Pro 9000 Mark II for glossy. I can print up to 13x19 on these printers if needed. My film workflow is simple (develop -> scan in -> Lightroom for catalog and some basic edits -> Photoshop and plugins for detailed edits -> output file for printing). <br>

6.) For prints that I send out to my lab I could do anything from 5" x 7" to 30" x 45". I base my pixel calculations on a 300 dpi print scenario. When I run short of pixels for the really big prints I have used Genuine Fractals in a few select cases from onOne Software for my DSLR images as well as my scanned film images with very good success so far. I've only done this less than a dozen times over the last year. <br>

7.) My local pro lab charges $10 for 120 MF 4000 dpi scans. A Nikon 9000 new runs about $2200 so it would only take a couple hundred scans to offset the cost. I am still on the fence about this option for purchase because of the state of scanners and the lack of OSX Snow Leopard support for the Nikon 9000. <br>

Thanks again for everyone's help, input and comments. I look forward to any observations and suggestions that you might have.<br>

Thanks,<br>

Tim</p>

<p> </p>

[StuartMoxham]

<p>The optics in the epson scanners are not great and usually anything above 2400 ppi with a v700/v750 does not add any real resolution to the files you might on a good day with the wind in the right direction in March see a touch more detail scanning at the higher resolutions but its usually not worth the effort of dealing with the larger bloated files.</p>

<p>With 6x7 or 6x9 negs or slides you can still make quite large prints scanning with a v700/v750 as you have seen for yourself. The v750 could be all you need. The v700/750 is better than the v500 that tops out around 1800ppi although Epson make claims of 6400ppi. I make quite small scans with my v500 big enough for web uploads and small prints but with B&W films I'm drifting back to darkroom printing for my small B&W print and use the v500 to choose which negs to print.</p>

[rob_piontek]

<p>Here is a site with resolution test results for various consumer scanners:</p>

<p>http://www.filmscanner.info/en/FilmscannerTestberichte.html</p>

<p>Epson V700 at around 2300dpi, Nikon 9000 at almost 4000 dpi, and the Flextight even higher. </p>

[timlayton]

<p>Stuart and Rob, very good info, thank you. I am learning a lot here. <br>

I do have a question about pixels in regards to scanning from film vs. DSLR pixels. I've read more than once that the pixels from the scanned film is not equal to the DSLR pixels. Can someone provide some detail here to the differences? <br>

Here is what I think I know:<br>

I shoot 6x7 MF mostly. On a 6x7 film scan at 2400 dpi on the Epson V750 I get a file that is about 6600 pixels by 5250 and about 180Mb based on my scan parameters (multi-pass, etc). The exact number varies a little based on the capture size, but this is good enough for discussion purposes. <br>

I personally use the 300 dpi rule, meaning that I divide the pixel count by 300 and that tells me the maximum size of the print I can get at that resolution. Using these numbers I could get a print that is about 22" x 17". I could lower the dpi to 250 and effectively get a larger print (26" x 21") and in many cases that is more than "good enough". I personally never try to go below 250, but I have read that some printers will interpolate the 8-bit sRGG file to either make the print bigger or lower the dpi value to achieve a larger print. I want to avoid both of those scenarios if I can at all costs.<br>

If I multiply the 6600 x 5250 I get 34Mb approximately. Does this mean that by shooting film and scanning it it at 2400 dpi that I effectively have an image that is comparable to a 34 MP digital camera? I've been told that film pixels that are scanned are equal to about 1/3rd of DSLR pixels. I have no idea if that is true or the details behind this. Does anyone know?<br>

The reason I am trying to understand this and get to the bottom of it is so I can make some decisions about my workflow and simply just work within the limitations of my equipment (film or digital). For example, if the above scenario of scanning my 6x7 film is close to the quality I can get from a 34 MP digital camera then I can live with the slower workflow and hassle factor in some cases vs. the huge expense of buying a 30+ MP digital camera. If it turns out that the 34Mb effective pixels are equivalent to 1/3rd of that value at 11 or 12 MP, then that is okay as long as I know and understand the details. There are times where I need to leverage the capabilities of my DSLR system or maybe I need the fast-paced results of digital, then I will use my Nikon D3S.<br>

Thanks for your help and I am interested in finally understanding the details between scanned film pixels and DSLR pixels.</p>

<p>Tim</p>

<p> <br /></p>

[StuartMoxham]

<p>Well I would say it will be somewhere in between. Not as good as a 34mp digital back probably not as good as 20mp digital back but likely better than a 12mp DX DSLR.</p>

[m_dawg]

<p>From what I've seen, color 6x7 scanned on a Nikon 9000 is similar in terms of absolute print quality to full frame digital, but once you get up close the digital lacks detail (it also lacks grain, though, and is to that extent better for wall-sized enlargements).</p>

<p>You can download a sample image taken from a 5DII or D3x and print it to see if you think it's good enough. My guess is it will be.</p>

[philip_wilson]

<p>In my experience a Nikon 9000 scan of a 645 film (e.g. Velvia 50) produces equivalent quality to my 5DII (same resolution etc...). My Fuji GX680 will blow away my 5DII. However, how much of this is the sensor / film size and how much is the glass is not clear. I believe that the 5DII is at or above the resolution of canon's best lenses and that this has an impact on ultimate resolution and quality. Scanned film obviously looks different to a digital picture and does not look as "clean" when you pixel peep. That said most analysis suggests that a scanned 35mm film has about the resolution of a 6MP to 12 MP digital SLR. This would seem to be in line with my personal experience where the 5DII is about the same as a good 645 scan.I disagree with M Dawg and can provide some scan images to show why.<br>

This is the scanned GX 680 shot - very highly compressed as it is a 600 MB file</p><div></div>

[philip_wilson]

<p>Here is a crop from the GX680 scan</p><div></div>

[philip_wilson]

<p>Crop from 5DII taken at the same time from the same place ( I used it as a polariod) with the 16-35 F2.8 II L series lens. While this lens has some issues it is the best Canon wide angle zoom by far.</p><div></div>

[philip_wilson]

<p>Before anyone asks the 5DII was focused and tripod mounted using MLU and a remote release. The issue why this is so bad is that the lens was at 35mm and F2.8 - probably it's worst settings - but even at F8 the shot would not have been in the same league as the GX680. Remember these are extreme crops - the GX 680 scan is 11969x8819 pixels so we are seeing 0.3% of the total image in the crop. Similarly the 5DII shows 0.53% of the final image.<br>

For those who will claim the 5DII can do better - here is perhaps a best case result for 35mm (different scene). This one is 331x 119 pixels or 0.2% of the full 5DII image. Here we have MLU and tripod mount with wireless release, 5DII ISO 100, 1/50 F8 and 100 F2.8 L IS lens</p><div></div>

[philip_wilson]

<p>I could go on but in my experience lenses limit DSLR resolution - as the Moraine Lake 5DII crop shows (this is a $1500 lens) and even in the best case (the 3 Sisters crop is using a $1000 Macro lens) the 5DII can only match 645 film scans. I am sure a Nikon D3X Canon 1DsIII and Sony 900 will all show similar results as the sensor resolution is very similar and these are the best DSLRs for resolution. The 645 scan just shows much more detail. Even with advances in scanner technology I doubt if 35mm will match a scanned 6x8 (or 6x7) film scan as the lenses will limit it's performance. I own the Canon 7D (18MP on APS-C) and just like the DXO Mark tests it's resolution is much lower then the 5DII - limited by the lens resolution. The Fuji GX680 has a film area 5x a full frame DSLR and this is the reason why it performs better.</p>

[timlayton]

<p>I have attached two files. <br>

The red/orange flower is a lilly and was shot with my RZ67 Pro II using a 140mm macro lens at f/32 in natural light at 14 seconds. I scanned with the Epson V750 and used the following settings: 24bit, auto sharpen, image-type standard, mult-pass 4 times. The pixels are 6533 x 5260. The scan took 4 minutes and 21 seconds and the file size was 98 Mb. I did scan it at 48-bit and the file size was 196 Mb with almost no noticeable difference to my eye. The film was FujiChrome Provia 100. If this were real-world condition I would edit this image, but I wanted to upload as is.<br>

The second flower is a gold calla lilly. It was taken with a Nikon D3X at ISO 100. The pixels are 6048 x 4032 which is the default for a full frame on the D3X. I used the Nikkor 105mm f/2.8 macro lens at f/22 and 1/60. I used two studio soft boxes for lighting. </p>

<p>RZ67 Pro II Scanned with Epson V750<br>

<img src="http://i605.photobucket.com/albums/tt140/timlayton/RZ-1-24k-24bit_6533x5260-72dpi.jpg" alt="" width="1023" height="824" /><br>

Nikon D3X<br>

<img src="http://i605.photobucket.com/albums/tt140/timlayton/D3X-6048x4032-72dpi.jpg" alt="" /></p>

 

[StuartMoxham]

<p>Tim you have a D3x and the V750? What are your conclusions based on what you shoot. Do you find any advantage shootng film and scanning it with the V750 over the D3X. If you have both you are in a great position to compare the two in real world use, shooting what and how you normally shoot. Please let us know what you find and uplaod some examples for us.</p>

[barry_sanford]

<p>Tim, the Nikon 9000 does indeed work very will under Snow Leopard. It's only Nikon's bundled software that stopped working. This is really no big deal as Nikon stopped any new development on Nikon Scan a long time ago because Vuescan and Silverfast were beginning to dominate the market. Either of these can leverage 100% of the 9000's feature set. That said, I did like using Nikon Scan and wish Nikon would have continued its development and support. Unfortunately, Nikon has never been known for its software prowess.</p>