Anyone tried copying MF negs/slides with a 5D2?

Discussion in 'Digital Darkroom' started by jamie_robertson|2, Oct 7, 2009.

  1. I have a load of MF negs and slides and can't justify splashing out the sheckles for a decent MF film scanner. Just wondering if anyone has tried copying MF slides with a Canon 5D2 and a lightbox and, if so, what the results are like.
     
  2. I looked around and there didn't see anything that would make me buy.
    Have you looked at one of the scanning services like scan cafe? I'm really thinking of doing that with a lot of my film that still needs to be scanned.
     
  3. I have done this - with good results - although not with a Canon. I think you really need a macro lens to do it properly, and a solid tripod or copy stand.
     
  4. OK, many thanks. For my needs I won't be using a scanning service. I already have a 5D2 which is why I was asking. I may give it a try this week using my macro lens and a lightbox. It will be interesting to see the result.
     
  5. Jamie, have you tried to look for a MF format slide duplicator that fits to the front of the lens? I use one for 35mm with great results. Scanning is just too costly, especially if you have a notebook full of slides both 35 and 645.
     
  6. Hi Ed,
    No, I had no idea such a thing existed. I have seen countless 35mm slide duplicators but never one for medium format. I'll look out for one of those. I suppose it wouldn't be too difficult to make one if necessary.
     
  7. Jamie, I have a several hundred MF transparencies I'd like to digitize. Your post prompted me to do a quick Google search using "medium format slide duplicator" and I found the link below. The description in the search results say medium format, but it does not specifically say that on the webpage.
    http://www.kauserinternational.com/Photography/Ohnar/Copier/Ohnar%20Digital.htm
     
  8. Bill:
    Looks too small to be a Med Format duplicator.
    I check with buyer (friend of mine) for major photo chain with 35 yrs in the biz and he has never seen one.
    The light box with macro lens is a good option. That was what I used before acquiring my 120 Scanner.
    Tips:
    • Make sure the lighting is even with a light meter reading across the area to be used.
    • White balance the light source without any slides
    • Level the camera and the surface of the light box.
    $0.02, hope it helped.
     
  9. Jaime, I have never seen one but am looking. The light box is ok except having the film absolutely flat. I would imagine that you would have to tape the film down. ????? ed
     
  10. Bill Mussett: Many thanks for that link but after studying it I think it is for 35mm only.
    Danny Wong: I agree, I have also never seen a MF slide duplicator. Thanks for the lightbox tips.
    Ed Lustoff: Thanks, I'll either tape them down or put a P size clear filter on top of them to keep them flat.
    Les Sarile: I have all three formats in my archive 645, 6x6 and 6x7.
    I am not worried about 35mm. I have my old Canon FS4000 film scanner for that.
    I am hoping to be able to duplicate my MF slides/negs to mainly use for web and A4 prints. Any larger prints will eoither be made directly from the slide/neg or from decent scans made by an outside service.
    I appreciate what you are saying about the pixel count of the 5D2 compared to the various sizes of scanned files. I don't expect to be able to match the detail from a high quality film scanner. However, I do have some 4000dpi scans of some of my 6x7 slides made at a pro lab. It will be interesting to compare these scans with duplicates made using my 5D2. If I get round to trying this I will post the results on here.
     
  11. Thanks Les,
    Like I said, I am not expecting miracles. I do have a flatbed with a transparency hood but it broke last week (thankfully). It was so bad at scanning transparencies that I rarely used it. I have no intention of spending cash on a MF film scanner so I'll see what the 5d can do.
     
  12. Thanks Les, I had no idea the V500 was so cheap (although I'm sure it will be much more expensive here in the UK like most things). I already have a decent macro lens so I'll give the camera a go to see what happens.
     
  13. Well, typically, the V500 is £183 pounds here in the UK which equates to $293 USD. Now you see why the UK is called "Rip-off Britain". If I shot a lot of MF I wouldn't hesitate to invest in the CS9000. My 5D2 gets me as close to 6x7 without all the hassle with film so I'll never shoot MF film again unless it's for a bit of fun.
     
  14. Yes, the US vs US price situation has got slightly better recently. Previously, if something cost $500 in the US it cost approx £500 ($800) here. The 5D2 was horrendous. I paid £2200 for the camera body when it retailed (at the time) for $2800 in the US (£1750). The prices have equalised now but it was annoying at the time. I could have bought a good lens with the change if I lived in the US.
    Anyway, thanks for all your advice. I'm in Sweden next week but when I return I will try copying some MF slides and post them on here.
     
  15. I could not recommend an Epson for MF. Nikon is the minimum for me YMMV.
    I wouldn't dismiss using a service as too expensive. You are not required to scan all your film at one time after all. And some appear to have a good value. I also have a lot of film to scan, and I do have a good scanner to use. But, time is the issue. Can't find enough time to scan, and if I did it would take away time from taking photos or printing or post processing.
    Even if you can find a solution do you have the time to spend on it? If you do find a solution please post it with your results. I may be interested in giving it a try if its not too expensive.
     
  16. I did try the do-it-yourself method last night using my 5DII, 24-70 lens and a Cabin light panel rated at 5000K. I need to fine-tune the mechanics of the set-up, specifically a better method of keeping the transparencies flat. Though the 24-70 has a macro focus setting, I do realize it is not a true macro lens. With this lens the transparency image covered about 1/3 of the sensor area. I cropped the transparency image out of the resulting image to a size of 1000px x 1000px. There's nothing special about that size, it's simply what I chose to use. The resulting transparency-to-digital image would certainly be acceptable for emailing, cataloging or perhaps small prints.
     
  17. Well, I couldn't resist it. I have just tried a very quick duplication of a 6x7 transparency using my 5D2. I compared it directly to a 4000dpi scan done at a pro lab of the exact same slide. I'm not sure what scanner was used at the pro lab but the scan was done 3 or 4 years ago. The file size of the pro lab scan was approx 270MB. By the time I had cropped off the edges of my 5D2 duplicate the file size was 47MB. Both files were TIFFs.
    The result? I am utterly astounded by the quality of the 5D2 duplicate. It took me completely by surprise. Is it as good as the pro lab scan? Believe it or not, the edges of the 5D2 frame are actually BETTER and the centre of the frame is slightly worse but there is surprisingly little in it.
    Before you get excited, I haven't got the time to sort out any example crops for posting on here yet. I will try and do that by this weekend before I go away. And before you all slate me and tell me I'm hallucinating I understand that a number of factors may be affecting my findings. I understand the pro lab scanner may not have been a Coolscan 9000, I fully understand that the pro lab scan might be inferior somehow (operator error or transparency not flat during the scan?). I do have a small number of other pro lab 6x7 scans to compare with so I will do a more thorough test when I have the time.
    The only conclusion I can come up with at the moment is that my intial quick test shows that I can EASILY produce nice A4 prints from my MF slides using this method. That is what I wanted so I am delighted. Whether these duplicates will stand up to a decent scan from a Coolscan 9000 is a different matter altogether. It's very doubtful indeed but I think it will be an interesting test to see what the 5D2 is actually capable of.
     
  18. Jamie, why is your 5D MkII TIFF so small? I normally get 110mb plus for a TIFF or 25mb for a RAW.
     
  19. Les: I am not at home at the moment but the pixel dimensions of the lab scan were around double that of my 5D2, at least 8000x4000 from memory. I will confirm later. I have no idea what scanner was used by the lab. I used them because they came out as the best UK scanning company after a magazine review.
    Scott: I was referring to the size of the finished file when opened in PS, not the size of the file on disk. I had to crop quite a bit from the left and right edges of the frame because 6x7 doesn't fit exactly into 35mm format. This resulted in a 47MB image. Also, the file was 8-bit, not-16 bit. Normally my full frame 8-bit TIFF files from my 5D2 open out to about 60MB in PS.
     
  20. 645 MF scanned at 4k dpi will be much bigger (9440x7080) than if you shoot it with the 5d2. If you do, get a good light table with even illumination and a good level. I did a project archiving 6x6 negs this way and the main advantage I saw is it's incredibly fast compared to scanning. Construct an L-shaped guide on the surface of the light table so you never have to move the lens.
     
  21. L-shaped guide if your transparencies are already mounted I should say. It's kind of a pain otherwise, as they usually will not lay totally flat. Even a little curl is going to screw it up.
     
  22. OK, I've spent a little more time on this.
    For clarification, after cropping out the black surrounds from both the pro scanned image and the 5D2 image so that just the actual photograph is displayed, I get the the following pixel dimensions:
    5D2 image: 4530 x 3678 (8 bit TIFF file)
    4000dpi scanned image: 10703 x 8654 (8 bit TIFF file)
    Just to illustrate, here is a thumbnail of the slide I have duplicated:
    00UhkM-179209584.jpg
     
  23. Now for the interesting part. First, I resized the 4000dpi scan to the same dimensions as the 5D2 duplicate in PS. No image adjustment has been applied at this stage (no sharpening, no levels, no colour balance, nothing)
    Here is a 100% crop from the resized file (4000dpi scan):
     
  24. And here is the same 100% crop from the 5D2 duplicate image (again, no manipulation at all, no levels, no sharpening, no colour balance):
    00UhkY-179211584.jpg
     
  25. Now, here is another 100% crop from the bottom of the same slide. This is from the 4000dpi scan:
     
  26. And now the exact same 100% crop from the 5D2 file:
    00Uhki-179211784.jpg
     
  27. Now for some sharpening in PS.
    Here's the same 100% crop from the 4000dpi scan with some unsharp mask applied:
    00Uhko-179213684.jpg
     
  28. And now for the same 100% crop from the 5D2 image. The exact same amount of unsharp mask has been applied:
    00Uhkq-179213784.jpg
     
  29. Lastly, let's look at those central crops again, this time with some unsharp mask applied. This is from the resized 4000dpi scan:
    00Uhks-179213884.jpg
     
  30. And now the same 100% crop from the 5D2 file with the exact same amount of unsharp mask applied:
     
  31. *sigh*... try again...
    00Uhkx-179213984.jpg
     
  32. So, there you have it. This is only one slide and I don't know which scanner was used so don't take this example as the norm.
    I have made two very large prints from the 4000dpi scanned file in the past (approx 30x24 inches) and they were lovely. So, I know in my own mind that the 4000dpi scan is reasonably good no matter what scanner was used. My first impression was actually one of amazement. I expected the 5D2 duplicate to be apalling compared to the scan. I was initially astounded at how good the duplication was. As you can see from my above crops, the central portion of the frame showing the cars is visibly less defined on the 5D2 file. This is quite obvious. However, I wouldn't exactly call it terrible. In fact, I would expect the same result (or possibly worse) from most flatbed scanners. I do own a flatbed with a transparency hood but it is quite old and it could not hope to compete with either of these two slide duplicates.
    The part that most surprised me was the crop from the bottom of the slide. Amazingly I would hand the trophy to the 5D2. It's not hugely better but I definitely prefer it to that of the 4000dpi scan, with or without sharpening applied.
    My conclusion so far is that this method will easily give me what I need... the ability to make decent sized prints (at least A4 size) from the 5D2 duplicates. The best bit about this method is that it is simply so quick compared to scanning. Even if I could justify a Coolscan 9000, I'll bet I can duplicate at least 10 transparencies in the time it takes the Coolscan to do one. Sure, time isn't everything but it's an excellent bonus.
    I'd be interested to hear your thoughts. Please remember that I am not here to parade the abilities of the Canon camera. Everyone knows what the 5D2 and Coolscan can do so don't start a raging battle. This is just a fun experiment and the result turned out way better than I could have hoped for.
    After I've heard a few of your comments I have one more bombshell to expose about my working method that may surprise you.
     
  33. Jamie - your results don't surprise me at all. Resolution is not a pixel dimension nor a line pair so it's useless to guess based on such simplistic measurements. It is a MTF curve where the response at low and mid frequencies is much more important to image fidelity than the response at high frequencies. Likewise, the curve for low contrast detail is more important than the curve for high contrast detail, even though test charts like the USAF one are typically at least 100:1 contrast. (That's way too high to be a useful predictor of resolving power in the real world.)
    If you don't understand how all of that relates to your results or the thread, don't worry. The short of it is this: a 5D2 can extract all the detail you need for common print sizes as you've found.
    You did this in a single shot, right? Try a true macro prime lens and stitch the 6 frames it takes to fully cover the 6x7 film frame with the lens at 1:1 magnification, carefully focused with an aperture just short of hitting diffraction. PS4 stitching is pretty much automatic and painless. You won't be tempted to spend money on another scanner.
     
  34. Three months ago I took delivery of a Canoscan 8800F scanner. It cost $165 and after using it for these three months on my collection of about 5500 slides and 700 sheets of 4x5 bw negatives, I can only say this is one of the best photographic purchases I have made in the 50 years I have been looking into camera viewfinders.
    My film of choice until four or five years ago was Kodachrome Pro 64 and 200. B&W varied between Agfa Agfapan, Ilford Delta, and Kodak, Tri-X. This scanner gives me 48-bit color depth in TIFFs so that I can post process the way I want to, not the way someone else thinks I want.
    I had quite a few of my slides scanned but was never happy with that. Now, the advance of electronics and resolution solving in modern optics has made it possible to get spectacular results without spending a small fortune getting started.
     
  35. I've done this with a Nikon, various slides and negs and lenses. The bottom line is that the results can be very good, but getting the process fast enough is a constant challenge (slide scanners aren't too fast either, but still). A dedicated copy stand would be good. Of larger slides, I've taken several exposures and stitched together to get better resolution, which requires work but leads to good results. What is critical is getting the film flat and aligned with the camera, something which is easier said than done. Another is using a good lens. I've found that at these magnifications dedicated copy lenses can be much better than regular macro lenses.
     
  36. Here's a bit of additional info:
    Amazingly, the light table I used was the lid off my old broken flatbed scanner. The scanner packed up last week so I ripped the lid off, turned it upside down and used it as a small lightbox.
    The lens I used was... wait for it... a £99 Cosina 100mm f3.5 macro!
    The tripod I used was an el-cheapo, it was so wobbly you could see the image moving in the live view screen.
    Just imagine what the result would have been like with a decent lightbox, a decent macro lens and a decent tripod.
     
  37. You got to have a continuous spectrum light source; fluorescent lights typically found in light boxes (whatever their kind) don't reproduce color well. I recommend flash with appropriate colour gels.
     
  38. Why not send out to Scancafe to have all of them done at 4000dpi without any effort?
     
  39. "Why not send out to Scancafe to have all of them done at 4000dpi without any effort?"

    1. I live in the UK
    2. Price. 4000dpi scans in the UK are expensive (just like everything else). Average cost for a 4000dpi scan of a MF slide is at least £3 ($5 USD)
     
  40. Thanks Waldo, interesting article.
     
  41. Um, am I missing something here? The example above looks significantly better - I mean obvious even after a single glance, at the small sizes shown online - for the scanned image over the one shot with the camera, both before and after unsharp masking. That said, the example image is perhaps not that well chosen; the focus is at around infinity, and it's obviously a hot environment giving you a lot of heat blur so the original negative is unlikely to have all that much fine detail to begin with.
     
  42. Medium Format scans at Scancafe are 0.99 $US.
    Using a digital camera you are adding a resolution limitation, plus all the flaws particular to the camera and lens, plus a high rate of outoffocus, etc...
    I would definitely use scancafe or buy a cheap flatbed. Even a flatbed scanner should do better than the 5DII+lens+focus+no-ice, etc.
     
  43. A V500 on ebay UK can probably be had for £100.
    Using a 5DII to digitize fine MF film captures is quite a waste.
     
  44. I casually used a D700 with a 50mm to copy/scan 6*7 negatives, although not from a lightbox but from the projection of the negatives in an enlarger. It worked out ok, since I wasn't really going after the ultra tack sharpness of the original ( Mamiya 7 II ) negatives anyway.
    00Ui53-179343584.jpg
     
  45. A number of folks (including photo.net wizard Ellis Vener) are using old Beseler slide duplicators. Has light source built in, filtration, bellows. You can go up to MF 6x7, or even 4x5 with an attachment. They aren't made any more, but show up from time to time on ebay. I got mine for $50 and it's built just like a Beseler enlarger, and takes Beseler 67 negative carriers. Works great.
     
  46. Flatbeds offer only convenience; a 5D2 with proper accessories, lens and technique will easily give a better result than a flatbed. Quality-wise the comparison is more along with a Nikon 9000, but convenience needs to be considered too as scanning takes time and effort, no matter how its done.
     
  47. Oscar, the OP was referring to medium format. And Epson flatbed can probably resolve somewhere between 2000-2400dpi plus has ICE.
     
  48. the edges of the 5D2 frame are actually BETTER and the centre of the frame is slightly worse but there is surprisingly little in it.​
    This is probably because the film is not completely flat. The standard approaches of either wet mounting or a sandwich with anti-Newton ring glass will both fix this. Quite the bother however. Stopping down the lens will help but trades off resolution.
    Focus stacking is a good alternate approach. This also compensates for unintended tilting between the camera focus plane and the light table.
    The idea is to take multiple exposures at slightly different focus offsets, then blend the sharpest part of each to form the final digitized image. In practice, might as exposure bracket, say up to +2 stops at each focus offset as well. This digs out otherwise unavailable shadow details from the more dense sections of the slide.
    The align_image_stack and enfuse utilities bundled in with the open source Hugin is an easy way to automatically process the whole stack in one go. Search for "align_image_stack" in this thread for the actual command line parameters. The context is a bit different, but look further up that thread to see what this can do for a compact digicam at 800ISO.
    Daniel Taylor's suggestion for multiple planar stitches is good too. Combine both approaches, and you can probably expect final images as good as those from a Nikon scanner.
     
  49. "Using a digital camera you are adding a resolution limitation, plus all the flaws particular to the camera and lens, plus a high rate of outoffocus, etc... "
    Well, I can understand what you are saying about resolution limitation but pixel count means very little. A 4000dpi scan of a 35mm frame is around 26 megapixels. The original 5D is only 12 megapixels but wipes the floor with 35mm. Once film grain becomes visible in a scan film has nothing left to give. My quick 5D2 example above is obviously inferior to the scan but lets remember that it is a QUARTER of the resolution of the original scan. Regarding out of focus shots, that's only going to happen if I am either drunk or seriously incompetent. Manually focusing on the 10x Live View guarantees perfect focus unless you've got fingers like cow's udders.
    "I would definitely use scancafe or buy a cheap flatbed. Even a flatbed scanner should do better than the 5DII+lens+focus+no-ice, etc."

    The reason I am doing this is because my own flatbed is crap. It couldn't hope to compete with my 5D2 duplicate. Yes, today's flatbeds are far better and although I don't have a V500 I find it hard to believe it can do much better than I've already done with the 5D2. If it did outdo my 5D2 duplicate it would be just about as good as that original 4000dpi scan done by a dedicated professional MF film scanner.
    I do agree that lack of ICE is a disadvantage. However, I could probably duplicate the slide and manually remove the dust in the time it would take a Coolscan 9000 to do the scan.
    I'm not trying to be clever. As I said at the start of my post, all I want is to make decent A4 prints and use my MF photos on the web. My method has proved to be quick and capable of doing just that. If I wanted A2 prints it would be a different matter.
     
  50. "This is probably because the film is not completely flat. The standard approaches of either wet mounting or a sandwich with anti-Newton ring glass will both fix this. Quite the bother however. Stopping down the lens will help but trades off resolution."
    Transparency was perfectly flat, aperture of f8 was used.
     
  51. Jamie,
    If you think a 12mp camera is far superior to a 35mm 4000dpi scan or that a good flat bed would be just as good as the Coolscan 9000 you are incorrect and should experiment more before embarking into a large number of scans. Now if you only need it for web and A4 prints it probably won't matter other than you would leave out shadows and highlight texture behind by not using a dedicated film scanner.
    Robert's comment on the film's not being flat was regarding your 4000dpi scan. What equipment did you use?
     
  52. "Jamie,
    If you think a 12mp camera is far superior to a 35mm 4000dpi scan or that a good flat bed would be just as good as the Coolscan 9000 you are incorrect and should experiment more before embarking into a large number of scans.
    "

    Many 12mp cameras don't get anywhere near 35mm quality but the original 5D is easily superior. I know, I have vast experence with decent 4000dpi 35mm scans and the 5D. The 5D wipes the floor with 35mm film. Those who say otherwise haven't used the 5D or haven't used 35mm. As for the 5D Mark II, it is in a completely different league to 35mm.
    I did not say a flatbed could match the Coolscan 9000. I don't even know where you got that idea from. That would obviously not be the case.
    "Robert's comment on the film's not being flat was regarding your 4000dpi scan."
    No it wasn't. I suggest you read Robert's post again or ask Robert himself.
    "What equipment did you use?"

    You obviously haven't read this post very well, the equipment I used has been discussed already.
     
  53. Jamie,
    I just read pro-lab and didn't see the equipment.
    Can you post any of your experiences of the 5D next to a 35mm 4000dpi scan to illustrate what you mean?
    My 35mm TMAX, Velvia, Ektar, etc have far more detail in them than my Coolscan 9000 can capture.
     
  54. It is good to steer away from digital vs film.
    People get sensitive when a discussion turns into comparing film vs digital.
    Comments like the 5DII is in a complete different league than 35mm film are invalid assertions that lead to unproductive discussions. You can always scan your TMAX and Velvia 35mm with an Imacon or better and dig more detail than a 5DII (or any DSLR) can capture. What we are discussing is the quality of the scans not the quality of the film capture.
     
  55. "Can you post any of your experiences of the 5D next to a 35mm 4000dpi scan to illustrate what you mean?"

    Sure, here's a 100% crop from a 5D image without any sharpening applied:
    00UiCV-179395684.jpg
     
  56. And here's a 100% crop from a 4000dpi scan of a 35mm transparency (Provia 100F, ISO 100, 24-70mm f2.8L). This is also unsharpened but it has been resized (i.e. scaled down) to match the size of the 5D2 crop.
    00UiCa-179395784.jpg
     
  57. Thank you. Both look good but it is difficult to compare with different subjects. Do you have a direct comparison of the the subject?
    Here is a sample of 4000dpi scans (at 100%) of different 6x7 films with my Coolscan 9000. Also the same take was made with a Canon 40D (not to compare - just for reference).
    http://shutterclick.smugmug.com/Photography/Resolution-and-Diffraction/6302153_PLzKe#407096809_FeZJ4-O-LB
     
  58. Also here is another sample scan with the Coolscan 9000 of 35mm Velvia next to the same setup with the 40D (different cans). Shot in ideal conditions so the lens was not limiting factor on the 40D but just its resolution. In general the 5D would be less demanding on the lens but that didn't come into play in this test.
    00UiCv-179399584.jpg
     
  59. Thanks Mauro, interesting article but comparing the 10MP 40D to a coolscan scan of a 6x7 frame seems idiotic. It was obvious the results would not compare in the slightest. If they did it with a 5D2 it would be far more relevant.
    Your example of the 40D compared to 35mm Velvia is an excellent illustration. Unfortunately I do not have identical images to compare the 5D to a 35mm scan. However, I do think the full frame 5D would perform way better than the 40D in that test. Not sure what version of Velvia was used (50 or 100?) but surely it would be fairer to compare an ISO 100 film to the ISO 100 40D.
     
  60. Yep I used probably used Velvia 50.
    Here is a comparison of the 5DII and a Coolscan scan of Ektar 35mm (not quite as good as Velvia). Sadly different subjects:
    http://shutterclick.smugmug.com/Photography/New-2008-Ektar-100-35mm/6499685_dJwsh#676248506_pCZoc-O-LB
    And here the 40D next to 35mm Ektar:
    00UiDT-179401584.jpg
     
  61. Hi Jamie
    I have just completed copying a large bunch of MF & LF negs using a macro lens and the results are great, I think transparencies are going to be a bit more difficult for color correction.
     
  62. Peter: That's good to hear. Glad you're getting success too.
    Mauro: Yep. 35mm Ektar seems to give better results compared to the 40D in those direct comparisons. I agree.
    I'm a bit wary of those shutterclick links though. The so called 6x7 shot of those autumn (fall) leaves is obviously a 35mm shot that has been sized down (the shot is identical to the 35mm shot, even the leaf position and depth of field is the same).
     
  63. Not to be weary. All those shots are mine.
    The 6x7 is a resize of the 35mm to show what a 20 megapixel scan would look if it had been taken with 6x7.
     
  64. Before this veers off too much, it's worth remembering what Kirk Tuck and others have been pointing out lately: for the vast majority of camera users (including professionals), for almost all of the time, any modern digital camera or film camera will give you all the resolution you need. Around 8-10mp is sufficient for all but really large printing, and is something all dslrs and film cameras can do with a good lens, low iso and good shooting practice. And of course most images today are never printed. They're used online in one way or another, and there 10mp is not only sufficient, it gives you lots and lots of margin for cropping, noise removal and so on. What medium and format to choose is mostly about other qualities (color fidelity, dynamic range, noise/grain, "feel" and so on).
    Copying with a camera is perfectly fine; it will give you all the detail you need. After all, normal 1:1 macro lenses were originally meant for precise copying too. The issue is really mostly if it's a good way if you need to copy more than just a few images. For larger amounts of images a scanner is probably a faster and more reliable way to get consistent results.
     
  65. The only practical MF slide duplicator I know, is the NOVOFLEX Castel-Cop-Digi, for slides up to 6x7cm. To keep the excellent quality of the slides, an excellent lens should be used for copying MF slides with a DSLR.
     
  66. Transparency was perfectly flat, aperture of f8 was used.​
    If you're not getting equal sharpness across the field, then consider that film scanning plane might not be quite perpendicular to the lens boresight. This was the problem that I ran into last (100mm f2.8 macro on a Canon XT trying to stitch scan 6x7 Neopan 400.) The nastily designed, re-purposed Epson film holder didn't help in keeping the film flat either.
    Getting tolerances tight on jerry rigged mechanical fixtures isn't easy. In my case the plentiful film grain clearly highlighted areas of trouble, manifesting itself as unsharp grain in sections of the digitized frame. In any case, my recollection was that focus stacking worked well enough in addressing this.
     
  67. Jamie, your results do not surprise me. I've been telling people to use a good P&S to copy prints for a while. I think that that the 5D MkII has an amazing sensor, combine that with a decent glass and you can get decent results. At least much better than a cheap flatbed scanner that has a cheap sensor and plastic lens. You also avoid scratches and dust as major issues. Film flatness may be less of an issue as well. Framing and perspective are more of an issue and keep this from being real easy IMO. But, good results and sounds like it will work for you.
    As your results show, the 4000dpi scan has a little more info, and that may be desirable on some photos and scanning would be the choice for those photos. But for a lot of photos this may be the best way to go for some people. For me, I may try scan cafe and not worry about capturing stuff myself. 4000 dpi won't be necessary for most of my shots but limiting my time on the project is a major selling point.
    Les, personally I'm a detail nut. I can't wait for a 40mp DSLR and when MF digital breaks 100mp (if I could ever afford that). I haven't shot 35mm film for like 15 yrs or more because it could not do what I wanted. 35mm film is limited in the sizes it can print, even with great technique and great gear your limited to a little more than 11x14 in a print. But, given real world conditions a 11x14 is not likely and 8x10 is more realistic for most images. But, I have been back to DSLRs for like 5yrs because they get better results for some subjects than film and MF is just a pain for some situations. During that time I had used MF for my important shots, like landscapes and such. Now, I may use the 5D MkII more. To say that 35mm film is better is to ignore what DSLRs do very well and how that relates to image quality. I've printed 10mp files to 24x36 (for some subjects) and have had MUCH better results than I ever did with 35mm film. Resolution is not the most important factor. To be frank, I don't see any situations where 35mm film does better than the current generation of DSLRs. IMO scanning is a necessary process, but less desirable because it is a second generation of an image. As soon as technology progresses MF film will be passed as well. And I can't wait.
     
  68. Les, I'm sorry but that is an absolute nonsensical method and means absolutely nothing about image quality. All you are quoting are scan resolutions compared to the pixel dimensions of a DSLR frame. What you need to be looking at is the resolving power of the negative/slide you are actually scanning, not the pixel dimensions of the scan.
    Let's say you scanned a 110 negative. Let's say you scanned it on a drum scanner at 10,000 dpi. That would give a final scan with pixel dimensions of around 5000x7500. Are you seriously saying that it would give superior image quality compared to a Nikon D3x file? Heck, a 6mp compact would blow the 110 out of the water, let alone a D3x.
    If it was possible to scan a 35mm Velvia transparency at 50,000dpi would you then say that 35mm outresolves the detail of a 60MP MF digital setup?
    35mm is good, I still use it and I enjoy it, but DSLRs have long since left it behind.
     
  69. Well said Matt, I agree.
    Les, I am not here to bash film. I was out shooting Ilford FP4 yesterday on my old Yashica Electro. I am off to Sweden tomorrow and I will be using my Olympus MjuII (Stylus Epic) loaded with Delta 3200 (in addition to my 5D2 kit). I love film and I would hate it to disappear altogether. But film has grain, and once that grain can be seen in a scan it has nothing left to give no matter how many dots per inch you scan it at. 4000dpi is more than enough for all films. Film technology is no longer advancing (not much anyway) whilst digital is screaming ahead at light speed.
    As for your comment ("Perhaps vast experience with bad gear, bad use of gear or both?") my gear is fine thanks and I object to being accused of using my gear badly. I'll gladly put my 5D2 up against any 35mm kit and those people like me who have used both systems extensively will not only agree, they will be laughing at the suggestion that 35mm can in any way compete.
     
  70. Les, all those numbers mean are the size of the finished file. I could take a file from a 4MP compact and interpolate it in Photoshop to 25000x15000 pixels but it wouldn't make the image quality better than a MF digital system. Same with film, you can scan it at the highest dpi imagineable but once past 4000dpi or so you're wasting your time going any further.
    As for your new image comparing the dynamic range of Ektar 100 to various DSLRs, well that's very nice thanks. What will you post next? A comparison of turnip prices comparing Poland to Alaska? Excuse my sarcasm but dynamic range is not something we are discussing on this post. Perhaps you are bringing it into the discussion to divert attention away from your flawed pixel count theory.
    Anyway, like I said, I'm off to Sweden in the morning so I don't have the time (or the interest) to photograph coloured pencils with my 5D2. I'll be in Stockholm photographing more interesting things (brick walls, USAF charts perhaps? sorry... sarcasm again).
    To keep you going while I'm away, here's another person who has already compared the dynamic range of film to DSLR. Perhaps you'd like to meet up with him to talk it over. Don't forget your coloured pencils though... ;-) (link)
     
  71. Just before I dash off, show me a piece of 35mm film that can do this:

    [​IMG]
    Full frame (Canon EOS 5D Mark II)


    [​IMG]
    100% crop from same image, unsharpened.


    I took this shot earlier this year. I'm afraid there were no coloured pencils handy, just this boring bird of prey ;-)
     
  72. Les, show me your 35mm piece of gelatine that captures more detail than my humble bird of prey shot and I'll eat my own hair.
    You've posted plenty of other examples, why be afraid of this challenge?
     
  73. Jamie,
    Nice shot. The eye of the bird you posted is an example of how the camera sharpens after applying its own noise reduction (regardless of your not doing processing) that makes the image artificial - see bottom eyelashes.
    I tried to find an example of 35mm film scanned with the Coolscan with a similar eyelashes proportions but the only one I had handy was of TMAX 400 pushed to 800.
    Here it is----> a 21+MP scan of 35mm film iso 800 scanned with the Coolscan. Obviously you always have the film to get more detail if you upgrade your scanner years down the road (This is scanner limited and not film limited):
    00UimO-179637784.jpg
     
  74. 100% crop (remember it is at iso 800 and no sharpening):
    00UimR-179638084.jpg
     
  75. I'll try to find an eye in the same proportions on iso 100 and scan it at 8000 dpi. I'll do that only if you promise not to eat your own hair though - I read a girl in India had to have surgery for it.
     
  76. Clearly by your eagle shot the 5DII has an outstanding power to capture detail. It is a blessing of technology for all of us.
    That said, it certainly does not capture as much detail as 35mm film (though it is comparable to a 4000 dpi scan of 35mm film - scanner limited capture). Also I am not sure it would be better than a flatbed that has ICE and decent parallel focus for the OP's task.
     
  77. Here what I did was to sample down my shot to the resolution the 5DII would capture in ideal focus across the frame (not realistic) conditions with its resolving power of 2800 lines over the 2 1/4 inches of medium format (1250 lines per picture height) and sample back up to compare to the Coolscan 9000.
    Using the eyelashes theme:
    00Uio8-179647684.jpg
     
  78. As you can see, even with theoretically perfect results, the 5DII would be too lossy to consider it as the tool to digitilize someone's entire medium format film library.
     
  79. A flatbed scanner could just about double the effective linear resolution of the 5DII for this task.
    A Coolscan could quadruple it.
     
  80. To keep it in perspective, if you will print at A4 size it won't matter:
    00Uipd-179651684.jpg
     
  81. Another for a different size monitor:
    00Uipj-179651984.jpg
     
  82. A4 is about letter size so the 5DII capture should be plenty. If later on you decide to print significantly larger you may need to evaluate and decide whether a re-scan is needed.
     
  83. ok....now that the rampage of off topic comments have passed by.....someone above said not to use light boxes because of the flourescent tubes not being able to reproduce color corectly. I know flourescent lighting in photography is a pain, but if the light in a scanner is flourescent and it scans color slides and negatives properly................what's the problem. Or am I missing something.
    And on a further thought.....my monitor is calibrated to produce the right colors and right luminence. Has anyone ever used a monitot to copy negs/slides. I mean, go into photoshop, create a 255, 255, 255 image, size it to your monitor screen, and you should have a perfect light table. Albiet keeping the neg flat is still a concern, but all the rest of the concerns should be eliminated. Right? Or am I missing something here also?
     
  84. Thomas, not all flourescents are created equally. They don't give you a whole spectrum, but only light at some frequencies. A light box flourescent is probably selected to give us a decently accurate representation of color, but the DSLR or scanner has sensor sensitivities different from ours, so the images may get a color cast even though they look correct on the light table to our eyes. The scanner avoids this by matching the fluorescent type and scanning sensor of course (as well as recalibrating itself on startup).
    A screen is almost never neutral. "Pure white" isn't. Don't forget that the calibration is (or at least should be) made in relation to the ambient light in the room. That said, I doubt there'd be much visible color cast left if you manually tweak the color balance of the resulting images afterwards. Unless accurate color representation is important to you it's certainly worth a try.
     
  85. Janne....I calibrate my monitor in a total dark room....and I also edit my pics in a total dark room. So my white should be really close to actual white I would think. Yes?
    Anyhow, i understand the scanner light and light table now.....thanks.
    One more thing....forget color for now....besides 99.9% of my color is done with a digital camera these days....film for me in color is pretty much gone. But I shoot tons of B&W film in medium format. So, given that I would mostly be doing 5D copies of B&W MF negative film....color balanced light pads should be a non issue then. right?
    Only reason I ask is altho I'm happy with my Epson 700 scanner's scans of my MF film, I'm moving into probably a smaller apartment, and ANY reduction in equipment would be a blessing. That Epson 700 ain/t exactly small. So a littlle slim 4x5 light pad would be a blessing if it worked.
     
  86. Wow, got to say the images posted here just confirm what I experience. The current gen of DSLRs pretty much have more resolving power, not to mention less grain, than film. Its not even close, at least to me. And, yes I still have a film camera, but its MF. For scanning I have the Nikon 9000. It does a good job, but it can't pull all the detail out of the film which is a little frustrating. So, I'm not trying to bash film, it still has its place. But to say 35mm film out resolves 21+mp DSLR, well it does not reflect my experience. IMO the photos presented so far are arguments for DSLRs and against 35mm film.
    Jamie, that eye detail is amazing. Love the crop, how does it look printed?
     
  87. Ridiculous. There are plenty of reasons why one might wish to 'scan' film in a camera that have nothing at all to do with the foolish and ultimately personal arguments about whether a digital camera can 'beat' the performance of film. Perhaps it's time some of you film devotees simply admit you like film better, and stop trying to prove to everyone, at any hint of an opportunity, that film will always rule. Or whatever. I know you don't think that's what you're doing, but that's the way it comes across.
    Simple fact is, the ability to scan film very quickly to make working prints or integrate film into a digital workflow is very valuable for some. And that value has nothing whatsoever to do with the ultimate quality of the scans. A well made camera scan is capable of making pretty darn good files, but the real value, likely, is in helping to archive film in what will--for most photographers--ultimately be a digital world.
    So rather than taking your time degrading the quality of a lovely scan to show what a 5DII "would" have captured (!) how about addressing the topic of the thread. That is to say, to add value to the conversation about scanning MF film with a 5DII? Les and Mauro, all respect; it's clear you care a great deal about the points you're making. They're just misplaced.
    Ridiculous and unhelpful.
     
  88. Scott,
    Of course you can "scan" your MF film with a 5D2. If you expect it match the quality of a good film scan, then no, it won't
    Anything else?
     
  89. Earlier in the thread someone mentioned using a flash as a light source instead of a light table. Any thoughts on what a reasonable setup would look like? I have a DSLR, macro lens and flash but am stuck as how to evenly illuminate a MF slide while holding it flat.
     

Share This Page