Minolta Scan Dual IV and shadow detail with negatives - Software solution?

Discussion in 'Digital Darkroom' started by hique, Jan 29, 2006.

  1. I've been testing my Minolta Scan Dual IV with color negatives.

    For slides it works fine and I am satisfied (not much shadow detail
    but one can't expect too much from a slide).

    With negatives though it is a complete junk. Ok, not a complete junk
    but very inadequate when dealing with shadow detail.

    Comparing it to Fuji's Frontier scan, I realize that Frontier
    preserve the shadow detail...even if it doesn't show in your monitor,
    but ajusting levels and using shadows/highlight function gives you a
    lot of shadow information that didn't show up at first place.

    Also those scans don't have their histogram clipped. It's like
    Frontier scans the maximum shadow and highlight detail it can
    and "compresses" the tonal range in a more contrasty final image, but
    still possible to extract that shadow detail.

    In the other hand, Minolta Dimage seems not to compress the tonal
    range but to choose, to select one part of it and transfer into the
    final image. The result is that the histogram is always clipped and
    the shadows or the highlights can not be extracted.

    While experimenting I tried to scan the negative as positive film and
    the result preserves the whole tonal range, not clipping the
    histogram, but the result is a very bizarre and mis-colored image.

    So it seems that the hardware is capable of scanning the whole tonal
    range (scanning as positive could show that) but the software simply
    can't deal with it when scanning as negative film.

    Finally my question: Is there a software solution to this problem?
    Does any company makes another software for this scanner that would
    result in better negative scans? What is your experience with that?

    I am really dissapointed that the original software is so terrible.

  2. Try Vuescan. A very viable alternative.
  3. I will try that. I decided to post an example of the scans. Notice that the shadow is almost plain black when scanning as negative. Scanning as positive conserves the shadow information but the overall image is very bad, lacking contrast and with wrong colors.
  4. I'm working on a Photoshop plugin to help with negative scanning. The idea is to scan the negative as a positive, and then invert properly using the plugin to get good results.

    Do you have Photoshop and Windows? If so, I can give you a copy of my plugin to test. Also, if you send me your original scan, I could post the results after running it through my plugin to see if it will give good results for you. For that, make sure to scan as a positive using 16-bit linear color depth, no auto exposure, and set the manual exposure to +0 for R, G, B (the defaults).
  5. I've noticed how well these minilab scanners can detect noise
    levels in shadow and hilites and will roll off their tonal response
    curves accordingly. Very smart machines.

    I got the same just barely detectable shadow detail without
    clipping as you got, Marcio. If I tried to open them up in PS, the
    noise shows up as seen in your scans. I really don't see much
    difference between the Frontier original and your Minolta original
    except in color balance which the Frontier's cool cast is balanced
    for its warmish printers. However your minolta did clip to 0RGB
    in the two black square shaped objects in your scan where the
    Frontier held up.

    Can you set a minimum black point in the Minolta? That seems
    all that's needed.
  6. Hi Marcio,

    In Minolta Scan Utility, I'd suggest:

    * scan as color slide

    * slide auto exposure off (see exposure control settings below)

    * output 16 bit linear (or 16 bit, if you prefer)

    Exposure control:

    -MAS: +2.0

    -RED: -1.3

    -GRE: -0.2

    -BLU: +0.8

    Note, above settings are what *I* settled on, with *my* scanner, for *one* particular film. You would need to go through the following exercise to determine your settings:

    Above settings determined by scanning leader and adjusting the settings to bring all 3 channels to have few or no pixels at 255. This is determined by moving the mouse around around preview of the leader and observing the readings. It can be verified by reviewing the histogram of the scan histogram, within Photoshop. The object of this exercise is to neutralize the orange mask, and it is quite effective.

    If you "play by the rules" with MSU, telling it you're scanning color negative film, you *will* get half-decent result, but closer examination will reveal blown highlights. You're not getting the most out of your film. FWIW, I am still struggling with this, well, have put it aside for a while. Also, scan times with the above workflow tend to be a fair bit longer. Not that serious with the Dual IV, but almost unusable with my 5400.

    Now, what to do with the 16 bit linear (or 16 bit).

    * You can work either in Photoshop. The former will require you to first apply a profile (such as Minolta's supplied Posi Linear Profile), and then convert to something such as sRGB or Adobe RGB.

    Then for both, you might try invert, followed by some variation of Photoshop's Autolevels, Autocolor, or Curves, or Saturation adjust, or all of the above.

    * Alternately, you can use the 16 bit linear as a Vuescan raw file, for a scan-from-disk workflow.

    Also, you might have a read through:

  7. So many good advices.

    David's idea on the actions for photoshop and Mendel's sort of "manual profile" when scanning as positive film are both very good ideas. I would love to test the actions David.

    About setting the black point...I really can't, not with the original software.

    I tried VueScan as our fellow user suggested. It is a much better software. Gives the user much more control (including setting the black point). I guess VueScan fixes my problem.

    I would like to sincerely thank everyone for the fast and helpful support. I hope I can help you all back soon.

  8. I swear by Vuescan and have used it exclusively in several Minolta scanners for about 3 years. I scan negatives as negatives and positives as positives and feel I can't improve on that. Vuescan allows you to adjust the endpoints vis a vis the histogram so you are insured of no clipping.

    Don't try to compare an unadjusted scan from your ScanDual with an auto processed scan from a Frontier. All you want your ScanDual to do is to give you a file without the histogram clipped on either end that you can adjust in Photoshop. I find a good looking scan does not make the best final print. A scan that looks flat because it captures all the data is the key.

    I suggest you download the free Vuescan demo and try it. Like it, buy the key to remove the watermark. If not, just delete it and be out nothing but your time.
  9. I have read many discussions on this topic. There are several people that have found their own methods using PS, and there is quite a variety of methods. If you don't mind the learning curve, these methods are worth trying, otherwise I agree that Vuescan is the way to go (with negs... slides you don't really need it) Personally, I love Vuescan, but since I like to learn how to do things myself (in other words... the hard way) I'm going to try the "do it yourself" methods. The good news is that you can scan as positive, then try several methods "post-scan" (keep backups of the original scans) Each film will behave differently, so something I'm also trying is to take some shots of a grayscale stepcard, using the films I have used before, and in a variety of lighting (ideally this would have been done when you took the original photos, but you can "re-create" this by shooting the grayscale in full sunlight, overcast, and shade, including a note in each shot saying which light was used) This will give a general idea of how each film behaves, specifically how it behaves when scanned on your scanner. The scans of the grayscale will also help you judge how well your attempts at converting positive scans (of negs) is working. You can also include shots in incandescent, fluorescent, ect. if you work with these much, but situations like sunsets are different... they tend to be very subjective and more a matter of personal taste. To be honest, I'm still learning all this myself, but I have been keeping up on the discussions for several years. If I have time, I'll try to find some of them and post links (to this thread, or maybe start a new thread if needed... keep an eye on this forum) Like I said, Vuescan is a wonderful solution (and is also available for Linux, for those of us going that route) so don't hesitate to pay for it... totally worth it!
  10. Hi!
    Just forget the twain driver from Minolta. Use VUESCAN, the difference is quite amazing.

Share This Page