Vuescan & Erik Krause's Advanced Workflow = BROKEN

Discussion in 'Digital Darkroom' started by rishij, Nov 19, 2008.

  1. Hi,
    <p>
    I was a big fan of Erik Krause's Advanced Workflow for scanning color negative film with the Nikon film scanners.
    <p>
    It went something like this:
    <p>
    -Preview<br>
    -Lock Exposure<br>
    -Preview again<br>
    -Lock Film Base Color<br>
    -Amongst the 'film base colors' in the 'Color' tab, the 'film base red' is always of the highest value. Divide
    it by itself to get 1; insert this under 'red analog gain'<br>
    -Divide 'film base red' value by 'film base green' value; insert this value under 'green analog gain'<br>
    -Divide 'film base red' value by 'film base blue' value; insert this value under 'blue analog gain'<br>
    -Uncheck 'lock film base color'<br>
    -Preview again<br>
    -Now, when you check 'Lock film base color', under the 'color' tab, 'film base red' should equal 'film base
    green' should equal 'film base blue'<br>
    -If not, repeat above procedure until they do.<br>
    <p>
    This was a way of getting rid of the orange mask from roll to roll. The above process was carried out for any
    given roll, and then the rest of the roll was scanned with 'Lock exposure' activated as well as 'lock film base
    color' activated, with 'film base red/green/blue' values set to '1' under the Color tab. This led to
    color-unbiased scans of negative film by using the individual channel analog gains on Nikon scanners.
    <p>
    This no longer works on the latest Vuescan (I've tested on 8.4.60 and up). Presumably because the software is
    automatically doing some orange mask subtraction when 'color negative' is selected as 'media type'... negating
    the need or option for red/green/blue analog gain adjustments to get rid of the orange mask.
    <p>
    Can anyone else confirm this?
    <p>
    Can anyone else comment on HOW Vuescan determines the orange mask color? Does it scan spaces in between frames?
    If so, how accurate is it, and is there really any advantage then to Erik Krause's advanced workflow using
    red/green/blue channel analog gains to get rid of the orange mask cast using Nikon scanners?
    <p>
    Roger Smith & Mendel Leisk, in particular, dying to hear your comments :)
    <p>
    Cheers,<br>
    Rishi
     
  2. "Amongst the 'film base colors' in the 'Color' tab, the 'film base red' is always of the highest value." The above is not always the case for me. I'll cut-and-paste from my spreadsheet of the few color neg rolls I've redone with Super Advanced Workflow, at the end of this posting. I know Photo.net wraps single carriage returns, so hopefully it will behave... Also, doing and the SAW and *then* doing Lock Film Base Color seems to me (but I'm not sure) to be superflous, ie: doing SAW takes over the task of LFBC. I'm still a little puzzled by LFBC, at least when outputting only Vuescan Raw File, which are supposed to *ignore* Color Tab settings, and I believe this is true. There is one exception to raw output ignoring Color Tab settings: "Color|Scanner Color Space", which *will* change the raws, if they're saved at "Save". It's imperative to leave that setting on "Built In". This is contrary to Vuescan documentation, but is my experience, and I've read of others corraborating this. Regarding using SAW and setting Input|Media to "Color Negative": I wonder about this, specifically, I think I'll repeat the exercise, but set media to "Image". I do know Vuescan shifts the green and blue channels dramatically if media is set to Color Negative Regarding Vuescan's ability to "lock film base color" on the fly, from a regular frame of film, I don't think it does that well. Ed Hamrick suggests using clear leader to accomplish this, as being the best approach, hence "Advanced Workflow". Do this: preview a regular frame with LFB unticked. Now tick it and refresh if needed. No change, right? Check the film base values in the Color Tab. They're markedly *different* than the similar values obtained previewing the leader... Also, I'm going to post a couple of Raw File Examples. The first, in this post, is with Vuescan Advanced Workflow. I'll post a second, which is with Erik's Super Advanced Workflow. It's frame 1 of roll 1. My Super Advanced Workflow calcs follows (the few rolls I've redone this way): (note, roll 7 is the one that is "misbehaving", giving apparently underexposed raw file output, not sure what to make of it) on001 film base gain red 0.674 1.448 gre 0.686 1.423 blu 0.976 1.000 rgb exp 1.519 inf exp 1.173 on002 film base gain red 0.971 1.000 gre 0.754 1.288 blu 0.946 1.026 rgb exp 1.964 inf exp 1.166 on007 film base gain red 0.979 1.000 gre 0.867 1.129 blu 0.967 1.012 rgb exp 1.858 inf exp 1.166
    00RXxu-90171584.jpg
     
  3. Same frame, but Super Advance Workflow (these are reduced raw files, btw)
    00RXxy-90173584.jpg
     
  4. My third para should start:

    "Also, doing SAW"
     
  5. Did you save any .ini files for film types from older versions of Vuescan? If so, try reusing the analog gain values for your new roll of film, assuming it's the same type as the old. There shouldn't be too much variation from roll to roll.
     
  6. Just for kicks, try this: 1. Set Input|Media to Image (or Slide Film, with type Generic Slide Film, looks to close to or identical) 2. Untick any and all Exposures and Lock Film Base Color, then preview leader (crop to exclude border) 3. Adjust the 3 analog gain settings to get all 3 peaks near right end and sync'd. This will take multiple previews to show the results. Move mouse around preview and watch the values. Out-of-focus previews might be better for this. 4. Scan thus and save Raw only. I'm finding this to be very close to the color balance, but slightly "punchier", comparing to SAW with media set to color negative. Note, this is more-or-less per Bart van der Wolf's Scan Elite 5400 color negative workflow. So, Exposure not locked, media set to Image, and analog gain values (for me, based on my roll 1 leader) Here's a downsampled raw, same frame doing this workflow, again roll 1, frame 1: (BTW, this is using version 8.4.94)
    00RXzd-90189584.jpg
     
  7. The above worked very well with my problematic roll 7, where SAW yielded grossly underexposed raw files, ie: no pixels above 80.

    Setting Vuescan's image type to "image", and then ajusting analog gain to achieve white preview of leader, seems to be doing a similar neutralizing of the film base.

    *Not* locking exposure seems to be working as well. When you go onto real frames, Vuescan assesses exposure, frame-by-frame, but seems to keep applying the color bias you've set with the custom analog gain settings.

    One thing doing this workflow: the previews don't look to hot, basically like somwhat washed-out color negative images, but without the orange look. The Vuescan Raw Files look good though.

    I've yet to do some scan-from-disk from these, and am still sceptical: it's often the case that something looks promising, but then you hit snags. Anway, give it a try if you like, and see how it's working for you. I'll post some more once I do say a whole roll, output raws, and scan-from-disk from the lot.
     
  8. However good the raws look, the scan-from-disk results aren't that great. Put it aside for a while I think.
     
  9. Holy Moley!

    I must have been away from school that day!

    I have had Vuescan for a few months, but, I have never heard tell of this sort of thing. I looked-up Erik
    Krause's web page, but, it seemed to be all about stitching panoramas and contrast masking...nothing about Vuescan.

    Would one of you kind souls please point me towards a source for these Vuescan "Advanced" and "Super Advanced"
    workflows? I would like to read-up on this and give it a try.

    Thanks in Advance!

    Cheers! Jay
     
  10. Jay, Vuescan's Advanced Workflow is as close as the Vuescan help file:

    http://www.hamrick.com/vuescan/html/vuesc15.htm#topic12

    and for Erik Krause's Super Advanced Workflow:

    http://www.pcreview.co.uk/forums/thread-1939580.php

    (first hit on Google search of above string)

    Here's my take on SAW:

    **************

    Super Advanced Workflow:

    (Vuescan Advanced workflow, following up with adjusted anolog gain settings adjusted according to film base color
    settings)

    (Thanks to Erik Kraus, the author)

    1. Ensure Input|Lock Exposure and Input|Lock Film Base Color (if showing) are both un-ticked.

    2. Preview a clear leader. Set crop to a smallish but decently representative area near center, the uniformest dark
    area. For insurance, Set Crop|Buffer to 15% (this will insure 15% within your crop edge is ignored).

    2. Do a second preview. (Not 100% sure this re-preview is needed, but it doesn't take long) After the preview, tick
    Input|Lock Exposure. Do a third preview, and then tick Input|Lock Film Base Color.

    (Steps to here are essentially Vuescan's "Advanced Workflow")

    3. Record the values:

    Input|RGB Exposure

    Input|Infrared Exposure

    Color|Film Base Color Red

    Color|Film Base Color Green

    Color|Film Base Color Blue

    4. Untick Input|Lock Film Base Color, but leave Input|Lock Exposure ticked.

    5. Set the Input Tab's Analogue Gain for the the color channel having the largest film base color value to be: 1.

    6. Divide one of the lesser FBC values into the largest FBC value, and assign that result as the Analogue Gain for
    that lesser color.

    7. Divide the second lesser FBC value into the largest FBC value, and assign that result as the Analogue Gain for
    that lesser color.

    8. Untick Input|Lock film base color, but leave the Input Tab Exposure values set,

    To check: do a preview. Tick Input|Lock Film Base Color and verify in the Color Tab that all Film Base Color's are
    close to value 1.000. If they are way off, something is wrong, and you need to repeat the process. If they *are* close
    to 1, override them: set all 3 to values 1.000.

    Note, a spreadsheet is very useful, both for logging the values for various rolls of film, and to do the Analog Gain
    setting calculation.

    **************

    And yet another color negative scanning approach you might try:

    Click on the link:

    scanhancer_manual.pdf

    on the following Scanhancer site page:

    http://www.scanhancer.com/index.php?art=17&men=17
     
  11. Thanks Mendel!

    Cheers! Jay
     
  12. OK, so here's my problem guys:
    <p>
    When I go through the SAW procedure now, I get the following values for film base colors:
    <p>
    Film Base Color Red: .956<br>
    Film Base Color Green: .956<br>
    Film Base Color Blue: .954<br>
    <p>
    My point being, these values are insanely close to one another, resulting in hardly any application of individual channel analog gains. Like for the above values you get a Blue Analog Gain of 1.002. That's retarded.
    <p>
    BTW I'm scanning a leader of Kodak Portra 160NC.
    <p>
    I tried back to version 8.4.18. Same problem. I trashed my versions prior to that, unfortunately. Does anybody have a 8.3.?? for Mac?
    <p>
    So my conclusion here is that Vuescan seems to be doing some sort of automatic color negative mask subtraction when 'Input|Media' is set to 'color negative'.
    <p>
    Here's some irrefutable evidence to support my hypothesis:
    <p>
    Here's a snapshot midscan of Vuescan scanning the leader (part of the leader is exposed, but don't worry, when I actually went through the SAW, I had a frame of nothing but unexposed film), with 'Input|Media' set to Slide Film:
    <p>
    <img src="http://staff.washington.edu/rjsanyal/ScanningWoes/InputSetTo_SlideFilm.jpg" width=800>
    <p>
    Now, here's a snapshot midscan of Vuescan scanning the leader with 'Input|Media' set to 'Color Negative':
    <p>
    <img src="http://staff.washington.edu/rjsanyal/ScanningWoes/InputSetTo_ColorNeg.jpg" width=800>
    <p>
    Note that in both cases, I reset everything to 'Default Options'. So RGB exposure is 1, Lock Exposure & Lock Film Base Color aren't even options on the menu (yet). And Color Balance is set to 'none'. And Scanner Profile is set to 'Built-in'.
    <p>
    So, my conclusions is: clearly Vuescan is doing some automatic color subtraction if it's scanning the leader as clear (not orange) in the preview window. Note that in this particular version of Vuescan pictured (8.4.29), Ed Hamrick hadn't implemented the 'scan color negative as positive during scan'... so, in the preview, the film leader should appear as orange, no?
    <p>
    So, any ideas anyone? And does anybody have a version of Vuescan prior to 8.4?
    <p>
    Many thanks,<br>Rishi
     
  13. OK, guys, I figured out what was throwing me off.
    I contacted Ed Hamrick, and he mentioned that, for whatever scanners support individual channel gains, Vuescan has, and always has, applied a default gain to the green and blue channels to neutralize the orange mask. That is, none of the film is analyzed; it's just a generic application of gain.
    Now, why was I getting like completely equal film base color values?
    Well, it was just that whatever extra exposure Vuescan does for the green & blue channels happens to very effectively get rid of the orange mask on Kodak Portra 160NC.
    When I threw in a leader of Fuji Pro 160C, which appears much more red to the naked eye, I got film base color values of 0.96, 0.68, 0.568 (for R, G, and B, respectively).
    So it was just that whatever gain bias Vuescan applies to the green and blue channels effectively makes Kodak Portra 160NC leader appear as neutral gray :)
    Thanks everyone for helping out. Once again, I should have just figured this one out myself, but it helped that Mendel: you mentioned you don't always get the film base color RED as the highest channel -- that in itself hints at the fact that Vuescan, from the very beginning, biases the blue & green channels because, c'mon, all negative film leaders would otherwise have a higher 'film base color red' than 'film base color green' or 'film base color blue'.
    Cheers,
    Rishi
     
  14. Hi!
    I am trying to archive my colour neg. collection in TIFF or RAW. Trying all this workflows (SAW doesn't work in ver 8.5). Now I am trying workflow with manualy adjusting RGB gains (Lock exposure unselected, colour balance to none, output colour space = device rgb) and saving in 48bit tiff, input media to image. I have Minolta DimageScan 5400.
    Can anyone give me some recomendations for new Vuescan (where scanning in colour neg. doesn't work right).
    Some more questions
    Is locking exposure important?
    What's the best way to make a batch - or how to adjust scanned files in A-Lightroom or A-PS?
    Should i assign any ICC profile to this files?
    For Example: my rgb Gain values:
    R: 1
    G: 2.54
    B: 4.36
    Is that not too much? And values in preview are around 180??? I can't get them higher to be neutral.
    Many Thanks,
    Gregor
     

Share This Page