IT8 targets, don't see any real benefit

Discussion in 'Digital Darkroom' started by juke, Oct 23, 2009.

  1. Hi,
    I had my previous scanner calibrated with WF's IT8-targets and I have go thru calibration process with my current scanners (V700 & Minolta sc IV) including three scanning softwares.
    However, I have become more and more sceptic with the targets. They just don't seem to work correctly or I have completely misunderstood their purpose.
    For example, I scan Fujichrome Astia with V700 using Vuescan - the output file will have gamma1.0 and I assign the proper profile in the photoshop.
    That works well with the IT8 target and.... well, it works with any other color slide (astia, provia, sensia), but the color just aren't same as in original slide.
    The black is not really black, the whole picture lacks kind a vividness. And usually there's more or less magenta color cast in entire scan (especially with astia and provia).
    Now, correct if I am wrong, but I have understood that the purpose of IT8 target calibration is to get as good digital reproduction from the original color slide as possible. But it seems that the result is never any closer than just scan with fully automatic settings or scan RAW (or linear gamma, ..) and then correct all colors manually using curves tools.
    Where did I got it all wrong?
     
  2. If you have Silverfast, the IT8 calibration sequence is built into the software. The profiles can be used with other color managed programs as well as Silverfast. If you calibrate on your own, you must calibrate with most of the automatic adjustments and profiles turned off (other than exposure), and subsequently scan using the same settings, but using the profile. The profile is generated by comparing the scan with the reference file for that particular target, using software designed for that purpose.
    If your scanning software can't use an external profile (e.g., Nikonscan or Epsonscan), you can still scan with the automation off, then "Assign" the profile in Photoshop. It's best then to "Convert" the image to one of the standard color spaces, like Adobe RGB or sRGB.
    For better blacks and contrast, you can adjust the white and black points using Photoshop/Levels (or some other histogram program).
     
  3. Yes that's exactly what I do. I scan using 'raw' output (gamma 1.0) which I have used for creating profile.
    After a scan, I assign proper profile in photoshop and convert it to the working color space.
    But where's the beef? I really don't get any better results with profiles than with scanning with fully automatic modes with various softwares.
    Does anyone have simple examples of color slide scanned with calibrated process and then same slide scanned without any calibrations, just with the scanner's automatic levels & color settings?
    If you will gain some advantage with calibrated scan, please show it. Where should I see it? The color balance - no. The contrast - no. Tonality - no...
    Here's an example scans from 120 Provia 400.
    The first is scanned with Vuescan's raw mode and proper profile assigment:
    [​IMG]
    Here's same frame scanned with Epson's scanning software, no color management:
    [​IMG]
    Then first scan (IT8-target calibrated) with auto tone applied to get rid of magenta color cast:
    [​IMG]
    And the second one, Epson's "automatic" scan with autolevels applied in PS:
    [​IMG]
    Really cannot say that the one from color managed process has any way better colors. It has a bit more contrast than epson scan's automatic result but it also has a magenta tint.
    After simple adjustment in PS both looks quite same. The last one looks bit better to my eye.
    I would expeced a lot more difference.
    I got a bit suspicious for whole thing after I realized that same IT8 target in Faust's set is used for Velvia 100, Velvia 100F and Astia 100F films - and those films does not look similar in terms of saturation and colors on the lightbox. So why do they share same target?
     
  4. I'll 2nd that.
    I've scanned many many images without the targets. Then scanned them again with the IT8 targets via Vuescan. The results? Different, yes. Better, no.
    I've had other people look at series of them, blind. And do an A-B comparison. Nobody reliably picked any of them as being better than another.
    I always assumed that I was doing something wrong.
    Lars
     
  5. Did you generate the ICC profile with Vuescan? If so, I agree there's little real benefit. Try a third party program like Scarse or LCMS/LProf and assign them in Photoshop.
     
  6. In digital image processing you have to sometimes stop and weigh whether the effort put into getting that extra 10-20% better image quality is really worth the extra time creating a custom profile.
    I used an it8 that came with my 1997 Agfa Arcus II medium grade flatbed scanner to scan prints to create a custom profile and always ended up using the canned profiles included by the manufacturer because they produced better and more accurate color.
    All those images look almost identical except for the extra dynamic range in some, but that can be fixed with a curve adjust in the scanner software using its Auto setting or canned profile and saved and applied to other scans. Sounds a lot easier.
     
  7. Roger, I created the profiles with lprof and tested different versions, from rought profile to very exact. Nothing that could be seen in the results.
     
  8. The image you submitted as an example is about as plain-vanilla in color content as possible, making it very easy to scan. This is typical of street and landscape shots.
    You can see from the histogram that the R, G and B channels a nearly equal. You could hit the "Auto" button in Photoshop Levels and achieve the same results, which is pretty much what Vuescan and Epsonscan does to automatically adjust the color. There is a slight shift toward blue in the midrange red/blue (i.e., white) balance, as would be expected for a picture in open shade.
    Try shooting something with a dominant color, like closeups of flowers, subjects against a red brick wall, etc. That will test the scanning software more severely, and demonstrate that using a well-crafted profile will come closer to the original slide.
    00Uq01-183417584.jpg
     
  9. I agree with Roger. For whatever reason, you're not getting optimal results. There are many settings in the color profiling software packages. Perhaps something there is set wrong.
    Proper IT8 profiling is so powerful that it can make scans from across scanners (different brands/models) produce pretty much the exact same scan of a slide. For example, the following three images are the same frame of film scanned on 3 different scanners (a mixture of Imacons/Nikons) using properly color-managed workflows. No image editing was done at all on any of the images ; raw scanned images had appropriate scanner profiles applied, then converted to sRGB for web viewing. Download them & overlay them in Photoshop. Your eyes will be hard-pressed to tell any difference. To my eye, the scan (as viewed on a profiled monitor) looks exactly like what I see on the slide (on a 5000K white box). Uncanny, almost.
    [​IMG]
    [​IMG]
    [​IMG]
    The automated results from each scanner would, needless to say, produce different colors from the same slide across several scanners. Man, I just layered these in Photoshop again just to convince myself... blows my mind every time how perfectly the colors match up. From 3 entirely different scanners running 2 different software packages!
    You may notice some variance in the black level... this occurs typically for various reasons, first and foremost being due to optical flare that artificially lightens darker colors adjacent to lighter colors on the IT8 target... long story, don't even want to get into it right now :)
    Rishi
     

Share This Page

1111