Tolerance for Scanned and Ultrachrome Printed Transparencies

Discussion in 'Large Format' started by neil_poulsen|1, Jul 22, 2003.

  1. It's my experience that for transparencies to visually appear properly
    exposed, they need to be right on, probably within a quarter stop (or
    less?) of optimum.

    To my question, how close must one be in exposing a 4x5 transparency
    to be scanned to obtain an excellent photograph from a good printer
    like an Epson ultrachrome, a Lightjet, or Lambda? For example, can
    one be within a half-stop of optimum visually and still expect to
    obtain an excellent photograph from one of these printers. Can
    Photoshop be used to correct the fact that the original 4x5 transparency
    was a little off on the exposure? How much tolerance is there on the
    original transparency when scanning to obtain a digital print?
  2. Neil
    I share your experience as far as optimum exposure for transparencies is concerned, especially where slow films such as Velvia are concerned.
    However it is my experience that the ideal transparency for scanning should if anything be slightly underexposed. If there is any loss of information in the highlights nothing will correct this, whereas it is possible to extract a little more detail from the shadows if necessary using Photoshop.
    I have always taken varying exposures of all my work and certainly since I’ve been scanning I have to say I’m glad of it. A transparency that is bang on will look great on the lightbox but is not always the best exposure for scanning
    I would put the range of tolerance for a film such as Velvia at around minus1/2 stop to spot on but obviously this is dependent on the film used.
  3. I think it depends on the scanner you intend to use. Drum scanning or using one of the high end Imacons will allow you to dredge plenty of shadow detail out of the scan.

    Using an inferior scanner will be less successful in the shadows. For this reason, contrary to Keith's practice, I tend to err slightly towards over-exposure -- say by up to half a stop -- if the scene allows it without jeopardising highlights. Obviously if the scene has a wide exposure range, it can't be done.

    Furthermore, this approach is only useful if the transparency is only used as a step-up into digital print production. Once in Photoshop, some contrast and brightness adjustments are necessary to restore the final exposure, but this can be done selectively, preserving the shadow detail you have gained.
  4. Keith, I have to disagree with you (sorry on my clumsy english as 1st).
    Overexposed areas are lost, that is true. But most scanners can't "see" (or reproduce) that dark areas what slide may have. In my experience (some 25 years of photography and about 10 of scanning in "meantime") the best original for most of scanners will be the one without complete tonal range. Sure, it will need to have "the black" and "the white" point, but it will be the best if they are "placed" few percent under the true ends. Photoshop can "create" pure black and pure white from, let say 2-98% tonal range, but if scanner couldn't see the 99.5% (deep, deep shadow) no software will get it back.
    Please, don't take my words as a flame-call, I'm just sharing my experience... have worked on several top-drum scanners as well as on the best flatbed scanners, and I had to prepare scans for many kind of outputs.
    Today, I'm scanning most of my 4x5" slides on Linotype-Hell (Heidelberg) Topaz, but sometimes I have to go back on slow drum scanner just because that CCD can not recognize shades in the deepest density of my slides. Those are mostly slides from nature where I can't control light that much as I can in studio...
  5. if it looks good on the light table it will be good to scan, too. and as said above overexposed aereas are not to correct, not in a scan and not in a print, cause there is nomre information in the picture. in the low zones sometimes its amazing how much can read a drumscanner in this regions,- but ofcourse its better and easier if the slide is correct exposed. whats correct? + 1/3 til - 2/3 for velvia ( exposed as asa 32 ),- but it depends clearly on the subject. if you have very dark zones where you need details,- overexpose 1 step and pull 1/2 step. than the lights will not be burnd out and the slides will become softer. the ( nearly not visible with 1/2 step pull ) color shift will not mean much in the scan, this you can correct easily.
  6. Leigh and Janko, I accept your point that many scanners have difficulty with shadow detail, but given the choice of an underexposed or overexposed transparency I would still generally choose the underexposed shot.
    Much depends on the film used and to a certain extent personal preference. In an ideal world we all strive to capture all the detail, but if I really have to choose, I find a loss of detail in the shadows gives a more “natural” look than blown highlights.
  7. Keith, when you put it that way "the choice of an underexposed or overexposed transparency" I certainly have to agree with you. Underexposed shots are "better" originals for scanning process and they look more natural, but what I wanted to say is that neither is perfect. If it is possible, I will always try to create a little, but just a little bit "flat" slide. If the purpose is scanning and publishing/printing.
    Of course, for cibachrome or others similar methods I will try to make slide with both - black and white point (if exist in nature or if I want them in my shot:))
  8. if I really have to choose, I find a loss of detail in the shadows gives a more “natural” look than blown highlights
    Keith I absolutely agree. The technique I described can only be used when there is enough highlight headroom. I find that using grad ND filters in the field can sometimes yield a scene with only 3 or 4 stops of variation across the main scene. In scenarios like this, there is a range of safe exposures, and with an eye on a digital destination, I slightly expose without risking loss of highlight to give the scanner better odds.
    There is an interesting parallel in the digital world, described here.

Share This Page