Depth Of Feild

Discussion in 'Digital SLR' started by Edward Holt, Feb 27, 2005.

  1. 0m (Paul Mitchum) wrote in 0m:
    Yes - that is the definition of DOF.
    Yes - you assume that you don't change the distance.

    If you do change the distance - that is a useful meassure - but it is
    not according to the definition of DOF.


    /Roland
     
    Roland Karlsson, Feb 28, 2005
    #21
    1. Advertisements

  2. 0m (Paul Mitchum) wrote in
    No - no - no - of any print size. The value 1/1625 has nothing
    to do with the print size. The 8x10 is just a starting point.

    Which make the rest of your reply irrelevant :)




    /Roland
     
    Roland Karlsson, Feb 28, 2005
    #22
    1. Advertisements

  3. 0m (Paul Mitchum) wrote in 0m:
    Nope - DOF is defined to use a CoC of 1/1625 of the picture diagonal.

    You might not like the definition. You might like to measure something
    else. But - that is not DOF according to the definition.

    And - what you proposes cannot be engraved into the lens and is
    therefore in practice rather useless.


    /Roland
     
    Roland Karlsson, Feb 28, 2005
    #23
  4. I have used the DOF markings on the lesns to set the lens
    to the hyper focal distance, i.e. you put the farthest 8
    (if you use F/8) at infinity.


    /Roland
     
    Roland Karlsson, Feb 28, 2005
    #24
  5. Roland Karlsson wrote:
    []
    What does that say about the number of distinctly resolvable pixels which
    can be present in the image?

    Cheers,
    David
     
    David J Taylor, Feb 28, 2005
    #25
  6. Nothing.

    DOF is defined with regard to a viewer. It has nothing to do with
    film resolution and pixels. DOF is used to evaluate the quality
    of a picture. If you print the picture too small or use a to coarse
    film or digital sensor, then DOF loses its meaning. If you make
    a large enough print and the quality is high - then DOF is meaningful.


    /Roland
     
    Roland Karlsson, Feb 28, 2005
    #26
  7. Edward Holt

    Paul Mitchum Guest

    That might be right, but the definition of CoC is that it's the smallest
    practical perceptible angle of human eyesight, at a reasonable reading
    distance. That ends up as a certain fraction of a certain size print
    held at a certain distance. The fraction varies depending on who you ask
    about it, but the fraction is still derived from print size and
    distance-from-face.
    For what?
    I never claimed relevance. :)
     
    Paul Mitchum, Feb 28, 2005
    #27
  8. Edward Holt

    Paul Mitchum Guest

    Yup. :) Try it and see! Find a somewhat fuzzy image, print it out tiny,
    and and then big. Put them next to each other on a table. Compare. One
    will appear sharper.
    Nope. :) The markings on the lens *might* *assume* to use that
    fraction. DoF isn't defined that way, however.
    The definition of DoF is this:

    The area in front of and behind the image subject which appears to be
    sharp.

    This means that since a given print with some bokeh viewed from, say, 3
    feet away will appear to be more sharp than if you view it 3 inches from
    your face, the DoF of that image changes as you move it. Ditto for
    resizing. And if you resize *and* move it, then all hell breaks loose.
    :)
    Ahh! Then you see my point! You don't discard it as incorrect, you
    discard it as impractical.

    What's engraved into the lens is a safe assumption. And that assumption
    is a certain fraction of a print at a certain size held a certain
    distance from the face of the viewer. Like I said. :)

    But the reality outside that assumption is not useless, either... For
    instance, if I were to get an adaptor and use a medium format lens on my
    digital SLR, the lens' DoF indicators would be meaningless (because
    they're probably calibrated to something like a print size 5x that of
    medium format film, which is huge.). But since I know how DoF actually
    works, I can do that math if I need to, and end up with a pretty picture
    of the dandelion in the foreground and the mountain range in the
    distance, all in focus. More or less. :)
     
    Paul Mitchum, Feb 28, 2005
    #28
  9. Edward Holt

    Paul Mitchum Guest

    On a re-read, I see your point.
     
    Paul Mitchum, Feb 28, 2005
    #29
  10. Well, I think it does.

    By saying that anything 1/1800 of the diagonal cannot be separately
    resolved (or whatever the figure is), then it's like saying that anything
    1/1500 of the width, or 1/1000 of the height can't be resolved either.
    See what I'm getting at? Defining DOF with a particular fraction image
    diagonal is implying that there are a particular number of pixels
    resolvable in the image as presented to the viewer. I'm not saying it's
    1500 x 1000 pixels exactly (would need to do the maths), but I'm sure it's
    related.

    Cheers,
    David
     
    David J Taylor, Mar 1, 2005
    #30
  11. 0m (Paul Mitchum) wrote in 0m:
    As I understand it the definition goes something like this.

    1. If you look at a print at normal viewing distance (25 cm),
    then a reasonable sized picture to look at is 8x10.

    2. At normal viewing distance the resolution is
    approx 1/5 mm for normal sight.

    3. This will give a resolution to diagonal of 1/1625.

    4. If you look at larger prints, then you adjust the viewing
    distance so the picture has the same FOV, i.e. the resolution
    to diagonal is still 1/1625.

    5. If you look at smaller prints, then the DoF really does not
    hold. And it is not really important to optimize smaller
    pictures for quality anyway.

    This leads to the conclusion that 1/1625 is the magic number
    for CoC - and then you have a fixed DoF scale on any lens as
    soon as you know the sensor size.



    /Roland
     
    Roland Karlsson, Mar 1, 2005
    #31
  12. Edward Holt

    DoN. Nichols Guest

    [ ... ]
    [ ... ]
    O.K. Working from that diagonal, and an 8x10 print (with
    presumed 1/4" borders, so it is in reality 7.5x9.5"), we come out with a
    diagonal angle of 37.92 degrees, and applying that to the 1625
    resolution, it would require a minimum resolution at the camera of
    1282x999 pixels to support that (about 1.28 MP) -- and probably better
    at double that (as linear dimensions) or four time that as area, so we
    need perhaps 5.1 MP to support an 8x10 at the standard viewing distance.
    (And actually, a bit more, because some of the picture must be cropped
    to fill an 8x10, as the aspect ratio is not the same as the standard
    35mm frame of 24x36.

    But it would also be the same minimum for a larger print viewed
    from the proper distance for that size. So you don't need more than 6MB
    unless you are going to be viewing it too closely. :)

    Of course -- if you are cropping more deeply into an image, you
    need even more resolution.

    Enjoy,
    DoN.
     
    DoN. Nichols, Mar 2, 2005
    #32
  13. (DoN. Nichols) wrote in d.com:
    Nothing wrong did I find.
    Did :)
     
    Roland Karlsson, Mar 2, 2005
    #33
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.