Lens perspective distortion and DSLR crop factor

Discussion in 'Digital Photography' started by pgmacdonald, May 14, 2007.

  1. pgmacdonald

    pgmacdonald Guest

    Hi all,

    I attended the Rocky Mountain School of Photography last week and
    their equipment guru made a claim that is inconsistent with my earlier
    understanding of an issue related to Crop Sensor DSLRs. Maybe someone
    out there could clarify for me.

    I have always been under the understanding that if I took a Canon
    24-70mm f/2.8 zoom and took photos with the following two bodies, the
    resulting images would be exactly the same if we were to ignore pixel
    density and quality differences of the zoom at different focal
    lengths:

    Canon 30D @ 31.25mm focal length, 5 feet from subject
    Canon 5D @ 50mm focal length, 5 feet from subject

    According to the individual at Rocky Mountain, the lens (perspective)
    distortion normally associated with wide-angle shots (traditional 35mm
    film) would be evident on the 30D example above because this
    perspective distortion is inherent to the wide angle lens. My
    understanding was that this was false and that the distortion is a
    result of distance to subject. Could somebody please clarify this
    issue for me?

    Regards,
    Paul
     
    pgmacdonald, May 14, 2007
    #1
    1. Advertisements

  2. pgmacdonald

    Mike Russell Guest

    You're correct - wide angle distortion is a function of distance to subject
    only, and the zoom settings you describe should perfectly cancel out the 1.6
    crop factor, modulo lens distortion and shifting of the nodal point.

    This is a very common misconception, similar to the belief that DOF is less
    for a telephoto than a wide angle.
     
    Mike Russell, May 14, 2007
    #2
    1. Advertisements

  3. wrote:
    : Hi all,

    : I attended the Rocky Mountain School of Photography last week and
    : their equipment guru made a claim that is inconsistent with my earlier
    : understanding of an issue related to Crop Sensor DSLRs. Maybe someone
    : out there could clarify for me.

    : I have always been under the understanding that if I took a Canon
    : 24-70mm f/2.8 zoom and took photos with the following two bodies, the
    : resulting images would be exactly the same if we were to ignore pixel
    : density and quality differences of the zoom at different focal
    : lengths:

    : Canon 30D @ 31.25mm focal length, 5 feet from subject
    : Canon 5D @ 50mm focal length, 5 feet from subject

    : According to the individual at Rocky Mountain, the lens (perspective)
    : distortion normally associated with wide-angle shots (traditional 35mm
    : film) would be evident on the 30D example above because this
    : perspective distortion is inherent to the wide angle lens. My
    : understanding was that this was false and that the distortion is a
    : result of distance to subject. Could somebody please clarify this
    : issue for me?

    : Regards,
    : Paul

    The answer is kinda maybe.

    Yes the field of view will be the same for both and at these values the
    spherical distortion (most noticable at extra wide angle or fisheye)
    differences will be minimal. So the two images will be close enough to
    identical as to be not noticed. But as you get further toward any extreme,
    lens based distortions can become more problematic. For example if it was
    only field of view, a fisheye photo of a closeup human face that shows the
    dramatic exagerating of physical features that is then cropped to an image
    with the same field of view as a longer lens image would be identical. But
    the simple cropping will not correct the exageration of features. So at
    extremes there will be a difference. But when the lens lengths are fairly
    close, the crop of one to match the FOV of another will be very close to
    eachother.

    So yes it is probable that the two images you propose will be visually
    similar enough that you wouldn't likely notice the difference. Of course
    each combination of camera and specific lens will have its own set of
    plusses and minuses and so even using the same body, several different
    lenses at the same focal length could have some slight variation in the
    image distortion from trying to represent a spherical view on a flat
    medium.

    JMHO

    Randy

    ==========
    Randy Berbaum
    Champaign, IL
     
    Randy Berbaum, May 14, 2007
    #3
  4. pgmacdonald

    Scott W Guest

    You are right, he is wrong.

    There might be some small differences in the amount of pincushion
    distortion, but this is not the
    same as perspective.

    Scott



    Scott
     
    Scott W, May 14, 2007
    #4
  5. Ah, the problem may be that there are TWO definitions of distortion.
    To an optical engineer and lens designer, distortion means that the
    field angle to any point in the image is not exactly the same as the
    field angle of the input point. This kind of distortion IS reduced by
    reducing the format used by a lens, and hence by the cropping done by
    digicams. That is, the distortion increases with distance from the
    optical axis, which intersects the image plane at the center of the
    image for most cameras. Thus cropping does not use the wide field
    angles and does reduce this form of distortion.

    Perspective distortion is another issue. This is the APPARENT
    distortion of perspective caused by viewing an image from a different
    effective vantage point than what it was taken at. This is especially
    due to the fact that wide angle prints would require viewing from a
    distance closer to the print than our eyes can focus at. This type of
    distortion is ONLY a function of the taking and viewing distances, and
    has no dependence on focal length of lens.
     
    Don Stauffer in Minnesota, May 14, 2007
    #5
  6. That would be pretty much always true if both cameras were using the
    same lens, but the 30D was using less of the image circle than the 5D.

    However, that's not the case here. The two cameras are using different
    focal length lenses in order to match the field of view, and so the field
    angles at the corner of the image are exactly the same for both of the
    proposed cases. So either camera could see slightly more distortion
    than the other, depending on the distortion performance of the two lens
    designs.

    Dave
     
    Dave Martindale, May 15, 2007
    #6
  7. You're right, he's wrong. The wide-angle perspective distortion is only
    caused by a difference in field angle between taking and viewing the
    image. In this case, you are carefully matching the field angles
    between the two cameras, so you will get exactly the same perspective in
    images from both. If you make prints of the same size and view them
    from the same distance, you will get the same amount of perspective
    distortion from both.

    On the other hand, the image shot with the 30D will have slightly
    greater depth of field if shot at the same aperture as the 5D one.

    Dave
     
    Dave Martindale, May 15, 2007
    #7
  8. pgmacdonald

    Colin_D Guest

    If both - or any - cameras are at the same distance from the subject,
    the perspective drawing is the same for all, regardless of focal length,
    wide angle or not, and regardless of image or film/sensor size. Any
    part of the scene common to both images will have the same drawing,
    which is determined solely and only by the camera position, and nothing
    else.

    It is commonly stated that certain lenses produce optimum perspective
    for portrait shots, typically 85mm on a 35mm camera. This is fallacial;
    the reality is the desired perspective is obtained from the shooting
    distance, and that distance needs an 85mm lens to fill the frame. If
    you shot with a shorter lens from the same viewpoint, the perspective
    will be identical - but the image will be smaller. If you go closer to
    fill the frame, then the perspective will change, and this gives rise to
    the commonly but wrongly held belief that the lens controls perspective.
    It does not; distance from subject controls perspective.

    Minor distortions with particular lenses are not relevant to your question.

    Colin D.
     
    Colin_D, May 15, 2007
    #8
  9. pgmacdonald

    B. Dover Guest

    Actually the distortion you see in a wide angle lens image does not
    exist. It just looks like there is a distorted wide angle look and if
    you know how to look it wouldn't look like there is wide angle look
    to look at.

    Ben
     
    B. Dover, May 20, 2007
    #9
  10. Yes, with two qualifications:

    1) The lens might have different rectilinear distortion characteristics
    at different zoom settings. So, for example, it might have some barrel
    distortion at 31.25mm that it does not have at 50mm.

    2) To get the same depth of field, you would have to adjust the aperture
    to compensate for the difference in focal length. Since subject-camera
    distance is the same, you would need the absolute size of the aperture to be
    the same. So, for example, if the 31.25mm setting was at f/4, then the
    aperture would be 7.8125mm in diameter. To obtain the same diameter at a
    50mm setting, you would need f/6.4.
    Well, now you know one instructor not to believe.
     
    Andrew Koenig, May 20, 2007
    #10
  11. As others have already pointed out - he's wrong.

    The only reason I chime in is to point a link to a pair of
    photos that I took to dispel this myth:
    http://hannemyr.com/photo/crop.html#per

    It is comparing a 135mm on a "full frame" with a 28mm on a 4.8x
    crop sensor.
     
    Gisle Hannemyr, May 21, 2007
    #11
    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.