perspective w/ 35mm lenses?

Discussion in 'Digital Photography' started by PrincePete01, Jul 16, 2004.

  1. PrincePete01

    Nostrobino Guest

    In a NEWSGROUP? Why?
    Nostrobino, Aug 5, 2004
    1. Advertisements

  2. PrincePete01

    mcgyverjones Guest

    I've been trying to keep out of this but have finally been dragged in. You
    mentioned earlier that you have not used a view camera. This experience
    would be valuble in demonstrating how a long lens can "duplicate the
    perspective" of a wide angle.
    First, define long and wide lens. This is the heart of the flaw in your
    argument. Lets use 35mm format as a standard.

    Imagine attatching an 150mm lens to an 8X10 view camera.
    Mount a 35mm back to the camera and take a picture.
    Looks like what you'd expect a 150mm shot to look like in 35mm.
    Now, without moving the camera, remove the 35mm back and insert a sheet of
    8X10 film.
    Well look at that, wide angle! The perspective has not changed.

    If you say it has, you've redefined the word "perspective" to create your
    point. Perhaps you are confusing it with composition?
    mcgyverjones, Aug 5, 2004
    1. Advertisements

  3. PrincePete01

    Jeremy Nixon Guest

    But why? More to the point, *how*? I mean, if you put your eye right where
    the lens is, and look over at the disc, you'll see it in the shape of an
    ellipse, because you're looking at it from an angle. (You will know it's
    a circle, because your brain will fill that in, but you will see an ellipse.)

    So how does a camera lens magically see a circle? The fact that the image
    plane is parallel to the disc's surface can't change the laws of physics and
    make the lens see the disc from somewhere other than where it is, so why a
    circle? How a circle?

    The answer, of course, is that the lens has distorted the image in such a
    way that exactly cancels out the angles involved (assuming a good lens, of
    course) in order to create a flat image. The projection used to create the
    flat image is the opposite of the distortion created by the lens seeing
    different parts of the scene from different angles.
    That, I think, is a major cause of your misunderstanding.

    Think of it another way: imagine all the rays of light entering your lens,
    and then reverse them, trace them back out to where they came from. The
    rays were coming in from all directions, within the field of view, right?
    All the various possible angles the lens can see contributed. Now, what
    geometrical shape do you get when you take lines in every possible
    direction? A sphere! Your field of view, then is a section of that
    sphere. Your picture, then, is created from a section of that sphere,
    which is not flat, and therefore must be flattened.
    Jeremy Nixon, Aug 5, 2004
  4. If you aim the camera at the disc, so it's on the optical axis of the
    lens, the camera will see an ellipse too. No difference in appearance.
    No "distortion".
    The camera "sees" a circle only if the camera is re-aimed so the film
    plane is parallel to the disc surface. In that case, and in that case
    only, the imaging geometry says that the image will be round. But this
    is not distorted. If you take the image and print it on a piece of
    paper, and then view the print from an eyepoint that makes the print
    cover the same visual angle as the original scene, and places your eye
    on the centre axis of the print, this completely recreates the original
    scene geometry. If you look over at the disc, you'll see an ellipse -
    the same shape and size as in the original subject.
    But this is not distortion by anyone else's definition. This is just
    plain geometric imaging. The image is *not* distorted; it's an accurate
    mapping of the 3D world onto a flat plane through a point.
    No, you get a sphere only if you trace each ray out to a fixed distance,
    rather than the actual distance it came from. The original image was
    *not* a sphere. You *can* print the image onto a spherical surface, or
    project it onto a spherical screen, and then when it's viewed from the
    appropriate point it will give a correct view of the scene. Such
    spherical prints/screens are useful in covering a wide field of view.

    But a flat print is normally a lot more practical, and when the flat
    print is mounted flat and viewed from the correct position, it
    reproduces all of the angular relationships in the scene just as
    correctly as the spherical image. In addition, the flat print also
    looks OK from many non-optimal viewpoints, while the spherical print
    does not.
    The sphere exists only in your imagination, unless you make a spherical
    print or project onto a spherical screen.

    Dave Martindale, Aug 5, 2004
  5. Yup. That's my conclusion after reading many of his messages.

    After all, I can say "you will never have the same perspective on image
    resampling algorithms that I have". That statement may be completely
    true, and yet has nothing to do with the current discussion because I'm
    using a different meaning of "perspective".

    Nostrobino is using "perspective" to mean something related to what's in
    the image, or perhaps how close your eye needs to be to the print for
    things to look correct. It is *not* the usual photographic meaning of
    perspective, which is concerned with how apparent size changes with
    distance, and which objects block the view of what parts of other

    Perhaps we should just say that "Changing field of view changes the
    Nostrobino Perspective of the image, but does not change the
    photographic perspective."

    Dave Martindale, Aug 5, 2004
  6. PrincePete01

    Jeremy Nixon Guest

    That depends on your definition of "accurate", no? I mean, there is no
    vantage point, no angle, which will make my eyes see an object the way it
    looks at the corner of a wide-angle photograph. It just doesn't happen.
    Stuff never looks like that, with the angles all wrong and peoples' heads
    stretched out. So how is that accurate?
    Well, of course the sphere is imaginary; I'm trying to model the way a lens
    sees the world, with light rays coming in from all possible directions and
    then being mapped onto a flat surface. That looks like a spherical projection
    to me. It's not the way a human sees the world, since we don't see an entire
    scene in one go, like a camera does, and thus we don't need to flatten the
    image like a camera does.

    It seems to me that the only way to map the 3D world viewed from a single
    point in all directions, without distorting the image in some way, would be
    spherical. I'm not saying that the accepted meaning of a "distortion-free"
    lens should be abandoned, but just that the term doesn't apply to this
    particular topic. Since such a lens is made to produce an image that
    looks right to us, would "corrected" be a more comfortable term?
    Jeremy Nixon, Aug 5, 2004
  7. PrincePete01

    MarkH Guest

    Ummm, you are completely wrong here! You are confusing camera POSITION
    with camera LOCATION. Camera location is unchanged when you alter the
    direction it is pointing. Camera position is definitely changed when you
    alter the orientation of the camera.

    You can reposition a camera by slightly altering the direction it faces,
    you don’t need to move it to a different location.
    MarkH, Aug 5, 2004
  8. This is of course a quite gratuitous insult to many posters who have
    over many years in this and other NGs demonstrated not only vastly
    greater knowledge of optics (and scientific definitions) than yourself,
    but also a huge willingness to give their time to pass their knowledge
    on to others less experienced. I thus have to take back my view a few
    minutes ago that you showed no evidence of being a troll; making ad
    hominem attacks against those more intelligent and better informed than
    yourself when you have failed to win an argument is of course very
    troll-like behaviour.

    At least it's good to see that you include yourself in the category of
    those who do not have a mind worth engaging. A touch of self-awareness,
    at least.
    David Littlewood, Aug 5, 2004
  9. This is quite trivially simple. Take a picture with a 150mm lens on a
    5x4 camera and one with a 50mm lens on a 35mm camera. Or 90mm and 28mm,
    or any other combination in approximately 3:1 ratio. Make appropriate
    allowances for differences in depth of field between the two lenses.

    The appearance, field of view and perspective of the two negatives or
    transparencies will all be identical; the 5x4 picture will of course be
    a lot larger - but that was not the point at issue. Your key assertion
    is that the focal length of the lens used affects "perspective". The
    fact is that in this test the focal length of the lens made no
    difference to the picture, except in its size. Even you have not, so far
    as I recall, asserted that "perspective" includes the size of the
    negative/transparency produced. There is therefore no way this result
    can be consistent with your key assertion.

    If you respond by saying "ah yes, but the 150mm lens is a "normal" lens
    on a 5x4 camera, and I really mean the focal length of the lens in
    relation to the size of the film (sensor)" then we will see you
    blatantly shifting your ground to avoid admitting defeat.

    If you would only see that you are using the word "perspective" in a way
    the rest of us do not, then there would be no significant dispute; in
    fact you have argued your case very cogently (and BTW I have seen little
    so far to suggest you are a troll) apart from this misunderstanding of
    vocabulary on your part.. Unfortunately, you appear to have the delusion
    that "perspective" means "all aspects of the appearance of the picture".
    Almost all photographers do not agree with you - which should have
    become apparent to you from the vastly over-blown debate.

    Where a person finds that (almost) no-one else in his field agrees with
    his definition of a word, then I think it behoves him to ask himself if
    he has made an error rather than assuming that all the world is out of

    However, if you still wish to be in a minority of one, there is nothing
    the rest of us can do to coerce you, and there seems no point in
    prolonging the agony.
    David Littlewood, Aug 5, 2004
  10. PrincePete01

    Nostrobino Guest

    Are we talking about two different things here?

    In the case of the discs, it is as Dave described. There is no distortion.
    Circles are represented as circles, PROVIDED the lens axis is perpendicular
    to the surface on which they are placed. A disc appears as an ellipse if you
    "look over at" it as you said, but as Dave has pointed out if the camera
    lens is tilted so that it "looks over at" it it will see the same ellipse.

    One problem with making this sort of eye/camera lens comparison is that the
    camera lens delivers a sharp image over a relatively wide flat surface
    (especially the extreme wide-angle used to make that figure), while the lens
    of the eye cannot. Thus if you place your eye at a distance from the discs
    equivalent to that of the camera lens, and continue to look straight at
    (i.e., line of sight perpendicular to) the surface on which they rest, you
    just cannot see the discs near the edge very well defined. When you "look
    over at" them to see them better, you have obviously pointed the lens axis
    of your eye in a different direction; therefore you see them just as the
    camera would see them if ITS lens axis were pointed in that direction.

    But now you're asking about "people's heads stretched out" (i.e., near the
    edges or corners of a wide-angle photo), and that's an entirely different
    thing because these are SOLID objects, like the ping-pong balls in the
    figure--which also appear "stretched out" radially in the same way.

    This is perspective (specifically, wide-angle perspective). You don't see it
    in the discs, because they are essentially two-dimensional things flat to
    the lens, and no two-dimensional thing has any perspective. (As a
    clarification: if you TILT a two-dimensional thing it becomes
    three-dimensional as far as the eye or camera lens is concerned, and then of
    course it does have perspective.)

    The "distortion" you see in people's heads (or those ping-pong balls) is not
    really distortion at all; it seems so only because your point of view is
    "wrong"--i.e., radically different from that of the camera that took the
    photo. If you move your eye closer and closer to the center of the photo,
    then you CAN "look over at" the apparently distorted heads (or ping-pong
    balls) and you will see the distortion gradually disappear as you do so. In
    the case of an extreme wide-angle shot this means moving your eye VERY close
    to the photo. In the case of that figure it will probably be impossible to
    get close enough because the picture is so small.

    No, it's not a spherical projection. There isn't any sphere, only objects in
    three-dimensional space.

    It's not the way a human sees the world, since we don't see an entire
    Actually it gets a lot more complicated than that, but I dread getting into
    it. Human vision (or any animal's) is quite different from a camera's, but
    the differences do not have to do with perspective.

    The most appropriate terms have already been given here, several times: The
    camera lens makes a two-dimensional representation of a three-dimensional
    world. That representation is (within reasonable tolerances)
    Nostrobino, Aug 5, 2004
  11. PrincePete01

    Nostrobino Guest

    I don't really doubt that, which is why I specified "conventional camera." I
    am aware that the movements of a view camera allow considerable manipulation
    of the image.

    With a conventional camera (i.e., one in which lens axis and film plane are
    in a fixed relationship), you cannot duplicate the perspective of a
    wide-angle lens with a long lens. That's what I'm saying.

    Sure it has.

    The perspective of the smaller part (the part first contained in the 35mm
    frame) has not changed, of course. But that is no longer the picture. A
    small part of any thing "X" is not X.

    Continue your experiment in this way: From one corner of your 8x10 negative
    or chrome, take a 24x36mm part. Any corner will do as long as it shows the
    actual perspective of some object or objects (not blank sky, for example).
    Enlarge it to whatever size suits you. Now put the 35mm back on the camera
    again and shoot the same object(s) from the corner piece, doing so without
    using any of the view camera movements--just a straight shot. Enlarge to
    suit as before.

    Does it look the same? No. Is the perspective the same? No. You cannot
    duplicate wide-angle perspective with a long lens, at least not without
    considerable manipulation of the image.

    Through manipulation of the image of course you can do almost anything--with
    my computer I can do zillions of things with an image that you can't do with
    your view camera, but none of these things really demonstrate anything about
    photographic perspective.

    No, perspective.
    Nostrobino, Aug 5, 2004
  12. PrincePete01

    Nostrobino Guest

    On the contrary, this has already been covered at some length.

    As explained previously, all mention of focal lengths has been based on
    "35mm equivalence," which is fairly standard practice in discussion of
    digital cameras. Except for dSLRs (for obvious reasons), digital camera
    users commonly use 35mm equivalents when referencing focal lengths.

    Here, rather than go over it all again I'll just paste in part of my remarks
    from yesterday:
    These things are all necessarily relative. With most digital cameras today,
    there are so many sensor sizes that actual focal lengths are essentially
    meaningless and are generally not even mentioned. If e.g. someone says his
    Minolta S414 has a 35-140mm lens, what he really means is that that's what
    it is in 35mm camera equivalence. I have one and I don't even know offhand
    what its actual focal length range is. I'd have to look. Similarly, I know
    my X-series Minoltas have "37-111mm" lenses and my F300s have "38-114mm"
    lenses. Again, what the actual focal length ranges are I don't remember and
    don't really have any reason to remember. My 7i and 7Hi cameras have
    identical lenses, and in this case I do happen to remember that the lens is
    7.2-50.8mm, but the important thing is that its 35mm equivalence is 28-200mm
    (and the zoom control ring is so marked, actually). The same lens exactly on
    my Dimage 5 is a "35-250mm" lens because of that camera's slightly smaller

    All this makes sense with digital cameras since most people are familiar
    with 35mm camera focal lengths, and would be hopelessly confused by true
    f.l. numbers if those were all they were given.

    However, all this has no direct bearing on wide-angle perspective, which is
    really a function of field of view. It's convenient to use focal lengths
    instead of angular measurement, simply because f.l. is the measure that
    people are most familiar with. More people will understand what you mean by
    "28mm lens" than would if you said "75-degree lens," and so on.
    Nostrobino, Aug 5, 2004
  13. PrincePete01

    Nostrobino Guest

    [ . . . ]
    This evidently is a very important thing to you, to believe what you think
    everyone else believes. It isn't important to me. You need to be part of
    "the rest of us." I don't.

    The important thing to me is to get at the truth. Often this requires
    relying on the help of others, which is something I always appreciate when
    it's given. Sometimes it requires that I think for myself, which is likely
    to be more work, but I am willing to do that when it seems necessary.

    Far from "[wishing] to be in a minority of one," my preference would be that
    everyone understood what I am saying and either a) agreed with me, or b)
    corrected me by pointing out the errors in my thinking. But such correction
    has to hold up to rigorous examination. Merely reciting nonsense you have
    read somewhere doesn't do it, particularly as in this case when I've already
    read that nonsense for myself, years ago, and concluded it was nonsense.

    Neither does specious reasoning impress me much (unless it's at least very
    imaginative), and certainly I am not dismayed by not being part of "the rest
    of us." If that is so important to you, then rejoice.
    Nostrobino, Aug 5, 2004
  14. PrincePete01

    Nostrobino Guest

    [ . . . ]

    Actually I thought you had mostly all gone, David. You seemed to have
    disappeared for the most part (other than one or two I plonked for sheer
    worthlessness). In fact your own comments clearly indicated you were having
    nothing more to do with this discussion. You had stopped replying to me, and
    these were what I very reasonably took to be your final comments on the
    matter, to Brian, two days ago:

    I think you are flogging a dead horse. This guy believes "perspective"
    means "everything which affects the appearance of the picture". He has
    been in a minority of one for several days and just keeps parroting the
    same stuff. I suggest you may be wasting your time arguing with him.

    After that, really, your present petulance and complaint of wounded feelings
    seems surprising.

    My remarks to BillyJoeJimBob were intended as, and I assume taken by him as,
    an expression of appreciation and nothing more. It is certainly true that my
    experience with others here ("the rest of us" in your words) was not
    encouraging in view of my original expectations of open-mindedness,
    intellectual honesty and reasonable inquiry.
    Nostrobino, Aug 5, 2004
  15. Look at my previous paragraph, which you snipped out. It says that if
    you print a wide-angle photograph on a flat surface, then look at that
    print from an eyepoint that is on the centre axis of the print at a
    distance that matches the visual angle of the print to the visual angle
    occupied by the original scene, you see everything with *exactly* the
    same shape and angular size and relative position it originally had.
    That is the sense in which it is accurate.

    Now, if you look at the print from somewhere off-centre or at the wrong
    distance, it will be distorted with respect to the original - but that's
    your fault. Your "undistorted" spherical image has exactly the same
    problem: there is one and only one eyepoint for which the image appears
    correct. The flat print is completely undistorted from one eye
    position, which is the best you can expect for any shape of print.
    It is a spherical projection. But there is no sphere involved
    It may seem that way to you, but please explain how the spherical
    mapping is less distorted than the mapping to a flat surface. What is
    your measure of "distortion"? Suppose you have two images of the same
    scene, one onto a flat surface and one onto a (part of) a sphere. Both
    are done with perfect geometric accuracy, so there is no lens
    distortion. The one on the sphere preserves angular distances, so a
    constant angle in the subject is a constant distance in the image, but
    it maps straight lines into curves. The one on the flat surface does
    not preserve angular distance, but it does map straight lines to
    straight lines. Why is one of these called "distorted" while the other
    is not?
    "Distortion-free" is a specific term long used to describe lenses, and
    since we are discussing lenses any attempt to redefine it will create
    confusion. "Corrected" is also already well-defined with respect to
    lenses; it talks about how well some particular lens problem has been
    removed by a particular design. (The problem might be spherical
    aberration or geometric distortion; it doesn't matter).

    If you want a good term to describe the fact that all lenses (not just
    wide angle ones) produce images that are incorrect when viewed from the
    wrong eye position, you should invent a new term for it that doesn't
    conflict with established lens optics terms. But since all lenses will
    have this "problem" for any shape of image plane, not just flat or
    spherical, how useful will the term be?

    Dave Martindale, Aug 5, 2004
  16. PrincePete01

    Frank ess Guest

    Not to mention your pretentious pomposity.
    Frank ess, Aug 5, 2004
  17. PrincePete01

    Nostrobino Guest

    Perspective is inseparable from field of view, is what I've been saying.

    Yes, there are many different definitions for "perspective."

    Have you never heard anyone speak of wide-angle perspective? Is my mention
    of it absolutely the first, in your experience?

    Do you claim--as some of "the rest of us" here have claimed--that a
    wide-angle photo looks just like a long-lens photo? (I am not one of "the
    rest of us," you understand.) That is, there's no such thing as a wide-angle
    look or a telephoto look?

    In your view, perspective is NOT something related to what's in the image?

    And how angles appear to change, and how parallel lines appear to converge
    at greater distance. All of that is part of what I mean by perspective.

    Well, you could just say that "Changing the field of view changes the
    perspective, but not The Rest of Us Who Read Somewhere That What We See
    Isn't Really So perspective.
    Nostrobino, Aug 5, 2004
  18. PrincePete01

    DSphotog Guest

    He must be related to George Preddy.
    DSphotog, Aug 5, 2004
  19. PrincePete01

    Nostrobino Guest

    How is it a spherical projection?

    I would take "spherical projection" to mean something like a Mercator
    projection, which obviously this is not.
    Nostrobino, Aug 5, 2004
  20. I'm being generous. It might be called a spherical projection in the
    sense that, if you represent the position of the object relative to the
    camera in spherical polar coordinates (two angles and a distance), then
    the position of the image point corresponding to the object is
    determined by the two angles (but not the distance).

    But you're right, it's not really a projection of a sphere onto a flat
    surface, and not a spherical projection in the cartography sense.

    Dave Martindale, Aug 5, 2004
    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.