Canon PowerShots not fully manual?

Discussion in 'Digital Photography' started by Dogger the Filmgoblin, Jul 20, 2004.

  1. Sorry ... but whether or not you're right on the larger issue, you're
    dead wrong on this point. With all due respect, you don't seem to have
    full understanding of what a lens does. Lenses cause light rays to
    converge, ultimately intersecting at a single point. Normally the film
    plane or CCD is placed somewhere beyond this point of convergence,
    where the light rays have inverted and started to diverge again,
    making the image upside down and backwards. If you take the lens out
    of its assembly where this distance is calibrated, it is quite
    possible if held at the correct focal distance from the surface, to
    make ALL of the light from the lens (including the sun if it is part
    of the image) converge upon a single point. Take a look at a diagram
    of a lens's operation sometime.

    DB.
     
    Dogger the Filmgoblin, Jul 27, 2004
    #21
    1. Advertisements

  2. SNIP
    Beware of foot in mouth!
    Only if the image source is a point! The smallest focused image of the
    sun is a reduced size version of the sun disc. The size of the
    projected disc vs the original 'disc' diameter is proportional to the
    focal length divided by the distance to the sun.

    Bart
     
    Bart van der Wolf, Jul 28, 2004
    #22
    1. Advertisements

  3. Been there, done that, got the Master's degree in applied optics,
    but no T-shirt. :(

    Perhaps you should take a lens out and try and focus the light coming
    from the light bulb in a table lamp to a point. You can't. Trying to
    focus the light from the sun to a point is just as impossible. You
    will be able to focus it to a fairly small area, but you cannot focus
    the extended source called the sun into a single point.

    For that matter, you can't focus the light from a distant star to a
    single point, nor can you focus light from a laser to a single point.
    Those two cases will focus to much smaller areas than the sun will,
    but neither will be a point. If you take diffaction into account,
    pretty much nothing will focus to a single point.

    BJJB
     
    BillyJoeJimBob, Jul 28, 2004
    #23
  4. At the nodal point of the lens all of the lines intersect. If they
    didn't, the image could not reverse itself upside-down and backwards.

    DB.
     
    Dogger the Filmgoblin, Jul 28, 2004
    #24
  5. Does light curve? (we're not talking about Einsteinian distances here,
    either). Since light travels in a straight line unless deflected, once
    it has passed through the lens, having been bent inward toward
    creating a smaller and smaller image the further you get from the
    lens, there MUST be a distance from the lens at which all of the light
    rays intersect. This is called the nodal point. The only thing that
    will prevent this intersection is imperfections in the symmetry of the
    lens, which are not enough to make much of a perceptible difference.
    Now, in actual fact it may be impossible for a human being to hold a
    lens at the perfect distance to place the nodal point on a surface,
    but you can get damn close.

    I don't know where you folks are getting your ideas of how light rays
    can converge infinitely without intersecting ... that is just not
    possible. The only other explanation for your viewpoints is that there
    is some misunderstanding in the terms that we are using.

    DB.
     
    Dogger the Filmgoblin, Jul 28, 2004
    #25
  6. Also: remember that we're talking about a magnifying lens here
    primarily, which is how this topic got started ... it was about
    comparing the time the light through a magnifying lens focused to a
    point takes to heat something up compared to the light going through a
    camera lens.
     
    Dogger the Filmgoblin, Jul 28, 2004
    #26
  7. No, with all due respect, you have no idea what you're talking about.
    Lenses cause light rays *from a single point in the subject* to
    converge to a small region, nearly a point, in the image if the image is
    in focus. But light from a *different* point in the subject converges
    to a *different* point in the image plane. The ratio of the size of the
    image to the size of the subject depends on focal length and distances
    between lens, subject, and image, and can be calculated by the usual
    lens formulas. However, any subject that is not a point source will
    produce an image that is not a point.

    The image *is* located in the plane where the rays from a source point
    all converge - that's what gives a sharp image. The image is inverted
    because of the ray path through the lens, not because the focal surface
    is located somewhere other than the plane of best focus. And there is
    absolutely no distance you can hold a lens from a surface where it will
    converge all the light to a single point.

    Of course, all of the above referrs to aberration-free lenses and
    ignores diffraction. A real lens will focus light from a single source
    point into an Airy disc pattern, not a point, due to diffraction. The
    presence of lens aberrations will spread the light out further yet.

    I can't explain how you have such a wrong understanding of how optics
    works. You must have glanced at a few ray drawings without really
    paying attention to the text, and come to a variety of incorrect
    conclusions. I suggest that you (a) take any real lens and try to focus
    all light from a scene into a single point, or (b) spend a little more
    time with some basic optics book.

    Dave
     
    Dave Martindale, Jul 28, 2004
    #27
  8. This is wrong on several counts. When you look at a lens ray diagram,
    it is showing rays of light from a single source point. These all
    intersect *at the focal plane*, which is a considerable distance from
    both the principal planes of the lens and the nodal points.

    The nodal point is defined as the location around which the lens can be
    rotated without causing the image to shift.

    The reason the image is inverted is simply that when the subject moves
    in one direction, the image has to move in the other direction because
    the light that makes up the image travels through a hole that's fixed in
    place. Even pinholes give inverted images - this behaviour isn't
    restricted to lenses.

    Dave
     
    Dave Martindale, Jul 28, 2004
    #28
  9. Yes, it does. Look up diffraction sometime. This is why light from a
    star (a point source for most intents) or coherent light from a laser
    can at best be focused to an Airy disc pattern, not a single point.
    Ignoring diffraction and lens aberrations and sticking to straight
    geometric optics, what you say is true for all of the light from *one
    point* in the subject. But the Sun is not a single-point source, so its
    light is focused to an infinite number of points in the focal plane.
    The image of the sun is simply a reduced-size copy of the source,
    complete with all its details. If you put a suitable sensor in the
    focal plane, you can see sunspots and solar flares. THERE IS NO WAY TO
    FOCUS ALL OF THE LIGHT FROM THE SUN TO A SINGLE POINT.
    That's wrong too. The rays converge at the focus, which is far from the
    nodal point. Even with the most precise positioning, the smallest image
    you can obtain of the sun is still a definite-sized disc whose diameter
    is easily calculated given the focal length of the lens.
    I don't know where you're getting your ideas about optics, but just
    about everything you have written is in fact wrong.

    A single point in the subject sends out light in all directions. The
    path of this light can be drawn as "rays". Some fraction of these rays
    enter the lens, and converge to a point in the focal plane. But at the
    same time, some *other* point on the subject is sending out light rays
    too, and all those rays converge to a *different* point in the image.

    Just look at the chief ray in any ray diagram. There is one ray whose
    direction is *unchanged* by the lens. Whatever angle with which the
    ray strikes the middle of the front principal plane, it emerges from the
    rear principal plane *at the same angle*. The intersection of the chief
    ray with the focal plane defines the focal point for all light from the
    same subject point. All of the other rays are bent by the lens, but
    they all must converge to a point that is on the chief ray. Now, if you
    move to a different point in the subject, the chief ray from the new
    point will reach the lens at a different angle, and it must intersect
    the focal plane at a different point.

    This is all the simplest, most basic geometric optics.

    Dave
     
    Dave Martindale, Jul 28, 2004
    #29
  10. This argument has made me think somewhat. So what is the nodal point
    if it is not the point at which all the rays converge? On all the
    diagrams it is marked as such. So what, exactly, is converging at the
    nodal point?
    Perhaps, and I am beginning to doubt my understanding now, but I still
    don't really grasp yours. Maybe I should find something with more
    detailed diagrams. Perhaps I shouldn't have gotten into this argument.
    But if you can explain to me simply what is the nodal point if not
    what I have always believed it is, then maybe I can eat crow and we
    can put this to rest.

    DB.
     
    Dogger the Filmgoblin, Jul 29, 2004
    #30
  11. Alright, this is helping me understand it a bit better. But I am still
    confused about the nodal point. I know that it is the rotation point,
    but what makes it so? I always thought that it was because it was the
    point of convergence, and that made perfect logical sense to me. If
    there WERE a point at which all light converged, it would follow that
    you could rotate around this point without parallax distortion. In
    your 'basic optics' description: what property gives the nodal point
    its special parallax-free property, if not the convergence of all the
    light? If I could understand that, it would be a lot easier to let go
    of my ideas about lens convergence; I learned them all through my
    study of panoramic stitching.

    DB.
     
    Dogger the Filmgoblin, Jul 29, 2004
    #31
  12. No, you should look at the diagrams again. The rays converge at some
    point on the subject in world space, and at some point on the focal
    plane (if the system is focused correctly) in image space. The lens
    nodal points are nowhere near the front and rear convergence points;
    they are usually buried inside the lens, somewhere near the front and
    rear principal planes.

    The rear nodal point is defined to be the point such that, if you rotate
    the lens around this point while leaving the focal plane stationary, the
    image does not shift. It is determined by measurment. But what is
    actually happening inside the lens is generally unknown, and we don't
    really need to understand that for most purposes.

    Here's an argument for why the rear nodal point should exist somewhere:
    Take a lens and mount it on a linear slide which is mounted on a pivot.
    Let the lens have enough freedom of movement so that it can move
    anywhere from well ahead of the pivot to well behind the pivot. Now
    position the lens on the slide so that the entire lens (all the glass)
    is well in front of the pivot (closer to the subject). Set up a focal
    plane that captures an in-focus image. Place yourself so you're behind
    the focal plane, looking towards the lens and subject.

    Now, rotate the lens/slide around the pivot - a few degrees back and
    forth is sufficient. Since the lens is well ahead of the pivot, a
    clockwise rotation around the pivot physically moves the lens to your
    right, while a counterclockwise rotation moves the lens left. The lens
    also rotates around its vertical axis, but most of the motion is
    translation left/right. Now, I hope you'll agree that when the lens
    moves right (CW rotation at pivot), the image will also move right, and
    vice versa. If you don't believe this thought experiment, set up a
    simple test version yourself.

    Now move the lens to the other end of the slide, so it is well behind
    the pivot. Repostion the focal plane so the image is in focus again.
    When you rotate the assembly CW, the lens now moves left, the opposite
    of the previous test. Again, since the lens is mostly translating
    left-right, the main effect on the image is to move the image left and
    right. Still with me?

    Now, we have a situation where, if the lens is far ahead of the pivot,
    CW rotation shifts the image right, while if the lens is far behind the
    pivot CW rotation shifts the image left. Since the whole system is
    continuous, there *must* be some point where you can place the pivot so
    that (small amounts of) rotation around the pivot do not shift the image
    at all. And this point is called the rear nodal point when found.
    Well, I tried above. Or check a beginning optics text, almost any one.
    You've probably been looking at diagrams specific to shooting panoramas,
    which may have been drawn assuming you *already* know optics and optical
    drawing conventions. Find something about lenses and cameras in general
    that doesn't even mention panoramas.

    Dave
     
    Dave Martindale, Jul 29, 2004
    #32
  13. It's not. The simplest handwaving explanation I can supply is: the
    light is at its most spread near the nodal point, nowhere near converged
    to a point. When you rotate the lens around the nodal point, some of
    the lens elements move right and some move left, and the two motions
    cancel out in such a way that the image does not move.

    There's a longer explanation posted in another branch to this thread.
    Are you talking about the front nodal point or the rear one? Are you
    talking about panoramic cameras that swing the lens while leaving the
    film stationary, or rigs that rotate both camera and lens together while
    exposing a strip of film, or multi-frame panoramas?

    In swivel-lens cameras, it's important for the lens to rotate around
    the rear nodal point with respect to the film to avoid image shift
    during exposure. In the other types of camera, the lens/film
    relationship doesn't change and the rear NP location doesn't matter.

    But in all of these cameras, you'd like the whole thing (or just the
    lens) to rotate around the *entrance pupil* with respect to the subject
    scene, since the entrance pupil is the centre of visual perspective for
    the image captured. This is similar to, but not quite the same as, the
    front nodal point.

    Dave
     
    Dave Martindale, Jul 29, 2004
    #33
  14. Actually, I was thinking about it lying in bed last night, and
    suddenly the things you have been saying about how only the light
    beginning with a single point can be focused on a single point snapped
    into place, and its relationship to the diagrams I had seen also
    resolved itself ... those diagrams were tracing all of the light from
    a single point ... I interpreted them as tracing the light of the
    entire image. And I may have misremembered where the nodal point is
    placed on these diagrams, to boot.

    I guess I am still not satisfied that I really know why the nodal
    point has the properties it has in relation to the whole image, but
    your reference to opposing parallax properties on opposite ends of the
    continuum is pretty convincing that there must be a nodal point in the
    middle of that continuum.

    I'm not sure anymore what bearing any of this has on the original
    discussion, but I certainly have learned something, and thanks for
    your patience ... and also for not getting snippy or escalating a
    conflict when one of my messages may have taken on a somewhat arrogant
    tone. You seem like a good guy. (And your knowledge is certainly
    convincing.)

    DB.
     
    Dogger the Filmgoblin, Jul 29, 2004
    #34
  15. The rear nodal point is definitely not where you want the pivot if you
    are moving lens and image plane as a unit, rather than moving lens and
    leaving the image plane alone. For your conditions, you definitely want
    the lens centre of perspective to remain fixed. I've seen information
    that says this is the *front* nodal point of the lens, and other that
    says it is actually the entrance pupil. I'm inclined to believe the
    latter.

    And yes, well-presented misinformation on one web site has a tendency to
    propagate to other web sites since people believe it. Meanwhile the
    optical engineers, who really know this stuff, don't generally
    participate in digital photography newsgroups and don't set up web
    sites. So you have to take *everything* you read on Usenet and on the
    web with a certain amount of scepticism. (This applies to me as well;
    I'm a computer scientist by trade, not an optical engineer. I just
    relate my own understanding of what I've read).

    Dave
     
    Dave Martindale, Jul 29, 2004
    #35
    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.