Full-frame sensors can't do wide angle - NOT!

Discussion in 'Digital Photography' started by David J. Littleboy, Sep 22, 2005.

  1. Phil Askey being wrong doesn't excuse morons who don't look up things in
    reliable references. Or quote advertisements as though they were technical
    sources.

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 25, 2005
    #41
    1. Advertisements

  2. David J. Littleboy

    BC Guest

    Brian Baird wrote:
    "> To say nothing of explaining how a high angle of incidence can even
    occur in

    Indeed. Again, the 35mm SLR format relies on reverse telephoto designs

    for wide angles, which takes care of the angle of incidence problem.

    I'd like to see how Doug explains telephoto lenses that produce
    chromatic aberrations with near-perfect angle of incidence. "

    Its good to see that at least a few people understand the reverse
    telephoto thing. With regard to telephoto lenses, many of these
    actually force a angle of incidence than you would expect simply to
    avoid vignetting at the lens mount throat.

    Another phenomenon that the sensor-induced "CA" folks would have a hard
    time explaining is the fact that many zoom lenses have a constant exit
    pupil distance throughout the zoom range, and yet the lateral color is
    not constant. Examples include the Nikkor 50-135/3.5 and 50-300/4.5.

    Brian
     
    BC, Sep 25, 2005
    #42
    1. Advertisements

  3. <G>

    Speaking of reverse telephotos, I looked up "telecentric" as best I could,
    and got something about lenses for industrial inspection/image processing
    whose magnification was constant with subject distance (obviously not the
    sort of lens I'm used to). From that I inferred that "one of the most
    traditional and respected lens makers in the world" had completely lost it
    and was spouting lying snake oil in a desperate attempt to get back into the
    game.

    Was that inference wrong? I.e., could you point me to a reference that
    describes how a 'telecentric' lens could be useful for pictorial imaging?

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 25, 2005
    #43
  4. David J. Littleboy

    Chris Brown Guest

    It's like rain on your wedding day, or something. ;-)
     
    Chris Brown, Sep 25, 2005
    #44
  5. David J. Littleboy

    Marli Guest

    Check out the corner shading... No thanks

    Great to have FF but they have to sort out the above first..
     
    Marli, Sep 25, 2005
    #45
  6. That's a 12mm FOV: anything better than cos^4 falloff (the theoretical
    falloff for a simple lens) and you have to say 'thank you'; that's a
    flipping amazing image by any prior photographic standards. And corner
    falloff is less than minimal with the 17mm el cheapo lens.
    Sounds like you're too cheap to buy a 5D.

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 25, 2005
    #46
  7. It has not much to do with FF.
    It's the lens that causes vignetting, it's the size of the sensor that
    records it.

    Bart
     
    Bart van der Wolf, Sep 25, 2005
    #47
  8. David J. Littleboy

    BC Guest

    David J. Littleboy wrote:
    "Speaking of reverse telephotos, I looked up "telecentric" as best I
    could,
    and got something about lenses for industrial inspection/image
    processing
    whose magnification was constant with subject distance (obviously not
    the
    sort of lens I'm used to). From that I inferred that "one of the most
    traditional and respected lens makers in the world" had completely lost
    it
    and was spouting lying snake oil in a desperate attempt to get back
    into the
    game.


    Was that inference wrong? I.e., could you point me to a reference that
    describes how a 'telecentric' lens could be useful for pictorial
    imaging? "


    The part about a telecentric lens having magnification constant with
    subject distance isn't really accurate. I think its more accurate to
    say that if you defocus a telecentric lens the image will become
    blurred, but will not change size. Also note that the industrial
    inspection lenses you refer to are generally telecentric in both object
    and image spaces, and therefore afocal for an object at infinity.

    True telecentric lenses are widely used in LCD and DMD projectors due
    to the nature of illumination used in these systems.

    In photography telecentricity, or at least quasi-telecentricity, does
    help eliminate any illumination falloff caused both by the sensor and
    by cos^4 effects.

    There is nothing wrong with using telecentric lenses for photography,
    but it isn't really necessary. As you've pointed out, the optical
    constraints imposed by an SLR viewfinder provide enough exit pupil
    distance to get good results with digital sensors. Sorry I can't
    provide you with a written reference for this, but I have been involved
    in several projects where an overly tight telecentricity spec turned
    out to be wrong.

    Brian
     
    BC, Sep 25, 2005
    #48
  9. David J. Littleboy

    Chris Brown Guest

    It's a 12mm zoom lens - what do you want, blood?
     
    Chris Brown, Sep 25, 2005
    #49
  10. David J. Littleboy

    Brian Baird Guest

    Wow... you're pretty vituperative and yet... you don't actually say
    anything of value. You're the troll, Dougie, not I.
    Ha. You started this, not me.
     
    Brian Baird, Sep 26, 2005
    #50
  11. [A complimentary Cc of this posting was sent to
    David J. Littleboy
    Actually, sensors also HAVE some "chromatic" component in the
    "abberation" they create. It may result in some "red fringing"
    (though I doubt is ever so pronounced to be visible by "naked eye").
    However, unless the lens is of fantastically good resolution
    the abberations of the lens are going to be much higher, so they will
    hide this.

    The "abberations" of the sensor are due to non-horizontal MTF curve of
    the sensor. The "chromatic" part of these abberation correspond to
    MTF curve depending on the frequency of the incoming light.

    [The physical reason for it is the electron/hole diffusion. A
    captured photon creates an electron/hole pair, which moves in
    semiconductor until it is "captured" by one of the capacitors
    (=cells); so if a pair is created near a boundary of a cell, it has
    a good chance to discharge the neighbour cell instead of the
    "proper" one. Photons of different wavelength are captured at
    different depth in semiconductor, so electrons/holes travel
    different lengths until captured, so those created deeper have a
    larger chance to diffuse further away.]

    Hope this helps,
    Ilya
     
    Ilya Zakharevich, Sep 26, 2005
    #51
  12. David J. Littleboy

    prep Guest

    Blooming. Could you explain how the lens COULD change this?

    BTW, CCDs and CMOS chips can be fabed with anti-blooming as part
    of the design. Cuts well capacity to hell though.

    --
    Paul Repacholi 1 Crescent Rd.,
    +61 (08) 9257-1001 Kalamunda.
    West Australia 6076
    comp.os.vms,- The Older, Grumpier Slashdot
    Raw, Cooked or Well-done, it's all half baked.
    EPIC, The Architecture of the future, always has been, always will be.
     
    prep, Sep 26, 2005
    #52
  13. David J. Littleboy

    Brian Baird Guest

    I don't think that it's as much as an issue with CMOS, which can do per-
    pixel noise sampling and reduction. In any case, blooming isn't a
    problem with any modern DSLR CCD or CMOS sensor.
     
    Brian Baird, Sep 27, 2005
    #53
  14. Here's my understanding of it:
    With a non-telecentric lens, all points in the image plane *except* the
    one in the very centre are illuminated by a ray bundle that arrives at
    the surface on a slant. Roughly speaking, this bundle appears to leave
    the lens via the exit pupil and converges on the destination point in
    the image surface. Because the centre axis of this cone of light is not
    parallel to the optical axis, if you throw the image slightly out of
    focus you *also* change the image size.

    This will be familiar to anyone who's tried to set an enlarger to make a
    print with some object in the image exactly some size, or anyone
    trying to frame a macro shot precisely. If you adjust focus, the size
    changes. If you adjust size, the focus changes.

    Now, imagine a lens whose rear element is larger than the image area,
    and whose optics are arranged so there is always *some* ray through the
    lens that arrives exactly perpendicular to the image surface. Further,
    imagine the designer has cleverly placed a stop inside the lens so that
    the *only* rays that reach the image plane are this perpendicular ray,
    plus a small and symmetric cone centred on that perpendicular ray.
    This is a lens that is telecentric in image space.

    Yes, it's going to "waste" a lot of light, since the light cone
    converging on any given image point is going to have a small f/number
    (e.g. f/8 or f/16) despite the large size of the lens elements.

    The unique thing about this lens is that as you adjust focus, the image
    remains the same size - because all of the little cones of light are
    parallel to the optical axis. So image size is determined solely by
    subject-lens distance, and you can adjust for best focus without
    messing up the framing.

    One use for telecentric lenses is optical printers - the optomechanical
    monsters used for film special effects before the age of digital
    scanning and recording. An optical printer may project the image of one
    frame of one strip of film as an aerial image in the same plane as
    another piece of physical film, and then photograph both of those
    together onto new film. More complex arrangements with B&W masks
    and adding images via beamsplitter are also possible. The images have
    to be aligned with a precision of a few microns to avoid seeing "matte
    lines" at the boundary between foreground element and background plate.
    In this environment, it's obviously very useful to be able to align
    images precisely, and then adjust final focus without messing up the
    alignment.

    Lenses that are telecentric in subject space are used in measuring
    instruments, where you're measuring the size of something by projecting
    its image on a screen with a scale, and you want it to appear the same
    size even if it's slightly out of focus.

    Dave
     
    Dave Martindale, Sep 27, 2005
    #54
  15. Thanks!

    This makes it pretty clear what the answer is, but you haven't _explicitly_
    answered the critical question here. Which is: "Are any of the Oly 4/3
    lenses actually telecentric?"

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 27, 2005
    #55
  16. [A complimentary Cc of this posting was sent to
    Dave Martindale

    Minor nitpicking:
    By definition of exit pupil, this holds for any lens.
    What you appear to imply is that the exit pupil is at infinity. Is
    this interpretation right?
    Looks like one of those pre-digitally-controlled-era compromises; like
    having equatorial mount for telescopes... "It is easier to design a
    very complicated lens, than design a proper coupling between focus
    ring and magnification ring." ;-)

    Thanks for very interesting explanation,
    Ilya
     
    Ilya Zakharevich, Sep 27, 2005
    #56
  17. I can't answer that. But it should be pretty easy to answer for anyone
    who actually has one of the Olympus lenses. Just set up a makeshift
    "optical bench" with the lens lying horizontally on its side on a
    table, casting an image on a piece of paper. Then insert something
    that casts a shadow between the lens and the paper. If the lens is
    telecentric, the shadow should be at exactly the same height as the
    light-blocking object no matter whether the shadow is at the centre or
    the top or bottom edge of the image.

    On the other hand, with a conventional lens, the shadow will be at the
    same height as the shadow-casting object only on the centreline of the
    image.

    Dave
     
    Dave Martindale, Sep 28, 2005
    #57
  18. The last sentence is true for any lens. But the first sentence is *not*
    true of a telecentric lens, since all the bundles are perpendicular to
    the surface. That's the difference.
    I believe so. One way to make a telecentric lens is to put a small
    stop at the focus of the lens groups that are ahead of (or behind) the
    stop. If that stop forms the pupil of the lens as a whole, it would be
    at infinity when viewed from the subject (respectively image) side of
    the lens.
    In fact, that's an interesting example. Until very recently, it *was*
    easier to build a precision equatorial mount that worked properly than
    an alt-azimuth mount for a large telescope. The AA mount is lighter
    and cheaper, but requires accurately coordinating the movement of 3 axes
    to take time exposure images of the stars. The mount has to move in
    altitude and azimuth continuously, plus the scope tube (or at least the
    camera) must rotate around its own axis, to keep the image stationary.
    Doing that with a fraction of an arc second precision is *hard*, as the
    early AA telescope builders found out.

    Anyway, I don't think you know what an optical printer looks like.
    There is no "magnification ring" because the lens is not a zoom.
    Magnification is controlled by adjusting the spacing between projector
    head, lenses, aerial image head, and camera. All are mounted to a heavy
    cast metal or granite bed. Focus is adjusted in exactly the same way -
    by moving one of the optical elements on the bed. There is no focus
    ring either. The thing is basically a large optical bench with many
    adjustments, except that several of the stations on the bench have full
    film transports.

    Here are some photos. Unfortunately, they don't show the optics very
    clearly:
    http://www.scifi2k.com/misc_html/articles/aricle_optical_printer.html
    http://en.wikipedia.org/wiki/Optical_printer

    Dave
     
    Dave Martindale, Sep 28, 2005
    #58
  19. David J. Littleboy

    BC Guest

    "Yes, it's going to "waste" a lot of light, since the light cone
    converging on any given image point is going to have a small f/number
    (e.g. f/8 or f/16) despite the large size of the lens elements. "

    I just wanted to add here that telecentric lenses don't have to be
    slow. Some of them are extremely fast. Microscope objectives and
    photolithographic projection lenses are two examples that are typically
    much faster than f/1 as well as being fully diffraction-limited. In
    the latter case this is an effective f/#, since these lenses are
    necessarily afocal for an object at infinity (i.e., telecentric in both
    object and image spaces).

    Brian
     
    BC, Sep 28, 2005
    #59
  20. [A complimentary Cc of this posting was sent to
    Dave Martindale
    Depends on what you call "recent". I think the Burakan telescope was
    the first big one with azimuth mount. What it was, design of '60s?
    (I remember its introduction being postponed again and again; then it
    fizzled when it finally was operational).
    This is why I put things into quotes. Thanks, but I have seen many
    optical benches in my life. The moment you can have
    computer-controlled movement of component, and reliable feedback about
    their position, you have digital equivalents of "rings". ;-)

    Yours,
    Ilya
     
    Ilya Zakharevich, Sep 28, 2005
    #60
    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.