Canon 5D vs. Medium Format (Film)

Discussion in 'Digital Photography' started by Progressiveabsolution, Sep 25, 2006.

  1. Not at all. 800mm may sound a lot but it's not really. I already take
    shots at 480mm equivalent with my 300mm lens and I need to get a lot closer
    for certain subjects.

    For example, there's a bunch of factories across the bay that belch smoke,
    and if you catch them at sunrise you can get excellent backlighting behind
    them. Trouble is, they're so far away they come out too small and a lot of
    cropping is required, so I end up with a low res image. No point telling me
    to go closer, I can't walk on water and a boat wouldn't be practical. If I
    drove to a nearer point the angle, composition and perspective would be all

    There's a lot of good coastline where I live, with lots of bays and
    headlands, and there are many interesting perspectives to be gained by
    shooting across bays, but again, I often can't zoom in enough to get the
    framing I want, and cropping loses too much resolution.

    Paul Saunders, Oct 2, 2006
    1. Advertisements

  2. At some point, haze will limit the reach as well.
    Thomas T. Veldhouse, Oct 2, 2006
    1. Advertisements

  3. I know, which is why I love shooting in winter when the skies are clear.
    Now that awful summer is over the light is clearing up once more.

    Early morning shooting after rain the previous day is an excellent time too.
    The rain clears the haze out of the air, and early is better since haze
    tends to build up during the day.

    Paul Saunders, Oct 2, 2006
  4. Progressiveabsolution

    Rich Guest

    Of course not, but there is more detail to be had than simply using the
    400mm lens as a 400mm lens. By artificially (a TC) increasing its
    focal length, you can produce more detail from a given subject you are
    shooting at.
    Resolution is regulated by Dawes Law.
    Resolution of a diffraction limited (very few camera lenses) lens is
    simple. 4.56/lens diameter in inches = the smallest detail in arc
    seconds or, 113/lens diameter in millimeters = the same thing. For
    instance, a 27mm wide lens (standard 50mm f1.8) should resolve
    detail 4.1 arc seconds across. But you will never SEE that on any
    current camera unless you interpose some kind positive lens (like a
    teleconverter) between it and the camera to produce the magnification
    needed to show the detail. It is ultimately determined by the camera's
    sensor, of course, pixel size, but no "stand alone" diffraction limited
    lens can be used to it's best resolution without increasing it's
    overall functioning focal length.
    Bad example. What I mean is that I've seen shots that were soft that
    had better resolution than shots that were sharp. This isn't all that
    That I agree wtih because you are talking about the same pixel count,
    you can't exactly resolve below that.
    Rich, Oct 3, 2006
  5. Progressiveabsolution

    Rich Guest

    The extreme of this being the prosumers of a couple years back with top
    notch lenses (this is when they could charge $900 for a prosumer
    because DSLRs cost more) would outdo DSLRs when it came to pure
    resolving power, because of the extreme pixel density. As long as the
    lens supported it. Of course, dynamic range and high ISO noise were
    the bad part of that.
    Rich, Oct 3, 2006
  6. You can't take multiple shots of action, like sports or wildlife.
    Examples (1D Mark II, 8.2 micron pixels, same size as the 5D):

    On these kissing birds, I had to lighten the background which was
    in shadow:

    On these bears, I had to lighten the faces which were in shadow:

    Roger N. Clark (change username to rnclark), Oct 3, 2006
  7. Hooray! We're agreeing!
    We're still agreeing!
    Which is exactly what I was talking about: using stacked TCs to increase the
    focal lengt. My (admittedly rough) calculation above (based on the 1600/(f
    number) approximation to the diffraction limit) has it that two 2x TCs put
    the diffraction limit well under the resolving capabilities of the 400D
    sensor. If you'd like to correct that calculation, I'd appreciate it.

    But if you want to argue that you can stack an infinite number of TCs and
    get continually better resolution, I submit that you are arguing an
    indefensible position, even if we're talking about theoretically perfect TCs
    (which TCs apparently are very close to being, for small numbers of TCs).

    And my second claim is that what you see with two 2x TCs on the 400D will be
    indistinguishable from what you see with two 2x TCs plus one 1.4x TCs on a
    5D. So if you have a 5D, there's no point in getting a 400D.
    Hmm. Seems were in agreement then. I guess I probably didn't explain well
    enough that I was talking about stacking TCs.

    David J. Littleboy
    Tokyo, Japan
    David J. Littleboy, Oct 3, 2006
  8. It really depends on what you do with your images on whether or
    not noise is important. For example, the are true resolution
    enhancement algorithms (as opposed to unsharp mask which
    only changes accutance, not resolution). But such resolution
    improvement algorithms basically trade noise for resolution.
    Your eye does too. For example, film is incredibly noisy, but the
    eye+brain tolerates it because of the spatial detail. Digital
    images can look better than film even though the digital image may be
    lower resolution than film because the digital has much lower noise.

    I push my system to its limits (1D Mark II), and I find noise
    objectionable even at ISO 100. Not on a snapshot, but on images
    I dodge, burn, and up size for large prints (e.g. 16x24 inches).

    Example resolution enhancement algorithm:
    Image Restoration
    Using Adaptive Richardson-Lucy Iteration

    Responding to other posts in this thread:

    With regard to apertures, noise, and photons/pixel, the situation
    from the film days is different now. Good digital cameras are
    photon noise limited for signal level above a few tens of photons.
    It is basic physics that a 50mm f/2 lens delivers more
    photons per pixel on a full frame sensor than a 50mm f/2 EQUIVALENT
    lens on a smaller format camera. Whether or not this is enough to
    be important for any application is dependent on the need and
    use of the image. But there is definitely a major difference
    in low light performance of a larger pixel camera.

    Digital Cameras: Does Pixel Size Matter?
    Factors in Choosing a Digital Camera

    In particular, see Figure 6, which shows "Sensor Unity Gain."
    This is the gain at which a 12-bit A to D converter converts one
    electron into one digital number (DN) in the output. The range
    of ISO at unity gain foes from about 100 for small pixel cameras
    up to 1600 for the 5D. (The 5D is the highest ISO at unity gain
    of any camera I have data for). This says the 5D has 16 times the
    sensitivity as the small pixel camera with unity gain at ISO 100.

    (The f/ratio myth discussion is on the above web page too.)
    Don't laugh! In scientific applications, when you have time to
    scan an image one pixel at a time, you can get a great signal
    and great signal-to-noise ratios.
    Such systems are in fact operational on spacecraft at this moment.

    Roger N. Clark (change username to rnclark), Oct 3, 2006
  9. Except that we aren't talking about the same pixel count. Due to the higher
    pixel density of the 400D, the same area on the 5D sensor only has 5MP
    compared to the 400Ds 10MP.

    Paul Saunders, Oct 3, 2006
  10. No, we're talking about similar pixel counts because I'm talking about using
    1.4x more TCs on the 5D than on the 400D, and projecting the image from the
    lens onto the same number of pixels.

    David J. Littleboy
    Tokyo, Japan
    David J. Littleboy, Oct 3, 2006
  11. There is if you need/want two bodies and can't afford two 5Ds. I still
    can't afford one 5D!

    If the results are indistinguishable, as you say in the case above, then the
    400D is indeed a good choice for a second body *for telephoto use*. Photon
    counts aside (which is an irrelevant argument if you're using the lowest ISO
    anyway), using a 400D with long telephotos gives you an extra f-stop for the
    equivalent focal length, which is no trivial matter with slow lenses,
    especially when you'd lose auto-focus by adding a TC to the 5D.

    I'm sure the 5D would be the better choice purely from the point of view of
    quality, but quality is not the only issue. Cost, f-stops and auto-focus
    are important factors too. And if noise is not an issue at ISO 100, then I
    don't see that using a 400D is a problem.

    Thanks for the interesting comments about resolution and so on, but you
    haven't convinced me that the 400D isn't a good choice for a second body for
    telephoto use. If maximum quality is required for a particular shot, then
    it's perfectly possible to put the telephoto lens onto the 5D instead, but
    for extreme telephotos and 2nd body convenience (and cost), I still think
    the 400D is a better choice.

    Paul Saunders, Oct 3, 2006
  12. I don't need to, I shoot landscapes. I rarely ever shoot action, although
    there are certain types of shot in which movement can cause problems, like
    swaying branches or crashing waves, or even fast moving clouds.

    But to be honest, there aren't too many shots that need significant shadow
    brightening, and in those that do, movement is not usually a problem, so I
    can take multiple exposures. But if I only have a single shot to work with,
    so long as the really dark shadows aren't brightened too much, it's not
    usually too much of a problem.
    Very nice shots.

    Paul Saunders, Oct 3, 2006
  13. Personally I hate unsharp mask. I never use it for "sharpening", because as
    you say, it doesn't really sharpen. (I do use it for local contrast
    enhancement using a very high radius.) I do all my "sharpening" with Focus
    Magic, which I find produces far better results. Just like USM, FM can
    produce nasty artefacts if overdone, but when applied subtly the result is
    much better IMO.

    Most of their example images deal with blatantly blurred images, but it
    produces excellent results on images that as sharp as possible to begin
    with. Might be worth you trying it on your fox photo.
    But only 1.3 times the sensitivity of the 20D, so not a huge difference

    Paul Saunders, Oct 3, 2006
  14. Yeah, but most camera lenses are *not* diffraction limited except at the
    smallest few apertures, where diffraction is bad enough to dominate
    Telescope objectives are designed for diffraction-limited performance.
    Part of this is because it's important - the highest power used for
    visual observation may be more than 10X the lowest power used, just by
    changing eyepieces and Barlow lenses. And partly because it's not that
    difficult, at least in the centre of the field, because the field size
    is only a few degrees and telescope objectives are often slow (f/10 is
    typical, f/5 is a very fast telescope).

    But (non-zoom) camera lenses are generally used at one magnification,
    or perhaps a 2:1 range with the help of teleconverters. And they cover
    larger angular fields and have much larger relative apertures, so the
    resolution expectations are less.
    As long as the sensor is setting the resolution limit and not the lens,
    you can stack another TC on and see more detail. But eventually, the
    image will be magnified so much that its resolution is less than what
    the sensor is capable of. Further magnification at that point is futile
    - you won't see any more increase in detail although everything gets

    Dave Martindale, Oct 3, 2006
  15. Progressiveabsolution

    Rich Guest

    Sure. You can stack up TCs (provided they don't degrade the image
    themselves) until the diffraction limit of the lens itself is reached.
    They used to say that with a 2mm exit pupil at the eyepiece, you would
    see all the detail a diffraction limited lens can deliver. This works
    out to a "power" of 50x or an effective focal length of about 2500mm
    for a lens aperture 100mm wide so in theory, a diffraction limited
    telephoto like that won't max out on detail (for the human eye anyway)
    until f25. This is probably somewhat "short" for sensors, which might
    require f50-f75 to properly image a detail with such a lens. You can
    do the math based on the pixel pitch.

    As for the 400D and the 5D, it is possible that by not "stressing" a
    telephoto lens by adding compound TCs, the 400D because of it's finer
    pixel pitch will always be able to outresolve the 5D because residual
    aberrations in the lens would take a lesser toll on the image.
    So by equalizing the magnifications so the 5D matches the 400D when it
    comes to pixel pitch and a selected image, the 5D will still not be
    able to resolve quite as much detail.
    But that is just speculation.

    Someone with a truly top notch telephoto (and I mean, the best one)
    could try this out by mating the telephoto with an eyepiece and
    shooting through it. But it would be hard getting the long effective
    focal length with a fast tele.

    Some experiments have been done in past years to do things like resolve
    ants crawling on
    a wall at a mile distant, things like that.

    There are three phases to magnification.
    1. Detail gain phase where each increase in focal length nets more
    2. Non-gain phase where each increase in focal length only makes the
    same detail larger.
    3. Degradation phase where detail due to over-magnifcation diminishes.
    Rich, Oct 4, 2006
  16. Progressiveabsolution

    acl Guest

    Of course, for someone who routinely probes the limits, a small
    difference there will be important. "Blown out of proportion" referred
    eg to replies given to various people who ask which camera to buy, and
    are told to "get x because it is less noisy than y". If someone's asking
    in a newsgroup about this, I doubt the slight difference between cameras
    with similarly-sized pixels will make all that much difference to him.

    Actually, you may have made a political mistake by labelling it a myth,
    since this seems to stir people into negative reactions wrt it. I
    imagine your life would be much easier if you had called the article eg
    "Some interesting Observations regarding the Significance of the f/ratio
    of Photographic Lenses, With Applications to Modern Digital Cameras"
    (plus it sounds like the title of a 19th century treatise, which is
    always nice).
    acl, Oct 4, 2006
  17. Progressiveabsolution

    Bill Funk Guest

    And single-pixel consumer cameras may not be far off:
    Bill Funk, Oct 4, 2006
  18. Progressiveabsolution

    Scott W Guest

    Using a telescope that you look through is not a good way to analyze
    an image capturing system. It is really pretty simple, this limit of
    resolution is the wavelength of light / f-number. So as an example if
    I have pixels that are 6.4 um in pitch then I know that the image will
    start to look soft at around f/11 and by f/25 there will be no more
    gain in detail captures by keeping the same aperture and going to a
    longer FL (or adding TCs).

    As a partial matter with 6.4 um pixels once you have adding TCs that
    drops you down to f/16 you are going to be getting about all that there
    is to be got out of a lens, and this assumes a really good lens. Larger
    pixels spacing would allow for a proportionally larger f/number. So as
    an example if you were using a sensor with 13 um pixels you could go at
    about f/32

    Scott W, Oct 4, 2006
  19. Progressiveabsolution

    Paul Rubin Guest

    Where do you get that? I always thought the diffraction limit
    depended purely on the aperture (diameter) and not the focal length.
    Paul Rubin, Oct 4, 2006
  20. Progressiveabsolution

    Scott W Guest

    It depends on where you are defining resolution. For angular
    resolution of the scene you are correct it is a function of the
    aperture, but when we are dealing with the resolution on the
    film/sensor it is a function of the f/number. This is pretty handy
    because for any imaging system there will be an f/number that going
    higher then will start to blur the image. This can be used for
    determining the limit of how much you can do with teleconverters.

    Now going to a longer effect FL will not reduce the angular resolution
    but it will needlessly throw away field of view with no more detail

    Scott W, Oct 4, 2006
    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.