Resolution with APS sensor vs. full-frame

Discussion in 'Digital Photography' started by Paul Flackett, Sep 25, 2005.

  1. Obviously I don't think I am going overboard at all, otherwise I would
    not have made the point. ;-) However, you are reading it with too much
    restriction to its meaning. Note that the comparison I made was *even*
    with a diffraction limited optic. Whilst most good 35mm lenses will be
    close to diffraction limited at f/8 or higher, they generally fall quite
    a long way short of that at larger apertures - and those shortfalls will
    be just as significant to the small pixels of the 20D as the diffraction
    limitations are at higher apertures. So it doesn't matter whether they
    carry the excess weight of an f/2.8 or an f/4 to stop it down or not -
    the performance limitation will still be more problematic on the smaller
    pixel camera.
    It will be at least proportional to the pixel area. In practice it will
    probably more than that since some of the architecture of the pixel
    itself, eg. tracks and transistor sizes, are fixed, making the ratio of
    storage capacitor of each pixel to be greater than the ratio of the
    pixel areas themselves. This is normal will all CMOS sensors.
     
    Kennedy McEwen, Sep 27, 2005
    #21
    1. Advertisements

  2. Paul Flackett

    ASAAR Guest

    And still much less than the difference multiple-stop difference
    between tiny P&S sensors and APS size sensors, as I said. If the 5D
    gets you about a one stop noise advantage over the 20D that's nice.
    And as I said, for many/most other situations it's a real advantage
    over the 20D's sensor. But in cases such as were under discussion
    where a particular image produced by one lens would utilize far
    fewer pixels with the 5D's sensor, the reduced resolution should be
    far more significant than a little extra noise.

    It appears to me that you're not disagreeing with what I've said
    (nor do I disagree with what you're saying) but you want to find
    ways to put the 5D's sensor in the best possible light.

    <pun intended>
     
    ASAAR, Sep 27, 2005
    #22
    1. Advertisements

  3. Paul Flackett

    Stacey Guest

    Skip M wrote:

    Sure, just seems odd to blow off FOV like it's a non-issue or the "So you
    buy a huge 4X more money lens to get the same FOV" comments like it's not
    even a consideration.
     
    Stacey, Sep 28, 2005
    #23
  4. Paul Flackett

    Stacey Guest

    Sure he is. If what he posted was a fact, a 20D would have a massive
    performance disadvantage vs a 10D. For that matter canon should start
    making a 2MP FF sensors again to get giant pixels?

    Like I said "5D's rule no matter what!"
     
    Stacey, Sep 28, 2005
    #24
  5. Paul Flackett

    winhag Guest

    If you're a wide-angle shooter it's the other way around. Having a
    smaller sensor causes you to spend extra money on a new wide angle
    lens.
     
    winhag, Sep 28, 2005
    #25
  6. Sure he is. If what he posted was a fact, a 20D would have a massive
    performance disadvantage vs a 10D.[/QUOTE]

    That depends on whether you consider the 1 stop resolution advantage to
    be massive or not, doesn't it.
    Using the same flawed logic in your argument, Canon should simply avoid
    large format detector completely and furnish their digital SLRs with 8Mp
    arrays of 2.5um pixels - approximately 9x6mm frames or around 1/6th of
    the APS frame size - since the optical limitations of small pixels are
    "not a fact"!! Just think, you could get all of the advantages of a x4
    EFL multiplier and have a standard 50mm giving you the FoV that would
    require 200mm on a full frame camera - even better than the limited x1.6
    of the APS format. There is a reason nobody does that though - it is
    because your argument is flawed.
     
    Kennedy McEwen, Sep 28, 2005
    #26
  7. Paul Flackett

    Stacey Guest

    That depends on whether you consider the 1 stop resolution advantage to
    be massive or not, doesn't it.[/QUOTE]

    1 stop resolution? I didn't realize resolution was measured in stops..
    Good grief.. Looks like the "full frame guys" are going to be quite
    irritating around here. So we go from trying to discuss if a cropped 5D
    shot would be lower resolution than a uncropped 20D one to this extream
    rant?
     
    Stacey, Sep 29, 2005
    #27
  8. Paul Flackett

    Stacey Guest

    Assuming they already own one plus the APS lens isn't heavier or more
    expencive to the same FOV FF one.
     
    Stacey, Sep 29, 2005
    #28
  9. Paul Flackett

    W.E. O'Neil Guest

    No, I think he is right. For your lens array to cover the same effective FOV
    it did when you were shooting 35mm or 35mm sensor, you are going to have to
    buy one more wider lens to maintain the original widest FOV you had
    previously. Otherwise the crop sensor is limiting the capability you already
    had (assuming 35mm sensor or emulsion)

    Where he is wrong is that he doesn't consider the advantage that brings to
    the shooter 1) elimination of glass edges in the WA shot, 2) more effective
    reach at the other end of your lens array and 3) the shooter may have needed
    a new body plus more reach and decided to buy a camera that granted that
    extra reach without having to buy a new telephoto. So spending the extra
    money isn't only about maintaining similar capabilities, but also about
    improving overall quality and capability across the whole lens array.

    My primary reason for excluding any FF body is the edges of the glass in
    anything 24mm and wider. It has always been a prohibitive shortcoming in
    35mm shooting. It is even worse in FF sensors because at the top of the heap
    the sensors are so very good. My preference has been to add reach (without
    buying another lens) but to eliminate corners in WA while keeping the same
    FOV (by buying a wider lens in order to maintain my previous widest
    perspective).

    At the top of the dlsr market Canon and Nikon have very different ideas
    about sensors and imaging. In Nikon's case they have determined that
    eliminating the corners during the compositional stage is an advantage that
    some people will find worthwhile. I happen to be in that camp, and so am
    shooting a d2x. But I think people who don't shoot 24mm and wider with much
    frequency are going to be better of with a FF body like the 1ds mkII.

    Pixel count isn't as critical to me as it is to some because at the levels
    of 12 and 16 mp it is the software that becomes critical, not the difference
    in resolving power. I think the noise issue is way overstated as
    well.....but hey, this is a newsgroup!

    But of course that is all just my opinion :^)
     
    W.E. O'Neil, Sep 29, 2005
    #29
  10. Just as people with a lens collection they are happy with migrating from
    APS-C to full-frame will only have to buy one lens (or one 1.4x TC (see
    below)), folks migrating from a filled out 35mm lens collection in their
    film days will only have to buy one lens.

    Unless you bet on the completely wrong horse, changing horses in mid-stream
    isn't all that painful. But people who bet on the completely wrong horse are
    going to squawk loudly and obnoxiously about the more sensibly designed
    systems. They can be ignored.
    Uh, that doesn't work. Please, think. (a) Lenses get worse overall as they
    get wider and you have to use a "wider" lens on the small sensor (while some
    lenses really do collapse at the far corners (beyond 18mm from the axis),
    most wides get funky either over the whole field or gradualy). (b) You have
    to use more magnification to get to the print from the small sensor (this
    one's a double whammy, because the smaller sensor has a finer pixel pitch
    requiring more resolution: a 35mm lens has a lot easier time providing 40
    lp/mm (5D limiting practical resolution) than a 24mm lens does providing 53
    lp/mm (20D ditto)). (c) What makes wide lenses get funky at the edges is
    that it's hard to make wide lenses without getting funky edges, so you can't
    even redesign for the smaller sensor. (Although redesigning does have the
    advantage of a more modern design.)

    2) more effective reach at the other end of your lens array

    This one's also less than claimed: if a lens is adequately sharp on a 1.6x,
    it will be adequately sharp on a FF sensor _with a 1.4x TC_. The extra
    sensitivity of the sensor will also exactly make up for the lost speed.
    If you've already got FF, a TC's a better idea than a small sensor. Of
    As before, there's no such thing as a free lunch, and if you see a free
    lunch, you should be suspicious. The 'crop off the bad parts' argument reeks
    of free-lunch-itis. (Note also that decent telephotos have almost no edge
    funkiness to cut off.)

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 29, 2005
    #30
  11. Paul Flackett

    W.E. O'Neil Guest

    Yes, it works perfectly. The bottom line, even considering your comments
    about wides getting worse across the glass, is that this is a comparative
    statement. The point that wides get worse across the glass does not diminish
    the truth that they get exceedingly worse at the edges, particulalry as they
    get wider, and crops sensors disallow the edges in the image. Either way you
    look at it the crop sensor is still using the best surface area of the
    available glass.
    But this is why people who do nothing but base their arguments/decisions on
    the hardware are more often than not thinking along the wrong path. At the
    level of hardware we are speaking about it is the camera software that
    balances out the subtle (and the things you allude to are very subtle)
    differences. Neither of Canon nor Nikon's top line DSLRs would be worth a
    crap based solely upon their hardware attributes. The day the software can
    truly correct glass edge fatalities, I will be a 35mm shooter again...only
    this time with a sensor :^)
    Well, that is not actually what makes wides crappy at the corners of images,
    but in the case of full or small sensors software in t.o.l. bodies is quite
    good at minimizing the effect. But to go further and eliminate crappy
    corners you have to eliminate the edges of the glass from the FOV. There's
    no alternative.

    The issue of small sensors eliminating crap corners is tangible and real.
    Make no mistake about that. I am not saying it's the end all, I am simply
    saying it is an advantage for those who require that advantage. For someone
    who doesn't even recognize crap corners, or for someone to whom they matter
    little, it's obviously a non-issue.

    I happen to be a corner-freak :^) Years of shooting behind a dome have made
    me that way. I suspect most other UW-photographers on the ng feel the same
    way about corners, even if they don't approve of my preference of sensor
    size.
    It has never been my choice to utilize a TC for critical photography. And at
    the level of minutia we are speaking here (because that's really what this
    is) a TC is completely unacceptable for my use.
    Sorry, I don't know what that means or it's relevance to this conversation.
    Apart from the fact that I know a number of pro underwater shooters who have
    migrated back to Nikon (from the 1ds mkII) after the release of the d2x. I
    think that has yet to be decided. I personally don't think 35mm size sensors
    are the future of digital photography. The future is in the glass, glass
    utilization and the body software. Sensor size and pixel count no longer
    hold the importance people defending their choice of camera people want to
    imply.
    Well, first of all it's not a free lunch. It does cause one to expense a new
    wider lens in order to maintain their previous widest FOV, that wouldn't be
    necessary had they chosen a 35mm sensor. Basically, that's a straw man
    argument from you. And while you knocked him down with vim and vigor, it
    isn't exactly relivant.

    And secondly, the use of only the better portion of glass is real and
    accurate. There is no way to deny that.....though people continually try.
    Please note in the post you responded to, that my comments were based upon
    wides equal or greater to 24mm. But also note where the small sensor is not
    advantageous for corner crop it is advantageous for implied reach.

    It does well at both ends of the spectrum and in my opinion is a much better
    choice at both ends. Where the 35mm sized sensor has an advantage, I
    believe, is in normal FOV photography. It is in this area where, based
    solely upon sensor size, the smaller sensor offers nothing more, and in fact
    maybe something less.
     
    W.E. O'Neil, Sep 29, 2005
    #31
  12. If you want to build a lens with a "35mm" FOV, the constraints that make the
    corners problematic in one format are going to make the corners problematic
    in another format. Consider a real-world example: I use the 35/1.4L on the
    5F and you use the 24/1.4L on the 20D.

    Here's your lens:

    http://consumer.usa.canon.com/ir/controller?act=ModelDetailAct&fcategoryid=151&modelid=7303

    And here's mine:

    http://consumer.usa.canon.com/ir/controller?act=ModelDetailAct&fcategoryid=151&modelid=7304

    We both shoot at f/8. The extreme edges of my image happen at 18mm, the
    extreme edges of yours at 12mm.

    Note that the thin dotted blue line in your lens is at 55% at 12mm, but the
    thin dotted blue line in my lens is at 60% at 18mm.

    Note that the fat dotted blue line in your lens is at 85% at 12mm, but the
    thin dotted blue line in my lens is at 83% at 18mm.

    But note that with my lens, I have 25% more (pixel) resolution, but with
    much fatter pixels, which means I don't need as much lens resolution as you
    do.

    I assure that this is a slam dunk for the 5D + 35/1.4 vs. the 20D + 24/1.4.
    I submit that thinking that there's a free lunch is a great way to make
    mistakes.

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 29, 2005
    #32
  13. 1 stop resolution? I didn't realize resolution was measured in stops..
    [/QUOTE]
    It is if you are considering the resolution of the diffraction limit for
    a lens. Resolution cut-off in cy/mm at the focal plane of a perfect
    lens is 1/(W x f/#), where W is the wavelength of the light being
    focussed. A 1 stop advantage in terms of the resolution of a
    diffraction limited lens is effectively a x1.4 gain in resolution.
    So using your own flawed logic back at you is a rant is it? In that
    case what was your initial response if not an unfounded rant?
     
    Kennedy McEwen, Sep 29, 2005
    #33
  14. In addition to which, the shorter focal length necessary to get the same
    wide angle FoV will require to have an even higher inverse telephoto
    ratio, just to leave room for the SLR mirror. That completely negates
    any of the size benefits the shorter focal length would offer, as well
    as making the entire design more complex, expensive and prone to
    aberrations.
     
    Kennedy McEwen, Sep 29, 2005
    #34
  15. Uh, but fair is fair: we have to admit that the EF-S lenses don't have that
    problem. The made-for-digital lenses from all the other MFRs do, though.

    (I'm feeling very pleased with myself that the very first MTF charts I
    looked at showed exactly what was expected: individual lens design
    differences could, of course, swamp the basic design considerations in
    specific cases.)

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 29, 2005
    #35
  16. Paul Flackett

    W.E. O'Neil Guest

    David, you continue to build straw men and then feel pleased that you have
    knocked them down. It doesn't bother me much, but I would prefer not to
    waste the time on that type of discussion.

    Thanks for the comments.
     
    W.E. O'Neil, Sep 29, 2005
    #36
  17. It's not a straw man: the idea that you can improve corner performance by
    using shorter lenses on a smaller sensor is simply wrong. The shorter lens
    on the smaller sensor still has to provide the same FOV, and the reason
    corners are funky is that it's hard to design lenses to provide a wide FOV.

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 29, 2005
    #37
  18. Paul Flackett

    Stacey Guest

    It is if you are considering the resolution of the diffraction limit for
    a lens. [/QUOTE]

    And the camera body has what to do with that?
    Who makes a diffraction limited lens for any of these cameras?
     
    Stacey, Sep 30, 2005
    #38
  19. Paul Flackett

    Stacey Guest


    So in another words, the canon designed for digital lens is a gem, ones made
    by any other manufacturer are crap... So IF there is -any- possible
    problem, a canon product won't have but if it's made by anyone else it
    will? This crap you post is too funny!
     
    Stacey, Sep 30, 2005
    #39
  20. Well, you're pretty funny: the Canon EF-S lens mount has a shorter
    film-to-lens distance than the other mfrs do, so for the issue under
    discussion (degree of retrofocusness) Canon's digital lenses can have better
    designs than lenses that have to meet the requirements of other SLRs. Heck,
    even the Canon EF mount is shorter than the others, that's why you can mount
    Nikon and Zeiss and OM lenses on a Canon.

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Sep 30, 2005
    #40
    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.