Olympus SLR boss says 12 MP is enough

Discussion in 'Olympus' started by rpd, Mar 11, 2009.

  1. A 48MP 24x36mm sensor works out to about 120 lp/mm. The Rayleigh
    diffraction limit for 50% MTF at f/8 is about 86 lp/mm for 550nm
    light. For red light at 120 lp/mm, you need to open up to f/4.5
    or so.

    That doesn't take into account any AA filter.

    In practice, people like Erwin Puts found that even 100 lp/mm was very
    difficult to achieve even under laboratory conditions. Eventually,
    focus accuracy and vibration become the gating factors rather than the
    sensor or lens.

    That isn't to say that a 48MP sensor would be worthless -- it could
    provide more data for sharpening algorithms and for color accuracy.

    Personally, I agree with Mr. Watanabe -- assuming that people are
    viewing the resulting print as a whole. But that assumption fails for
    group portraits and panoramas, which are often printed larger so that
    viewers can examine regions of the shot more closely.
     
    Michael Benveniste, Mar 12, 2009
    #21
    1. Advertisements

  2. rpd

    rjn Guest

    At some MP, you no longer need an AA filter at the
    Bayer sensor, because the lens is the AA filter.
    More MP, assuming they resolve something, is also useful
    when extracting regions of larger images. I'm often going
    back to old product images to extract details I didn't know at
    the time were going to be needed later.

    But only the whole, the chief beneficiaries of the MP Wars
    are the makers of flash RAM. Were it not for photo MP, a
    couple of them might be shut down already.
     
    rjn, Mar 12, 2009
    #22
    1. Advertisements

  3. rpd

    Marty Fremen Guest

    Nevertheless what he said about the impact of the OM1 is correct. Whilst
    Olympus were not unknown in photography circles, this was their first
    venture into SLRs and their compact SLR system took the world by storm and
    ended up being copied by every other SLR manufacturer.
     
    Marty Fremen, Mar 12, 2009
    #23
  4. rpd

    Get lost Guest

    In two years, the bare entry level 1.5 crops will be 15 megapixels. FF
    will be at least 18. Olympus will be further back. Which is a shame,
    since their lenses are most able to handle more pixels, especially at
    the edge because of their high correction and proper design.
     
    Get lost, Mar 12, 2009
    #24
  5. I think my math is OK, but my source is debatable. I've been using
    a constant of 0.38 as described here:

    http://www.normankoren.com/Tutorials/MTF1A.html

    Puts gives 94 lp/mm for f/8 and 555nm, so the "same ballpark" may cover
    it.
    The majority of subjects do not require pixel-level sharpness. I
    think that's Mr. Watanabe's point.
    The Zeiss claim is based on a detection level MTF of 2%, not an
    MTF 50. That's useful for interpretation of surveillance photos,
    but not really for "most applications most customers need."
    Mr. Puts was using slow black and white film in 35mm format, which
    should have been able to easily reach that 100 lp/mm mark.
    http://www.imx.nl/photo/Film/Film/Film/page35.html

    Paraphrasing a bit, there's no substitute for square millimeters. :).
    48MP on 4x5" film works out to about 31 lp/mm, which is a lot easier
    to achieve. One is typically not worried about handheld induced
    camera shake nor mirror slap with a large format camera either.
    Virtually all of photography, including depth of field calculations,
    are based on the acuity and field of vision of the Mark I eyeball.
    But you're actually asking a two-part question.

    1. Can the customers can detect the difference?
    2. Whether they can detect it or not, are they
    willing to pay more for the higher resolution?

    I call this second issue the "Monster Cable" question. But thinking
    about 4/3rds again, it may be that Mr. Watanabe isn't all that
    concerned about the pro market.
     
    Michael Benveniste, Mar 12, 2009
    #25
  6. rpd

    SMS Guest

    It's a good lens because to deliver similar results with the smaller
    sensor it has to be. That's one advantage of a larger sensor.
     
    SMS, Mar 12, 2009
    #26
  7. rpd

    SMS Guest

    I still use my Olympus XA!

    It's just too bad that Olympus didn't leverage their existing SLR base
    and continue with the same mount, rather than coming out with 4:3.

    If you look at 4:3 as a competitor, size-wise, to ZLRs, rather than a
    competitor, quality-wise, to digital SLRs from Canon, Nikon, Pentax, or
    Sony, there is definitely a value proposition for 4:3. You can get the
    tremendous advantages of an SLR over a P&S at about the same size as a
    ZLR, you just can't get the quality of an APS-C or full frame D-SLR.
     
    SMS, Mar 12, 2009
    #27
  8. rpd

    Bruce Guest


    Far too much hype there. Olympus made two SLRs before the OM Series.

    There was the Pen F/FT series of 35mm half frame SLRs, whose lenses had
    already earned Olympus quite a reputation by the time the OM-1 appeared.
    Then there was the Olympus FTL, a 35mm full frame SLR with the M42 screw
    mount.

    But yes, the small OM-1 SLR was indeed something new. It was originally
    called the M-1 and the system of cameras, lenses and accessories was
    known as the M System. Not surprisingly, Leica were not happy about
    this and threatened legal action.

    So, after some cameras and lenses had already been sold with the M-1 and
    M System designations, Olympus changed the name of the system to OM, and
    the camera became the OM-1. The few remaining M-1 cameras and M lenses
    are collector's items.

    There is no doubt that the OM-1 inspired small SLRs from other
    manufacturers - the Pentax ME and MX, and the Nikon FE and FM. But
    Olympus never managed to challenge Nikon for the professional market.

    Yes, a few pros used the Olympus outfits that had been given to them
    free of charge, but very few actually bought them with their own money.
    The OM System was more popular with enthusiastic amateurs.

    Here in the UK, the iconic fashion photographer David Bailey was
    employed by Olympus UK to advertise the OM System. But he never used
    the OM System for his own work. He used, and still uses Hasselblad gear.
     
    Bruce, Mar 12, 2009
    #28
  9. rpd

    Me Guest

    Is the high level of noise in the OOF areas typical for the a900 at ISO
    200? That's a distracting image quality fault more significant than any
    resolution advantage the image may or may not have. I guess a D3x would
    be very much better WRT this fault.
    BTW, I doubt you'd see any difference at all with an 18x12 print between
    12 and 24mp. If I resize that image to 12mp (bicubic), resize it to 24
    again, sharpen slightly to offset losses from interpolation, view at 50%
    side by side with original (a damn close approximation to an 18x12 - or
    slightly larger - print), then there's no discernible difference in
    resolution at all.
    But YMMV.
     
    Me, Mar 12, 2009
    #29
  10. rpd

    SMS Guest

    The Nikon APS-C sensor is 64% larger than the 4:3 sensor. The Canon
    APS-C sensor is 46% larger than the 4:3 sensor.

    I guess if you're comparing this to a full frame sensor that's 384%
    larger than 4:3 then you could claim "not much difference" but in
    reality the 46% and 64% is still significant.
     
    SMS, Mar 13, 2009
    #30
  11. rpd

    Me Guest

    Well I was trolling a bit there... Noise like that probably isn't going
    to be a problem in a print anyway, but you see it clearly at 100% screen
    view. Now if you're going to use 100% screen view to identify
    "resolution advantage", then for the hell of it you might as well use it
    to identify faults - and why not be merciless about it? The $64k
    question is how large do you need to print to see the benefit (and the
    flaws). I suspect that the answer is very very large indeed.
    Here's a guy who seems to have the gear and experience to do some tests
    with a700 vs a900 and various sized prints. He claims that at 18x12
    there's no difference unless you put your beak right close to the paper,
    which is almost as silly as looking at image detail at 100% view on
    screen (only not silly if you're in a "measurebating" contest - or
    printing very very large indeed.
    I'm not saying (and he's not either)) that there's no advantage to 24 vs
    12mp, just that it's a very small difference, and that in almost every
    case where 12mp really and truly isn't enough, then 24mp really won't be
    enough either.
    http://www.astroweb.no/a900/a700-vs-a900.html
     
    Me, Mar 13, 2009
    #31
  12. Fly? Quite possibly, but physical limits will prevent you from
    achieving the resolutions you claimed.
    As you say. Without see the picture nor any reference to it, it's
    impossible for me to disagree. But with a 24MP dSLR, Zeiss's own
    engineers were only to achieve a 10% MTF at 100 lp/mm, much less
    200 lp/mm. See: http://snipurl.com/8xdr7 [zeiss.com] (PDF).
    As opposed to your own undocumented claims?
    You shouldn't be. The f/8 diffraction limit for a 30% MTF is about
    155 lp/mm. By the Higgens formula, to achieve 100 lp/mm at 15% you'd
    need a film with a MTF 50 of about 130 lp/mm. Kodak only claims 125
    lp/mm for TMax 100. So under perfect conditions and with perfect
    processing, it's barely possible.
     
    Michael Benveniste, Mar 13, 2009
    #32
  13. Woah that looks really bad! I really looks like a good 12 MP image upsized
    to 24 MP. Look at the hairs. Mushy details all over.
    Wanna see a sharp high MP picture on pixel level?
    http://www.eliassegui.com/paloma02.jpg

    But nevertheless i´m with you - Olympus is wrong, there is no problem
    stuffing 21 MP or more on a FF sensor, but one thing is true: 95% of the
    users could easily get by with 12 MP.

    Cheers,
    Markus
     
    Markus Fuenfrocken, Mar 13, 2009
    #33
  14. rpd

    Mark Thomas Guest

    I understand (but am not totally convinced it is irreconcilable..) it is
    simply because with contrast detection, the camera's af doesn't know
    which way, or how much it has to re-focus, so it tries one way, dang,
    then the other to look for improved contrast. Whereas phase detect
    'knows' which way to go and also how far..


    It seems to me a little intelligence could be applied to the problem,
    and in fact the G1 already seems to get pretty kind comments about its
    focus speed. I think this issue is gradually on its way out.
     
    Mark Thomas, Mar 13, 2009
    #34
  15. rpd

    SMS Guest

    "A bit" would be like the difference between Canon's APS-C sensor and
    Nikon's APS-C sensor. The Nikon is a "a bit" larger.
     
    SMS, Mar 13, 2009
    #35
  16. rpd

    Me Guest

    LOL
    So a Zeiss zoom isn't good enough for 24mp - even when stopped down to
    f8 (APS-c) and f11 (35mm).
    Really?
     
    Me, Mar 13, 2009
    #36
  17. rpd

    Me Guest

    Look for it? Jeesh - you post a sample image to demonstrate something,
    and the noise in OOF areas is the first thing you see.
    Why?
    Doesn't the a900 have an exposure meter?
    D2x raw files show very very low read noise. At low iso I'm sure it
    would eat the a900 (and 5dII) for breakfast!
    Yes I was.
     
    Me, Mar 13, 2009
    #37
  18. rpd

    Bob Larter Guest

    They look pretty close to me, too. That said, the 'Paloma' shot looks a
    little sharper/cleaner around the eyelashes & eyebrows. Still, I'd
    consider both shots perfectly acceptable.
     
    Bob Larter, Mar 14, 2009
    #38
  19. rpd

    Paul Furman Guest

    What if you use a bellows to enlarge the projected image? I know that's
    kind of cheating but it gets to the heart of what the lens is capable of
    resolving. Or a teleconverter to assess resolution at a distance.

    --
    Paul Furman
    www.edgehill.net
    www.baynatives.com

    all google groups messages filtered due to spam
     
    Paul Furman, Mar 14, 2009
    #39
  20. rpd

    Paul Furman Guest

    4:3 -> 1.5x -> DX -> 1.5x -> FX

    -or close to equal steps like that:

    FX -> .64x -> DX -> .66x -> 4:3

    --
    Paul Furman
    www.edgehill.net
    www.baynatives.com

    all google groups messages filtered due to spam
     
    Paul Furman, Mar 14, 2009
    #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.