M4/3 Limited to 16MP, Why?

Discussion in 'Open Discussion' started by umanemo, Aug 2, 2012.

  1. umanemo

    umanemo New to Mu-43

    Aug 1, 2012
    Sorry I "Re-Posted" but I think this deserves it's own thread;

    I shoot to print at 1mt x 1mt. format. The subject requires size. So I could fall into the category of a "Landscape" photog yet I do not shoot Mtns or trees. Detail is what I sell. So;

    I am interested in the M 4/3 format for spontaneity and ease of travel. I have seen some Oly prints from the OMD and they are sharp and at 16MP! So the system looks promising. Panny should come forward with their GH3 this Fall and that I await. For my purposes I wish a 24 MP sensor and I'm in! But this is neither here nor there, my question is more scientific.

    I see a lot of confusing commentary on photo site count vs chip size. Simply I would do the math like this... Other than my 645, I have a old D-Lux5 (A/K/A Panny LX-5) I use for fun, it has a 1/1.7" 10MP sensor, the 1/1.7" sensor is 5.357 times smaller than M 4/3. I have reached 20 x 30 prints (Fuji Lambda) with that D-Lux5 with damn good IQ and without interpolation!

    So why don't the digital Guru's on these forums believe in the math. If the M 4/3 chip is 5.357 times larger than the lowly 1/1.7" (10MP) chip then shouln't it be able to populate up to 50 MP's with just as much comfortable inter-photo-site spacing that the 1/1.7 sensors photo-sites enjoy? (5.35 x 10MP = 53.5MP) I am sure we would all enjoy 334% better resolution from the M 4/3 bodies! (sorry, more math; 54 MP / 16 MP = 334) Does it not work this way or am I missing something?
  2. drewbot

    drewbot Mu-43 Top Veteran

    Oct 21, 2011
    Toronto, ON
    I believe digital sensors have a harder time dealing with noise as you increase the MP.

    A good example is the NEX-5N @ 16MP vs. NEX-7 @24MP.

    At higher ISOs, the NEX-7 turns to mush faster than the NEX-5N.

    And you know the high ISO race is all the rage these days...
  3. RT_Panther

    RT_Panther Mu-43 Legend

    May 4, 2011
    Guess what?
    The BEST NIKON CAMERA THAT MONEY CAN BUY only has 16mp - and this is on a much BIGGER sensor than the µ4/3 sensor :smile:
    • Like Like x 1
  4. Promit

    Promit Mu-43 All-Pro

    Jun 6, 2011
    Baltimore, MD
    Real Name:
    Promit Roy
    Nokia PureView 808.

    Silicon suffers from size. Producing big chips is difficult and very low yield. Compact and cell sensor tech tends not to scale well due to the random defects. Apparently it IS possible as Nokia has shown, but it is not easy to mass produce.

    More generally, building any sensor requires compromises. You sacrifice a lot of other things if you are going to push the pixel count so high. Nobody in the m4/3 world really wants to make that trade-off.
  5. David A

    David A Mu-43 All-Pro

    Sep 30, 2011
    Brisbane, Australia
    There's a 19th century recipe for rabbit stew that reportedly starts with "First catch your rabbit." In this case, before you put a 24mp sensor in a M43 body you have to have a 24mp sensor that is the right size. So far it seems that no one makes one.

    Sony put a 24mp sensor in the NEX7 which was released late last year/early this year. That's an APS-C sensor that's larger than M43. Until the OM-D which uses a Sony sensor, my understanding is that all M43 sensors were made by Panasonic and Panasonic haven't released a 24mp M43 camera so they're not making one. Sony don't release M43 cameras and the OM-D sensor was made specifically for Olympus.

    It's also worth noting that the M43 sensor is around 61% of the area of the APS-C sensor so the pixel density of the OM-D is roughly the same as that of the NEX7. I don't think Sony makes a sensor with a greater pixel density than that so they probably aren't in a position to offer a 24mp sensor for M43 at present, and probably wouldn't want to do so before they put a 36mp sensor in their own cameras and they haven't done that yet.

    It would take time, and cost, for either Panasonic or Sony or anyone else to tool up to release a 24mp M43 sensor. I have no doubt it will come, but M43 has just moved from 12 to 16mp in the last year and I don't think any manufacturer ever moved from 12 to 24 mp in one jump. We were obviously, with the advantage of hindsight, never going to see either Olympus or Panasonic skip the 16mp sensor and jump right from 12 mp to 24 mp.

    And as others above have remarked, the more pixels you squeeze into a sensor of a given size, the greater the noise issue you have to deal with. That can be dealt with in part with sensor design and in part with software but it does have to be dealt with and it takes time for manufacturers, both of sensors and of cameras, to do that.

    So if you want me view it will come, but it won't come this year or even next year. I think the earliest we would be likely to see a 24mp M43 camera is 2014 and it may not even happen then. Sensors and firmware need to be developed, costs worked out, and the 2 M43 camera makers, Olympus and Panasonic, need to do their sums and decide whether there's enough people wanting a 24mp M43 camera to enable them to release one at a price that is both profitable for them and acceptable to the marketplace. The OM-D is the most expensive M43 body so far and while it has certainly captured a market place, it has also gathered a lot of negative comments about its price relative to other M43 bodies. I'm not certain a 24mp M43 camera is a commercially viable proposition at the moment given that it would probably have to cost more than the OM-D.

    Those are my thoughts and they don't count a damn since the only thoughts that really count are those of Olympus and Panasonic since they're the only 2 companies making M43 bodies. If you want a reliable answer you're going to need to ask them and I doubt they'll tell you the true answer since if they do have something in the pipeline, they're not going to announce it before they can actually announce a release date for the camera they're putting that sensor in.
  6. flash

    flash Mu-43 Hall of Famer

    Apr 29, 2010
    1 hour from Sydney Australia.
    Real Name:
    I think you underestimate what a decent 16mp sensor, properly exposed, on a tripod and properly processed can actually do. A meter print is nothing for 16MP.

  7. jyc860923

    jyc860923 Mu-43 Hall of Famer

    Feb 28, 2012
    Shenyang, China
    Real Name:
    I think it's possible and desirable to have higher pixels sensor with proper technology, because Fuji's EXR looks promising to me, you can have larger picture in normal mode, or you can choose to use combining-pixels (I don't know how to name it) for the DR and cleaness with half or quarter sizes.
  8. Bokeaji

    Bokeaji Gonzo's Dad O.*

    Aug 6, 2011
    Austin, TX
    *eliots ssd cries*
    No mas megapicksells!

    - Eliot@Austin, TX
    • Like Like x 1
  9. drewbot

    drewbot Mu-43 Top Veteran

    Oct 21, 2011
    Toronto, ON
    My CPU and GPU are starting to cry with LR4 on my 2008 MacBook Pro.

    I also tried a D800 .NEF...that locked up my Mac nicely.
  10. Sammyboy

    Sammyboy m43 Pro

    Oct 26, 2010
    Steeler Country
    Hell, it was just a few short months ago folks were getting great images from their 12 MP cameras, now it seems they can't get any keepers if they filled up a 16 gig memory card when using a 12MP camera. Crazy.
  11. Tincam

    Tincam Mu-43 Regular

    Apr 25, 2012
    Real Name:
    I would LOVE an M-4/3 sensor with Nikon D4 sensitivity/capabilities, even if only 8-12mp.
  12. dhazeghi

    dhazeghi Mu-43 Hall of Famer

    Aug 6, 2010
    San Jose, CA
    Real Name:
    I think you overestimate the number of people who want a 50MP camera. Until they figure out how to do pixel binning properly, 50MP sensors mean very large files that will be slow and inconvenient to process. Moreover, in most cases, you won't see anything approaching 1.7x the linear resolution of a 16MP camera. The lens and photographer's technique simply aren't up to it.

  13. amalric


    Jul 24, 2012
    Rome. Italy
    As said above, each time you up the Mpx count, you have to denoise more and better. Not always the case, you get artefacts. So if you need that final product, FF is still a better bet.
  14. Ned

    Ned Mu-43 Legend

    Jul 18, 2010
    Alberta, Canada
    The images are sharp - that's what matters. That means that the resolution is there in the system, namely lens resolution and capture resolution. Pixel count and math simply do not tell the story when you're dealing with optics. You'd be surprised how large a print you can get out of a small pixel count if those pixels each have fine detail to them. People still get great large prints out of the 5MP E-1, using sharp, high-resolving Zuiko glass. On the other hand, some people can get mushy, washed-out photos out of a high pixel count system that just adds more pixels without definition from each other. It's too easy for manufacturers to use pixel count as a band-aid.
  15. Jermonic

    Jermonic Mu-43 Veteran

    Jan 14, 2012
    I've done good landscape prints at 1,5 m x 1 m with good results (1-2 meters viewing distance). That was using a Canon Powershot SX110IS @ 9 Mpixels.
    • Like Like x 1
  16. Sanpaku

    Sanpaku Mu-43 Regular

    Jul 24, 2012
    The megapixel race was driven in part by less informed consumers who thought increasing pixel density would always increase output resolution.

    Alas, as ably explained and demonstrated at Cambridge in Color: Lens Diffraction and Color, most current digital cameras are already bumping against the immutable laws of physics in resolution, as diffraction from the lens aperture creates "Airy disks" of photon path uncertainty, and sensor resolution falls with increasing f-stop.

    For a D800 with 4.9 µm pixels, diffraction becomes the limit to resolution starting at between F 8 and F11, becoming worse at higher F stops.

    For a NEX-7 with 3.9 µm pixels, or the OM-D E-M5 with 3.7 µm pixels, diffraction becomes a limit to sensor resolving power starting between F 5.6 and F 8.

    For my little 10 megapixel XZ-1 with roughly 2 µm pixels, diffraction becomes the limit beginning at F 4: I've got about 1 and a half stops of aperture latitude to work with before diffraction becomes limiting.

    But what about wider apertures (lower F-stops)? Physics provides lower constraints there, indeed a perfect 4/3rds size sensor with a perfect Summilux len's F 1.4 could theoretically go up to 400 megapixels and still gain image resolution (but only in a razor thin focal plane). Alas perfection is not for this world, and its usually useful to dial in depth of field to the subject at hand.

    So for a given sensor size (and hence associated) size of optics, we have little to gain in image resolution from increasing current megapixel counts for most all cameras. All additional pixels might do is allow electronic artifacts like color moiré and aliasing to be corrected.

    Only medium format and larger sensor format cameras still have headroom to increase resolution at the moment. A 60 megapixel Hasselblad has 7.4 µm pixels, and resolution is limited only above F 11. It could conceivably increase pixel count by 4-fold before hitting the sort of hard limits seen at OM-D E-M5 or GH-2 pixel size.

    Incidentally, at the Nikon D800s 36 megapixels, most users are discovering that their pro-grade lenses are now the limiting factor in image resolution. I speculate that the 35mm FF market has seen its last big jump in the megapixel race.
    • Like Like x 3
  17. mattia

    mattia Mu-43 Hall of Famer

    May 3, 2012
    The Netherlands
    Print size largely depends on viewing differences; I've printed up landscape panos at 2 meters wide (net resolution of about 50 megapixels, stitched from a 5DII) and you can get fairly stupid close and still see quite a bit of detail. Thing is, even at a significantly lower res (and thus fewer DPI/PPI) the images would have looked perfectly great. Because I don't stand 1 foot away to look at a 6+ foot wide print.
  18. Amin Sabet

    Amin Sabet Administrator

    Apr 10, 2009
    Boston, MA (USA)
    Allow me to play a little devil's advocate here...

    So far, increasing pixel density has for the most part been leading to increases in output resolution, and the only significant downside has been bigger file sizes. Meanwhile we get additional benefits, beyond resolution, such as less aliasing (as you pointed out). There has been a trend towards smaller pixeled cameras typically having slightly more image level noise under low light, high ISO conditions, but there have been exceptions to show that this need not be the case depending on a given implementation.

    f/1.4 is not necessarily such a razor thin DOF in our format when considering ultrawide and wide angle lenses, and a wide lens which is brilliant at f/2.8 would be useful for a lot of high resolution landscape photography. Theoretically speaking, there would be no image quality difference (resolution, diffraction, dynamic range, or anything else) between a 36MP MFT camera shot at a base ISO of 25 with a 12mm lens that offers superb sharpness at f/2.8 vs a 36MP Nikon D800 shot at a base ISO of 100 and a 24mm lens that offers superb sharpness at f/5.6. Only the load on the back of the landscape photographer would differ. Of course, no manufacturer has yet decided to make a 4/3 sensor with a base ISO of 25, presumably because this would limit high ISO performance, let alone one with 36MP.

    Diffraction does not impose a hard limit. Ie, you have to go well past the f-number where diffraction begins to affect resolution to get to a point where the higher resolution counts for nothing.

    I think we have a ways to go yet with megapixels before it becomes useless to go further. If we had a world class 200mm f/2.8 lens that could outresolve the central 50% of a 50MP Micro 4/3 sensor when used wide open, then increasing sensor resolution to 50MP would effectively give us about 70% more "reach" as compared to a 16MP sensor when dealing with fixed distance objects that don't occupy more than that 50% central portion of the frame (eg, the moon or a small bird on a branch high up in a tree).
  19. CUB

    CUB Mu-43 Veteran

    Apr 19, 2012

    No need to SHOUT! :smile:

    The "best" Nikon DSLR is not necessarily the most expensive (D4). It all depends what you want it for.

    If you need the maximum number of pixels, choose the D800 or D800E. They have the same 36 MP sensor and both cost a lot less than the D4 with "only" 16 MP.

    Apart from brief flirtations with Olympus, Pentax and Canon, I have used Nikon SLRs since the early 1980s and have used the F, F3, F4E, D3, D700, D7000 and now the D800. My current outfit is D800, D3 and D7000. The best Nikon DSLR for me is the D800. The best film SLR for me was the F4E.

    The OP should not even think about m4/3 for his specific task. Instead, he should consider a Nikon D800 or D800E. If they are too expensive, there is the entry-level D3200 with 24 MP and a body size that is not much larger than m4/3.
  20. umanemo

    umanemo New to Mu-43

    Aug 1, 2012
    I think what I am reading is that there is a "Chicken and the Egg" scenario forming here. If it is theoretically better to have more MP's for work like mine (Is it?) then I would have to lug around something like a E800 or Mk V. I am doing that now with the 645 and relative support gear.

    Also in that scenario I see that the hardware pipeline would also have to be streamlined. I.E. Faster and Bigger memory cards, more portable storage, and more efficient processing CPU capabilities and everything has to become more portable. I know the annoyances - when converting my D-Lux files from RAW even my quad core laptop gets boggy! Even when using Lightroom.

    So I guess that what I hear you saying is that I need to jump onto "Mr. Toads Wild Ride" and hang on tightly until the rest of technology catches up to us? Will larger format ever get there?

    Is there a platform that has greater portability and ease of use than full frame SLR's? If we go back to the Math and Theory then if everyone is raving about Nikon's FF 33MP sensors then wouldn't it seem that in proportion that the Mu 4/3 optimum relative to the Full Frame would be a little over 8MP given that Mu 4/3 is one quarter the FF size. Then I guess that 16MP would be crowding things too? Ok, I'm slow but I am getting it now. Looks to me that Oly and Panny are pulling off close to the impossible as it is. I getcha!

    I'll open the floodgates; What does anyone recommend I jump into? You know my output requirements and I always use a prime ultra wide (21 - 24mm) piece of glass. Fuji Pro 1 with the upcoming 14mm anyone?