Nikon 56 Megapixel camera?

Discussion in 'Nikon' started by ron_togger, Jan 18, 2013.

  1. My Nikon D5100 offers 16mpx across a 23.6 x 15.6mm area, 4828 x 3264. ...If the 5100's sensor had the remaining 12.4 mm added to the width, and 8.4mm to the height to become a full frame 36x24mm, the resolution would be roughly 36mpx, - as per the D800, right?
    Now, as we're already reaching 24mpx on DX cameras such as the D3200 and D5200, an FX version of that would be a whopping 56mpx...
    ...given that resolving power of most lenses is already called into question on the D800, I wonder if there are lenses that would (in theory at least) match the 56mpx resolution, or would we effectively come to practical limit of the FX format...
     
  2. ShunCheung

    ShunCheung Administrator

    We have discussed many times that you need really good technique and good lenses (top optics and stop them down to f5.6 or so) to get the most out of the 36MP D800. Otherwise, the difference between 36MP and 24MP is almost negligible in terms of real resolution difference that you can see, and the extra pixels merely take up additional space on your memory cards and hard drives.
    56MP on FX merely takes that another step further. If you need more pixels, you need a larger sensor. It is just like medium and larger-format film.
     
  3. What I'm hoping is that as sensors advance, instead of becoming larger (e.g. medium format,) they can become smaller so the cameras can be smaller.
    Kent in SD
     
  4. ShunCheung

    ShunCheung Administrator

    What I'm hoping is that as sensors advance, instead of becoming larger (e.g. medium format,) they can become smaller so the cameras can be smaller.​
    To a large degree that has already happened. A little over a year ago, I took some group pictures with the 10MP J1 and printed that to 8.5x11, and it looks very good to me.
    But at least for me, the J1 is kind of too small to hold. So smaller is not always "better."
     
  5. About time. We have been putting up with a mere 36MP on the D800 for what? Quite a few months now!
    I agree about sensor size. More pixels really needs more area. The laws of physics have something to do with it (they should repeal those laws!). It's the same with film - if you want more resolution without grain, it needs to be bigger.
     
  6. I agree with the above - there are a few laws of physics I would like to repeal...
     
  7. There was an analysis done a few years back about the tolerances in lens and camera construction, suggesting that the actual pixel count that's useful is quite limited. Not many lenses outresolve a D3200, and the number that could still do so across an FX frame is probably still lower. There are arguments for oversampling (it might be a way of mapping around small defects to increase yield, for a start), but with limits, and there are bandwidth and heat problems. If you'd like a 60MP image, I'm sure some of the medium format players will happily sell you their models from a few years back, if you don't mind shimming all your lenses to get the best out of them.

    There are repeated rumours about a 50MP-class sensor; Canon did one as a technology demonstrator a while back. Is it possible? Yup. Will it make the entire lens range look bad and potentially be a marketing problem? Probably.

    Though if someone decides to do a high-res centre region on an FX sensor, so you could combine a 24MP DX crop with a 36MP full frame (by edge resolution), that might be interesting, and stop me wondering about whether a DX back-up to my D800 has merit.

    Oh, and this is probably where people start referring to the Nokia 808 PureView.
     
  8. Nokia put 38 megapixels in their pureview 808 camera which has a sensor a little bit smaller than the CX cameras. That will be somewhere around 300 Megapixels on a FX camera.
    I think we will end up somewhere there in the future on FX, perhaps not exactly 300 megapixels but around 150 megapixels or so. But only when we have the technology to shuffle that amount of data in the cameras. With semiconductor performance doubling every 18 months it should be technically possible with 150 megapixels in three years but then the engineers need time to develop the product as well so say 5-6 years.
    Right now we may have 36 megapixels in the D800 but there are only 18 green megapixels, 9 red megapixels and 9 blue megapixels. With a quadrupling of megapixels we are doubling the resolution of the sensor. With 2x2 pixel binning on a 144 megapixel sensor we still end up with 36 megapixels of data to use but now each pixel will have full color resolution. AA filters will become a thing of the past and the lenses will even more than today become the limiting factor on performance.
     
  9. It's the same with film - if you want more resolution without grain, it needs to be bigger.​
    Interestingly, the lenses for medium and large-format are built to a lower resolution....making a lens to cover a 60mm x 60mm sensor at the same resolution of the Nikon V1's lenses (well the good one's at least!) would be pretty expensive.........kinda like satellite expensive!
     
  10. Andrew...
    Not many lenses outresolve a D3200, and the number that could still do so across an FX frame is probably still lower..​
    I'm getting a Deja-Vu from another thread here! http://www.photo.net/nikon-camera-forum/00bF5a
    I'm sure some of the medium format players will happily sell you their models from a few years back, if you don't mind shimming all your lenses to get the best out of them.​

    ....and just how many of my FX lenses are going to cover a 60mm x 60mm sensor?...never mind really, really crappy edge performance not to mention vignetting......:-(
     
  11. What about a square senzor of 36mm x36mm with the same pixel density like D800? A pro-grade camera with such as sensor, with the possibility to pick the ratio you prefer in portrait or in landscape like 1:1 5:4 5:3 3:2 etc... To not need to turn vertically your camera... no need for special bracket... no problems with the flash... to use the most of your lens... this may be interesting for me, personally. Having also the AF points equally distributed in a square.
     
  12. Will it make the entire lens range look bad and potentially be a marketing problem? Probably.​


    It won't make them look bad if they are not bad already but it won't make them look better either.
     
  13. "What about a square senzor of 36mm x36mm..." - Yeah, and why not make it a TLR design too Mihai? For goodness sake! You can crop a D800 image square and still have sharpness to spare.
    Most people I show the comparison A3 prints to, can't tell a picture taken with a 7 megapixel bridge camera from one of the exact same subject made with a Canon 5D and L series glass. Both shots processed from RAW files of course.
    More megapixels? Not needed IMHO. Not until optical formulation is revolutionised (again). Oh, and not until manufacturers can actually be bothered to do some quality control and stop chucking decentred garbage on the open market.
    The argument for or against format size comes down fairly simply to how much control over depth-of-field you want. Really small formats fall between the rock and hard place of not having a fast enough aperture to make DoF shallow enough, and of diffraction ruining definition at smaller apertures. So while 35mm film was utterly crappy because of its grain and accumulated micro-dirt and dust, we don't have those drawbacks with full-frame digital. My vote goes to full-frame as being a near ideal format (although not aspect ratio-wise). It gives wide control over aperture and depth of field, while now having sufficient resolution and low enough noise to satisfy most professional reproduction requirements. Anything further might be seen as gilding an already fairly beautiful lilly.
    DX you can keep for sports, wildlife and macro work - for the moment. That's until 4/3rds of next-to-nothing (or something even smaller) usurps its place.
     
  14. What about a square senzor of 36mm x36mm with the same pixel density like D800? ... To not need to turn vertically your camera... no need for special bracket... no problems with the flash... to use the most of your lens...
    Sure, that's possible. But you'll need an entire line of new lenses for that square sensor format. The current and legacy F-mount FX/35mm lenses, optimized to cover a 24x36mm format (43.2mm diagonal) are not designed to cover a 36x36mm format (50.9mm diagonal).
    So with existing lenses you'll have progressively worse vignetting (light falloff in the corners), or mushy, soft corners (very poor corner performance), or just plain black corners (inadequate image circle).
     
  15. "What about a square senzor of 36mm x36mm..." - Yeah, and why not make it a TLR design too Mihai?​
    But...I enjoy shooting my TLR more than my DSLR.....

    Anywho, my opinion is that Fuji and Sigma have it right in sensor design. Instead of trying to cram more pixels in, they are looking for interesting new solutions. Seriously, those two are the only ones I know who are really pushing the bounds of sensor tech. And man, do those X-Trans sensor look nice....

    As for stuffing more on a chip, if you shot 56MP, then your Raw processor had different binning algorithms (ala Fuji EXR modes) to bring it down to a 24MP image with exactly the right specs (wider DR, less noise, whatever you feel like), that seems like it would be an impressive tool. And it, to my guess, wouldn't be hurt by resolving power of the lenses as much.
     
  16. (43.2mm diagonal)​
    How about a 30.5mm x 30.5mm sensor then - would optimally use the image circle of current FX lenses and preserve the diagonal FOV. Of course 2:3 format is not 20.3 x 30.5 - oh the horror!
     
  17. RE: film size having to be bigger for more resolution/reduced grain.
    Since diving back into film, I have some observations about that. I shoot HP5 (ISO 400) as my regular film in my Voigtlander Bessa 6x9. When I shoot that same film in a Leica, the grain really is noticeable. So yes, a larger format gives improved image quality in that respect. However, it's not the only route to that end. Shooting an extremely fine grain film like Pan F or even Efke 25 in my Leicas, I end up with less grain and more resolution than I get with the Bessa or Rolleiflex. (Lenses are similar vintage and type.) My point is there is more to the equation than just the format size. I too wait to see what the next breakthrough development in lens technology will be. In 1890, it was new kinds of optical glass. In 1950, it was coatings. In 1990s it was CAD and aspherics. Hope I don't have to wait another 50 years to see the next major technology shift! We talk about the laws of physics as if they can never change. While the actual laws might not, the way we approach the design can and probably will change. In the1890s if you had shown them the 18 element lens formulas we use today in pro f2.8 zooms, they would have said that would be totally impossible. The laws of physics said that as light enters/exits each element, some of it is lost or deflected. At that time, a four element Protar was state of the art, and the six element Heliar was the limit. It stayed like that for nearly fifty more years. Bausch & Lomb and also Zeiss began experimenting with coating lenses with MgFl in a vaccuum, which dramatically reduced the amount of light that was scattered between elements. That suddenly changed EVERYTHING! Within ten years we had the Voigtlander Zoomar........
    Kent in SD
     
  18. I mean really guys. This has been there for some time now. Don't forget the venerable old Mamiya. B & H Says:
    The stats break down like this: 80Mp (10,320 x 7,752) are rendered by the 54 x 40mm CCD imaging sensor. The Uncompressed RAW file size is an impressive 165MB, while compressed RAW files are 107MB. Aptus-II backs provide 16 bit high dynamic range color and the ability to render awe-inspiring images, even when tightly cropped.​
    You guys are just not comitted enough to your hobby. A quick $35,045.00 and you can be the envy of the local camera club for a couple of months. $35,000.00 for the camera, lens and back and $45.00 for the used vest. You have to have the vest.
     
  19. As simple and convenient as it is, the analogy between film size and sensor size does not really hold true. As Kent mentioned, shooting something like PanF+ will allow a smaller format to have the sharpness and grain of a larger format. I have said before that PanF makes any camera look like it is shooting a format the next size up.
    However, all that comes with a price. We know this to be true, because if it didn't then the larger formats wouldn't exist anymore.
    I have used a LOT of different films. At one point or another, I've probably used most of the popular films in production right now, and shot them all above or below their box speeds. And while I haven't used more than ten or so different developers, I feel confident stating that you CANNOT get the grain, sharpness AND tonal range of a larger negative without having a larger negative. PanF+ is extremely sharp, and has very little noticeable grain. It also is very contrasty, and has a shorter tonal range than even some 35mm films. Tri-X can be developed to produce some wonderful tones, but is grainier and less sharp that other 35mm films as well. There is always a trade-off.
    Not being an engineer, I see no reason why this needs to be the case with digital sensors. I do see how this is the case right now, but I do not see why that cannot be overcome, even if it takes a very long time.
    As the film/sensor shrinks, contrast increases and tonal range decreases. For a real-world example, try painting a Rembrandt on a 4"x6" canvas and see how far you get. But sensor are fitted with microlenses, while film is not. Given a high enough level of sophistication, the microlenses could be small enough, and accurate enough, that they render detail on a nearly molecular level, allowing for the sharpness AND tonal range of a larger sensor. Yes a larger sensor would still be better, but at that point you're already outresolving the differences that the eye can see even at an extremely large print size, so it's an entirely moot point.
    On the other hand, film has a built-in limiter. You can never make resolve detail finer than the film grain, no matter how sharp your lens is.
    As it stands right now, a larger sensor will still yield better dynamic range, better sharpness, or more often both, assuming the lenses are capable of matching that.
    So even though we'll eventually get a mind-bogglingly good pocket-sized camera, I wouldn't hold my breath, or stop saving up for that FX camera.
     
  20. Given a high enough level of sophistication, the microlenses could be small enough, and accurate enough, that they render detail on a nearly molecular level, allowing for the sharpness AND tonal range of a larger sensor.
    Zack, I'm not sure what you are trying to say. The best of current sensors already record >50% of the light hitting the sensor surface (the quantum efficiency of the D800 sensor has been estimated as 56%; J1 57%). There isn't so much further to go regarding improvements in microlenses. With both film and digital sensors, the tonal quality is limited by the number of photons hitting the sensor or film. Digital sensors are more efficient than film but there is only so far you can go - with N photons the SNR can only be sqrt(N), or worse. This is a fundamental property that limits every system that records information by detecting photons.
    So even though we'll eventually get a mind-bogglingly good pocket-sized camera
    I'm afraid the current cameras are already equipped with sensors that are very close to as good as they will get. There may be improvements - certainly many more pixels, but with regards to signal to noise and tonal quality at commonly used ISOs today's best cameras will stand the test of time. There can be improvements in optics, and probably will be, that improve image quality from what it is today, but you can't escape photon noise.
     
  21. Just to head off more square sensor diversion: The expensive bit in a big DSLR is the area of the sensor. Most prints aren't square. By
    having a rectangular sensor, you reduce the wasted area (and hence sensor expenditure) at the cost of making the user rotate the camera at
    shooting time. Of course the aspect ratio of the sensor needn't match that of the print, and one could argue whether 3:2 is near the typical
    print aspect ratio (or a bit squarer to make better use of the lens), but I don't buy that trying to optimise the lens coverage is the determining
    factor in an ideal sensor. Except, maybe, for a small and cheap one - but compacts tend to use the squarer 4:3 anyway.
     
  22. However, it's not the only route to that end. Shooting an extremely fine grain film like Pan F or even Efke 25 in my Leicas, I end up with less grain and more resolution than I get with the Bessa or Rolleiflex.​
    Unless you use Pan F or Efke 25 in your Bessa or Rolleiflex.
     
  23. In technology advancement, integration has been one driving force. So the question might be: when do we get sensors optimized and paired with optics - as a one unit. In this scenario the sensor does not need to be planar. As sensor prices get lower, this may be one development track. Integrated software takes care of distortions etc.
     
  24. Kari: about 2009 (Ricoh GXR system). Dubious concept - there's not THAT much optimisation, each lens costs more by the price of the
    sensor, and the integration/body cost a lot. You get a body the price of a mirrorless camera and a lens the price of... a mirrorless camera. Big
    sensors are always likely to cost more, since Moore's Law doesn't help fixed sensor sizes. Making non-planar chips cost-effectively is
    tricky, though maybe a new (bio?) technology will appear.
     
  25. Andrew, I agree on that. I remember the Ricoh system, of wich I have only seen an ad. Ricoh was not optimized in the sense I meant. We'll have just to wait a bit more.
    Meanwhile - more photos.
    Cheers Kari
     

Share This Page