Nikon D810 with ISO 64 & 32

Discussion in 'Nikon' started by hjoseph7, Nov 11, 2014.

  1. Forgive me for being out of the loop but I didn't realize that the new Nikon D810 can be operated at ISO 64 even down to ISO 32. Wow that is amazing since only a few years ago the starting ISO range for some Nikon DSLR's was 200. I always wondered why the ISO numbers were getting larger and larger when I thought they should go lower just like in the days of film and slides.
    My question is does an ISO 64/32 translate into increased resolution ? Or is the lower ISO just there in case you want to decrease the shutter speed when shooting wide open ?
    I Remember that back in the days of film the lower the ISO the sharper the film. ISO 64 Fuji film was about as sharp as one could get, but you often needed a tripod !
     
  2. Normally, the "low" settings are only when you have so much light you HAVE to lower the ISO, because they are NOT "better". The best performance from most (all?) sensors is at base ISO.
     
  3. ShunCheung

    ShunCheung Administrator

    The base ISO for the D810 is 64. If you go to Lo 1, which is ISO 32, the quality may get slightly worse. The difference is probably negligible, though.
     
  4. My question is does an ISO 64/32 translate into increased resolution ? Or is the lower ISO just there in case you want to decrease the shutter speed when shooting wide open ?
    As Shun rightly points out , the D810's base ISO is 64. At all setting above or below that, any setting is the equivalent of turning up the gain (effectively, the volume) on a pre-amp. As gain is increased or decreased two things can happen- the effective dynamic range is decreased (with the d810 , up to about ISO 1600 this decrease is negligible) and there's a change in the signal to noise ratio. With the D810 and most of the latest generation full frame and APS-C cameras the increase on the noise side is almost non-existent from bae ISO to 400, and not very noticeable unless you are "pixel peeping" from 400 to 3200 and is easily handled with noise reduction algorithms in your raw processing software with out a loss of detail resolution.
    So back to your question. With the D810 at ISO 64 it's not that you have more detail, but that the signal to noise ratio is very, high and the "texture" of the photos looks smoother.
    You can't compare it to the look of film really. The two mediums are very different. Each medium has both positives and negatives.
     
  5. ShunCheung

    ShunCheung Administrator

    Each medium has both positives and negatives.​
    I thought only film had both positives (slide) and negatives (print film). Digital is always positive. :)
     
  6. Film actually only has negative. What you call positive is reversal or double negative. Pun intended though.
     
  7. Thanks Shun for catching the pun. Technically digital only has null and positive.
     
  8. As film ISO rises, the inherent sensitivity increases. With sensors it seems base ISO is it. Anything more is simply in-camera amplification.
    Does that mean with RAW files, the amplification is all done in post in the processing phase, ie in computer?
     
  9. No, the amplification is applied at the time of capture. High ISO is like recording a whisper (dim light). Low ISO is like recording a loud instrument (plentiful light). You need to amplify that whisper (dim light) in order to record it, and at the same time, noise is amplified and becomes a more significant part of the recording.
    Once you put the file onto the computer, the noise is already part of the recording (digital image).
     
  10. There are of course two kinds of gain - analog, and digital. Modern cameras like the D810 use both. From base ISO up to about ISO 1000-1200, the D810 uses analog gain. After that, it uses digital gain.
    There is no inherent reason why digital gain needs to be applied in-camera. It could just as easily be done using a metadata tag to indicate +N stops of digital gain are to be applied during capture. Getting the developers of capture software to all work with this convention is the sticking point.
    But /so long as you haven't blow any highlights/, you can un-do digital gain so long as it was applied in 1-stop increments. You can multiply a digital integer by two (shift left one bit), and then divide it by two (shift right one bit) without loss.
     
  11. The very specification of how ISO speed is measured dictates that a sensor's "base" sensitivity gives its optimum image quality. Above or below that will give a compromised image.
    The International Standards Organisation specification for digital sensors sets out a number of parameters for "measuring" image quality. Where all those parameters are at a maximum or minimum is deemed to be the sensor's base ISO speed. Anything else is less than optimal, although as long as the image quality remains within certain limits the speed (or rather gain applied) can be legitimately called an ISO speed. However, when noise or other IQ metrics are exceeded, then the speed should no longer be given as an ISO number, simply because the IQ has fallen below the standard specified by the International Standards Organisation. Hence the "Lo" and "Hi" markings at extremes of the sensitivity range instead of ISO speed numbers.
    Speeds lower than the base sensitivity are usually a cheat; inasmuch as the digital data captured is pretty much identical to the base sensitivity and it's only in JPEG conversion that brightness levels are mapped to a lower apparent sensitivity.
    Resolution has absolutely nothing to do with sensor speed, and is fundamentally fixed by the number of photosites that the sensor has - i.e. its MegaPixel count.
    Incidentally Harry, after recently duplicating some old Fujichrome 64 slides on my D800, I can tell you that modern DSLRs simply blow away the supposed "sharpness" of slide film.
     
  12. "Incidentally Harry, after recently duplicating some old Fujichrome 64 slides on my D800, I can tell you that modern
    DSLRs simply blow away the supposed "sharpness" of slide film."

    I pretty much agree RJ as that is my experience as well. But the absolute confirmation would be to shoot the same
    subject in the same light, at the same aperture with a very high quality lens like the Sigma 50mm f/1.4 A, the camera on a
    tripod, with various films (slides at various ISO ratings; color negative, and black & white negative film) and have Lenny
    Eiger of http://www.eigerstudios.com do the film scanning on his drum scanner.
     
  13. Here is a rather good explanation of how ISO works: http://petapixel.com/2014/11/12/understanding-iso-simple-explanation-iso-works-increasing-creates-noise/
     
  14. While the general though it that expanded ISOs do not 'add' anything to an image, DXO's test results on certain bodies, including the D810 indicate otherwise. If there is a difference one way or another, it would be so small it would probably be difficult to see anyway.
     
  15. The base ISO for the D810 is 64. If you go to Lo 1, which is ISO 32, the quality may get slightly worse. The difference is probably negligible, though.​
    That's not entirely true. LO1 deliberately overexposes the image (shoots at ISO 64 but metered for ISO 32), then applies digital exposure compensation to "pull-process". In doing this, you've lost one stop of highlight latitude compared with shooting at ISO 64 in the first place. You do gain some shadow detail through over-exposure, which is thrown away as part of the conversion back to ISO 32. I should know, but don't (or forgot), whether the NEF files have this scaling factor applied before saving, or whether they rely on the raw converter to do it - if the latter, at least you could override and keep that extra data as though you'd actually shot at ISO 64. (I vaguely know that some Fuji cameras rely on the raw converter for higher ISOs, which makes me think that Nikon actually scale the NEFs before saving, but I could be wrong. I'm sure someone on this thread can elaborate.) Nonetheless, you're losing a whole stop of highlight dynamic range compared with a "real" ISO 32, which is arguably not that negligible.
    As gain is increased or decreased two things can happen- the effective dynamic range is decreased (with the d810 , up to about ISO 1600 this decrease is negligible) and there's a change in the signal to noise ratio. With the D810 and most of the latest generation full frame and APS-C cameras the increase on the noise side is almost non-existent from bae ISO to 400, and not very noticeable unless you are "pixel peeping" from 400 to 3200 and is easily handled with noise reduction algorithms in your raw processing software with out a loss of detail resolution.​
    I disagree that it's not significant, Ellis. Up to a moderate ISO the noise is pretty invisible on a default image conversion, because the noise is less than you can detect on a conventional 8-bit display (and it's hard to see more subtle changes than that). However, the D810, like most Nikons from the D7000 generation on, does gain more dynamic range as it drops to native ISO. That makes it much more possible to recover shadow details by "selectively push-processing" (or pushing the shadow slider up in a raw tool, depending on how you think of it) at lower ISOs than at higher. This wasn't nearly so true of, say, the D700 or D3s generations, and still isn't as true of some Canons. With my D700, I was much less worried about auto-ISO because there really wasn't much difference between ISO 200 and ISO 800. With my D800, I really try to keep at base ISO if I can, because I get an extra two stops of shadow recovery by doing so, and I'm often in situations where I don't have complete control over my lighting. There's an example of how the base ISO dynamic range of the D800 helps compared with the 5D3 for shadow recovery here; this advantage of the D800 goes away as ISO increases, even if you can't see much difference on a normal exposure. I've relied on this for wedding candids in direct sunlight, where my D700 would have required fill flash.
    While the general though it that expanded ISOs do not 'add' anything to an image, DXO's test results on certain bodies, including the D810 indicate otherwise.​
    Really, Elliot? I can't see DXO having any measurements at ISO 32 other than for checking what the actual measured ISO is (47, same as ISO 64 - though I'm curious as to why they seem to find most cameras underestimate a bit).
     

Share This Page