1. Reminder: Please use our affiliate links for holiday shopping!

Micro 4/3 ISO = full frame DSLR

Discussion in 'Open Discussion' started by zuzullo, Nov 26, 2012.

  1. zuzullo

    zuzullo Mu-43 Regular

    Why is the ISO performance always the "catch" on M4/3 systems?

    P1470938.

    Wouldn't you rather have less Mpx but keep same ISO performance as Full Frame Pro DSLR?

    Imagine you take the 36Mpx sensor out of the Nikon D800 and place is inside the Olympus OM-D E-M5 or the upcoming Panasonic GX2.
    Of course that sensor will only capture the crop 2x format of the Micro 4/3 cameras.
    If you do the maths, that portion is about 10Mpx!
    m43+=+FF.

    Only 10Mpx! OK! But remember that you will get the same ISO performance as a Professional Full Frame camera, just in a tiny body, no?

    Sure we would love it! That is what micro 4/3 should be all about.
    Small and beautiful!
    Taking a tiny camera to the top of a mountain and enjoy the tour itself ;)

    The technology is here, no?
    Cheers everyone!
     
  2. addieleman

    addieleman Mu-43 All-Pro

    Aug 5, 2010
    The Netherlands
    Ad
    A µ4/3 sensor will not perform equally as a FF camera with the same pixel cell properties, because the µ4/3 file has to be enlarged more than the file of the FF camera to get to the same print dimensions. It's exactly like in the old days with film: the same emulsion properties on larger formats yield better picture quality.
     
    • Like Like x 1
  3. svtquattro

    svtquattro Mu-43 Veteran

    354
    Sep 24, 2012
    Vancouver, Canada
    It's a beautiful image there.
     
  4. RnR

    RnR Mu-43 All-Pro

    Sep 25, 2011
    Brisbane, Australia
    Hasse
    I think you misunderstand how light works. A FF sensor is close to 4 times larger in surface area compared to a m43 sensor. Which is about 2 stops more light. You can't break those laws of physics ;)

    Nice shot at any rate!
     
    • Like Like x 1
  5. littleMT

    littleMT Mu-43 All-Pro

    Apr 8, 2012
    Lucille Sanchez
    more Sony RX1 for me.
     
  6. goldenlight

    goldenlight Mu-43 All-Pro

    Jan 30, 2010
    Essex
    John
    No, it's close to 4 times the surface area.
     
    • Like Like x 3
  7. goldenlight

    goldenlight Mu-43 All-Pro

    Jan 30, 2010
    Essex
    John
    I can't understand why people expect it is possible to get Full Frame IQ from :43: It's horses for courses; for most people, if they are honest, :43: is more than adequate plus it has major size and weight advantages. Those who really do need Full Frame performance should buy a Full Frame camera.

    Would you expect it to be possible to tune a family saloon to match a Ferrari for speed and handling?
     
  8. bartjeej

    bartjeej Mu-43 Regular

    50
    Oct 9, 2012
    Addieleman is right. If you take the center portion of the FF camera and put it in a micro43 body, you don't end up with better (or worse) image quality than the native m43 sensor (assuming both sensors have equal efficiency).

    In effect it's the same as viewing the m43 file at 50% and viewing the FF file at 100%; any noise that is still there in the FF file is going to show up, whereas the (admittedly higher) noise is partially masked on the m43 file.

    great shot though!
     
  9. Promit

    Promit Mu-43 All-Pro

    Jun 6, 2011
    Baltimore, MD
    Promit Roy
    If you took a look at a similar D800 image at 100%, I think you'd find that it is actually about as messy as the m4/3 image. The reason a full frame D800 sensor works is BECAUSE you get to average out so many pixels, which naturally suppresses noise.

    Full frame pixels aren't made of magic.
     
  10. zuzullo

    zuzullo Mu-43 Regular

    Thanks guys! This is really something I need to understand:

    1 - Isn't the pixel-density directly linked to ISO performance?
    2 - "Less pixels = more light per pixel = better ISO"! No?
    3 - If you increase pixel size on m43 sensor, it will have better ISO performance! No?
     
  11. kinlau

    kinlau Mu-43 Top Veteran

    836
    Feb 29, 2012
    I have a 10 yr old 1Ds that sucks at any iso greater than 200, but that's not why I use it. By comparison, my GF3 is smaller and lighter than just one of the Canon's batt with much better high iso performance.

    Handling is a big reason to use one over the other. Sometimes you want the bigger size, sometimes you want the smaller.
     
  12. meyerweb

    meyerweb Mu-43 Hall of Famer

    Sep 5, 2011
    All else being equal, yes. None of the above relates to a comparison with a FF sensor, however. And "all else" is rarely equal.

    But if you took the technology in the OM-D and applied it to a sensor that traded off resolution in favor of larger photosites, then yes, noise should be lower.

    But a FF sensor gets it's noise performance not just from larger photosites, but from greater total surface area, too. So 1/4 of a FF sensor shouldn't be expected to perform the same as a FF sensor.
     
  13. madmaxmedia

    madmaxmedia Mu-43 Veteran

    335
    Feb 20, 2010
    1. It is a significant factor (perhaps the biggest) in ISO performance, but not the only one.
    2. More light per pixel= better ISO. But having more pixels as well will give even better results.
    3. Yes, but perhaps at the expense of overall image quality- it's a tradeoff.

    Let's take your hypothetical example above, of a Nikon D800 vs. 10MP Panasonic GX2. Let's assume that the GX2 uses the same sensor as the D800, just smaller (to eliminate differences between 2 different sensors). Let's also assume similar high quality lenses are on both.

    Images viewed at 100% should be close in amount of noise at high ISO. But...the D800 image would be 4x larger, and will have significantly more detail. In addition, if you were to downsample the D800 image to the same resolution as the GX2 image, the resulting image would have much less noise and better detail than the GX2 image. So is the GX2 technically equal or inferior to the D800 in high ISO performance? You could semantically argue for both, but in the real world the D800 will be better at low light performance.

    So while amount of noise decreases as you reduce pixels (to increase pixel size), higher resolution also has its benefits so it's ultimately a tradeoff of sorts.
     
  14. bartjeej

    bartjeej Mu-43 Regular

    50
    Oct 9, 2012
    When viewing an image at 100%, less pixel density means better high ISO performance (together with efficiency). However, most people care more about what the image looks like at a certain output size; for instance 20x15cm or 1024x768 pixels. If you want to view your image at 20x15cm, all of the pixels (however many they are) will be crammed into that space. More pixels on the sensor means a higher pixel density on the given output size, which means each individual pixel is less visible. That, in turn, means noise is less visible.

    You could achieve the same low noise by having fewer, but larger pixels, where each pixel has less noise, but any noise that is left, is more visible because each individual pixel is printed larger. So, in terms of noise at a given output size (instead of at 100%), pixel density doesn't really matter. More pixels does give you higher resolution if you make large prints (or have a very high resolution monitor) though, so there's actually an advantage to having high MP (provided your lens is good enough to take advantage of it).

    One downside of higher MP counts is potentially lower dynamic range, since more MP's means more electronic wiring, which gives off some noise (called read noise). The higher read noise means that the lower limit of the dynamic range (which is usually defined as the point where noise is higher than the amount of light signals gathered) is reached earlier. But in anything other than deeeeeeeeeep deep blacks, the read noise is almost irrelevant on modern sensors.
     
  15. goldenlight

    goldenlight Mu-43 All-Pro

    Jan 30, 2010
    Essex
    John
    Less pixels generally allows each pixel to be bigger which, all other things being equal, does mean less noise at high ISO. The reason is that each pixel is able to output a stronger signal that needs less amplification and it's the amplification which causes the noise to be more prominent. So limiting a :43: sensor to 10mp would in theory equate to 40mp on Full Frame. However, there is no free lunch and by limiting the number of pixels you are also limiting resolution. Thus, broadly speaking, an A4 print from the 10mp :43: camera should have similar resolution and noise characteristics to an A2 print from a 40mp full frame camera. The big camera clearly still has the advantage.

    What Olympus have at last achieved (with a lot of help from Sony) is a very acceptable compromise between resolution and high ISO performance, but the format won't ever get even close to matching the outright performance of Full Frame, nor should it ever be expected to do so.

    Back in my film days, despite the excellent transparency films available, I just wasn't satisfied with the quality I could get from 35mm and so eventually moved to medium format. The vast majority, however, were happy with 35mm despite it's obvious "inferiority," whilst a few found even medium format lacking and insisted on using large format. It's the same today with various digital formats, each one suits a different need.

    One thing that does puzzle me about the excellent picture in your opening post is why you should want to have shot it at a higher ISO but shorter shutter speed on a Full Frame camera to get roughly the same quality. Surely the whole point would have been to use the same ISO and get better quality?
     
  16. Amin Sabet

    Amin Sabet Administrator

    Apr 10, 2009
    Boston, MA (USA)

    Here's how it breaks down:

    35mm format is 4X the area, which is a 2-stop difference in light collecting area. However, you have to stop the lens down for two stops of light loss in order to get the same depth of field in a 35mm format shot that you would get with the 4/3 format, so for a given framing and depth of field, the image quality is the same when shutter speed is fixed.

    Take the photo in the OP as an example.

    On Micro 4/3 it required an Xmm lens at f/Y ISO 640 6s exposure

    On 35mm format to get the same image brightness, DOF, framing, and exposure time you would need a 2Xmm lens at f/2Y and ISO 2560 for 6s, and neither format would have an inherent image quality advantage over the other. The image quality advantage of larger formats only comes into play when a more shallow DOF and/or longer exposure time relative to the smaller format is acceptable.
     
    • Like Like x 3
  17. wclavey

    wclavey Mu-43 Regular

    45
    Sep 27, 2012
    Houston, TX
    Westley
    This was not exactly the scenario that the OP was posting. If we go with your suppositions (same sensor, etc.) but we put the FF sensor behind the m4:3 lens, then the image size on both sensors should be about the same... i.e., the image from the lens will cover the same area on each sensor. I believe that, in that case, because we assumed that the 4:3 sensor and the FF sensor were the same except for dimensions, the image quality of both images would be identical - - and the area of the FF sensor not exposed by the m4:3 lens would be black. It would contribute nothing to the overall quality of the FF image.

    All the arguments about the fewer multiplications of the image size to get the same size print are irrelevant in the problem as the OP posted it since the image on the sensor will be determined by the image circle of the lens, not some apparent equivalent overall image composed on each camera, using a lens that is of the right focal length and equivalent quality on each camera to create a roughly equivalent image on each. Of course, there will be a few other gating factors that effect image size, like the throat of the lens mount, etc...

    But roughly speaking, his question, as posed, with the assumptions made by madmaxmedia, should yield 2 images that are exactly the same. How this would work with real sensors with real specs, I make no claim for knowing...
     
    • Like Like x 1
  18. Yes, but with the Micro 43rds camera you have money left for gas and lunch.
     
    • Like Like x 1
  19. goldenlight

    goldenlight Mu-43 All-Pro

    Jan 30, 2010
    Essex
    John
    And a back and shoulders that don't ache! :biggrin:
     
  20. MAubrey

    MAubrey Photographer

    Jul 9, 2012
    Bellingham, WA
    Mike Aubrey
    No, it's all about the total photons. FF gets four times as many photons. Assuming equally efficient sensors, FF gets two extra stops of ISO (i.e. 4x)
    What matters is total photon count. Four times the size, four times the photons. The total number of photons is across the sensor what matters, not the number of photons in each pixel. Total photons competes with the read noise of the sensor itself. Lower read noise results in a cleaner picture. This is why the OM-D does so well, the efficiency of its sensor also it to make up for the fewer number of photons that hits the sensor, so much so that it is roughly equal in resulting ISO quality as the original FF Canon 5D (see DxOmark.com)
    If you get more photons on to the sensor by doing so, then yes.