Olympus E-PM2 sensor characteristics

Discussion in 'Olympus Cameras' started by Machi, Mar 15, 2016.

  1. Machi

    Machi Mu-43 Veteran

    May 23, 2015
    Multiple online sources provide informations about parameters of sensors of different digital cameras from almost every manufacturer so it's no surprise that majority of cameras made by Olympus are among them.
    On closer examination it's clear that most of them are based on results obtained by DxO-Mark.
    This is true for sensorgen.info or "Photons to Photos" (and it's the reason why Fujifilm cameras are missing in those lists).
    Only source which is using their own measurements is Polish page optyczne.pl.
    I was always interested about those measurements and about possibility to measure them myself. Luckily I found that with proper instruments it's pretty easy to measure basic parameters of sensor of every digital camera which can save images as RAWs.
    Two basic ways to measure multiple sensor characteristics exist, one is relatively quick, second one is more time consuming. For those interested description of them by Craig Stark is here.

    I wanted to do measurements for all ISO settings which are results of analog gain - ISO 200 up to ISO3200 (higher ISO levels are result of digital gain) so I chose the quick way.

    I measured parameters of my E-PM2 camera which has Sony's sensor IMX109.
    This means that similar results would be obtained also for other cameras with this sensor and same electronics (amplifiers, analog to digital converters). Other such cameras are Olympus E-M5 and Olympus E-PL5. Some other M43 cameras are using sensor IMX109 but there were some minor changes in electronics (E-M10) which leads to better usage of available range of "brightness" levels (ADU or DN).
    As many users know, M43 cameras are using almost exclusively 12-bit analog-to-digital converters (ADC). This means maximum of 4096 levels (0-4095) hereinafter referred to as ADU or ADUs (analog-to-digital units).
    In practice this number is lower. There is offset of 255 ADUs (0-254) applied in lower part of available range. This part contains usable information about noise (noise statistics) but information about incoming photons is below noise level and it's effectively lost.
    It's highly probably characteristic of ADC of older Olympus cameras that in some ISO settings they aren't using entire 12-bit range also at higher ADUs. My E-PM2 has at ISO200 maximum 4038, for RAWs from DPreview I measured 4039 (E-PL5) and 4045 (E-M5).
    Newer cameras goes probably always up to 4094 ADUs. For older cameras this is true only for some ISO settings.
    Usable range directly affects dynamic range (DR) and that's reason why my measurements works only with real range and not theoretical range for 12-bit image.​

    So how the results look?
    Surprisingly similar to what others extracted from DxO-Mark measurements.
    Here are results in table form. All of them were obtained at ambient temperature ~21°C.:

    Here are data for dynamic range in graphic form:

    View attachment 117081
    There are few possible reasons for little discrepancies between my measurements and those from DxO-Mark and some of them I want to examine closer in future (effects of temperature on read-out noise and non-linearity of CMOS sensor).
    Other possible reasons for differences are parameter variations between cameras of the same type or that DxO-Mark is using theoretical 12-bit range and not practical range.
    But generally I was surprised how precise measurements are possible under very primitive conditions at home.
    (BTW, DxO-Mark still has not published data for PEN-F. :) )
    Last edited: Mar 15, 2016
    • Informative Informative x 4
  2. coffeecat

    coffeecat Mu-43 All-Pro Subscribing Member

    Aug 4, 2012
    SW England
    Cool! I'm not sure I fully understood this on first read, but I do love a good science experiment (much to the annoyance of the program managers at work...)
    • Appreciate Appreciate x 1
  3. Machi

    Machi Mu-43 Veteran

    May 23, 2015
    I bought new E-M10II few days ago and I wanted to know if 16MPix sensor in new Olympus cameras (E-M10II, E-M5II, E-PL6,7,8) is the same as in my old E-PM2.
    So I did the same set of measurements and results are ........ almost exactly the same as in case of E-PM2!
    There is slight difference in full well and read-out noise (both are higher in E-M10II by few percent) but dynamic range is the same.
    Images with silent shutter shows the same results in all parameters but there is faint banding visible at (at least) some exposure times
    (it's visible in flat images taken at 1 second exposure time).
    I don't expect that this banding will be visible in most "real life" photos as it's really weak (less than 1%).
    • Informative Informative x 2
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.