DXO Mark results for EM1 MK II

Discussion in 'Olympus Cameras' started by Sootchucker, Nov 13, 2017 at 6:38 AM.

  1. Sootchucker

    Sootchucker Mu-43 Regular

    181
    Aug 9, 2012
    United Kingdom
    Now firstly don't get me wrong, I do love my Olympus system (OMD-EM1 and EM1 MK II), but was recently browsing the DXO website, and decided to plug in the models of 3 of the 4 cameras I own to see the results (they don't have results for the Fuji X Trans sensor of my X-T2 yet).

    These are the results.

    24515386618_7656e6729f_o.

    Now according to this, both my MK1 MK II and my D500 share almost identical low light ISO scores (1312 for the EM1 vs 1324 for the D500), with the GX8 trailing nearly a stop behind at 806. Trouble is though, I'm finding that my D500 wipes the floor with the MK II when it comes to high ISO performance. My cap for the EM1 II when shooting birds to retain maximum feather details is typically ISO3200 (preferably ISO1600), whereas on my D500, I've had great results at ISO6400 and even a cracking set of Jay images at ISO12,800.

    Obviously NR is applied to both, but I find the D500 cleans up much nicer whist retaining much more detail which is smeared out with the EM1 II (even with in camera NR set to Low). I always shoot Raw in both cases and develop in Lightroom CC (now Lightroom CC Classic). Furthermore, I don't really find the GX8 nearly a stop behind the EM1 II, with their noise patterns looking quite similar (maybe a slight nod to the Olympus).

    Now I expect the D500 with it's bigger sensor to do better and it does, but that's not what DXO Mark says ? TBH, whilst everyone raves on about how good the Fuji X-T2 is, whilst it's a great camera, I'd say it's high ISO noise levels appear not much better (if at all) than the EM1 MK II, and certainly behind the D500.

    Now I'm not saying I'm disappointing, as that's what smaller sensor technology gets you, and the size and weight of the M4/3 system over my APS-C Nikon system counter acts nicely the slightly lower ISO performance, but was just really surprised that DXO ranked them about the same ?

    Any one else think the high ISO performance of the EM1 II matches the best of APS-C (which arguably the D500 is) ?
     
    • Agree Agree x 1
  2. ijm5012

    ijm5012 Mu-43 Hall of Famer

    Oct 2, 2013
    Pittsburgh, PA
    Ian
    I too own an E-M1 II and D500, and my findings mimics yours in that the D500's high ISO files are much cleaner than my E-M1 II's. How DXO has come to their conclusion is beyond me, but I agree that it is a bit misleading. I've also found that at low ISOs, the D500 files don't contain as much noise in the shadows as the E-M1 II does.

    As far as the best APS-C sensor goes, I think that title still stays with the D7200, as it edges out the D500 in color depth and dynamic range.
     
    • Agree Agree x 1
  3. bikerhiker

    bikerhiker Mu-43 All-Pro

    Dec 24, 2013
    Canada
    David
    This is why photography is not about numbers, but rather it is about image quality. My friend and I had a discussion about this over at a bubble tea house just yesterday and and he is shooting for top agencies and going to top events still (Olympics etc..) for awhile. He owns and paid in full a D5, D4 an a D3 and he was comparing his D5 against his competition, the EOS 1DX Mk II and the D5 beat the Mark II for at least 3 stops at higher ISOs.

    In the end, it's all about your images and how clean they look to you and to everyone.

    Ultimately, the larger the sensor the better noise performance it is against a smaller sensor of the same generation of sensor technology.
     
  4. ijm5012

    ijm5012 Mu-43 Hall of Famer

    Oct 2, 2013
    Pittsburgh, PA
    Ian
    Agreed. This is why we need to use a bit of common sense as well, rather than blindly trusting test charts with values generate via a method we don't know about. Is it really reasonable to believe that a 20MP m43 sensor would perform just as well as a 20MP APS-C sensor when it comes to noise? Absolutely not, as the pixels on the APS-C sensor are larger, and therefore should do better with regards to high ISO (just like how the 20MP FF sensor in the D5 does better than the 20MP APS-C sensor in the D500).

    I also agree 100% that all that matters when you get down to it is that your paying clients/customers are happy and satisfied with the images. Whether those images were shot with a $6,000 FF DSLR or $1,000 m43 camera, as long as the images are of the needed quality, that's all that matters. Far too often we get sucked in to staring a 1:1 crops on a computer monitor, which is never how a final image is actually viewed. The only benefit you get out of staring at 1:1 crops on a monitor are that you convince yourself that whatever gear it is you're shooting with is an improvement over what you had before and that the cost is justified (or on the flip side, your current gear is inferior and you need that upgrade).
     
    • Agree Agree x 2
    • Winner Winner x 1
  5. Drdave944

    Drdave944 Mu-43 All-Pro Subscribing Member

    Feb 2, 2012
    All these tests are just a starting point. I don't bench my camera when I take pictures. It is almost impossible to set up two cameras with every parameter set the same,except in the lab. It is impossible to do when walking around. I usually take two cameras around and download the results to Lightroom in the same file. The difference in noise,IS and other factors from shot to shot is remarkable. One sensor is putting out 20MP M-43 and the other 36 MP full frame. The shots can't be easily identified on the screen without checking. Some times I say,Darn that Camera X. look at that noise! Then it turns out to be the wrong camera. It is all easily corrected with PP. People invest too much emotion in technical details. Get your camera and get out of the house.
     
    • Agree Agree x 6
  6. DWS

    DWS Mu-43 Regular

    116
    Jun 6, 2014
    I'll take an image that evokes emotion or thought over a technically correct or benchmarked “ideal” image each and every time.
     
    • Agree Agree x 2
    • Winner Winner x 1
  7. Turbofrog

    Turbofrog Mu-43 Hall of Famer

    Mar 21, 2014
    I've got to agree. The DXO disparity between the E-M1 II and all other M4/3 sensors is pretty much entirely unfounded. It makes me concerned that DXO is very easily susceptible to being "gamed." There is evidence that Olympus is doing some "nearest neighbour" signal processing in the RAW data to try and alleviate the influence of the PDAF pixels on image quality, with the effect of having some noise reduction. It's not particularly discernible in the actual images once they're demosaiced, but it shows up in the analysis of the RAW data, and probably makes that data look "better" than it is.

    Due to the lack of transparency in DXO's methodology, we of course can't tell how they actually determine any of their numbers. But given how no one is able to perceive a huge leap in IQ from the E-M1 II, especially not compared to something like the PEN-F or GH5 that have DXO scores 0.5-0.7 stops worse, it suggests funny business.

    I've long noticed inconsistencies in DXO's numbers, so this is yet more evidence to add to the pile. DXO should be regarded as just another source for data, and certainly not a definitive one.
     
    • Agree Agree x 1
  8. whumber

    whumber Mu-43 Regular

    38
    Jun 13, 2017
    The problem lies in how the DxO low light score is calculated. It's the based on the ISO setting where the output satisfies three minimum values based on SNR, DR, and color depth. It's really the color depth that throws a monkey wrench into the process and as a result in general the single value score is meaningless garbage as the OP has seen. As another example, the A7R gets almost the same sports score as the 1DX which is similarly misleading as the A7R is much worse at high ISO than the 1DX. Now with that said, the underlying data that DxO shows in the measurement charts is quite good; I just wouldn't recommend bothering with the distilled scores that they push.
     
    • Agree Agree x 2
  9. PacNWMike

    PacNWMike Mu-43 Hall of Famer

    Dec 5, 2014
    Salish Sea
    "Looking better" is the bottom line. More power to Oly for coaxing out the result in a small sensor and small body. That's not cheating, that's good engineering. up-R.
     
    • Like Like x 1
    • Agree Agree x 1
    • Funny Funny x 1
  10. ijm5012

    ijm5012 Mu-43 Hall of Famer

    Oct 2, 2013
    Pittsburgh, PA
    Ian
    Not if the reported ISO deviates from the true ISO, i.e. the camera is reporting 3200, but the sensitivity is more along the lines of 1600.
     
  11. Turbofrog

    Turbofrog Mu-43 Hall of Famer

    Mar 21, 2014
    Sorry, when I say making the data "look" better, I mean to the algorithms that DXO uses to analyse the RAW ones and zeros.

    Not "look" better to the human eye. Which is where the GH5 and E-M1 II are on even footing. Exactly as you'd expect.
     
  12. Gregory

    Gregory Mu-43 Regular

    97
    Mar 4, 2017
    So around 500 points on the DxO scale constitutes 1 stop of performance? Is there an exact formula to convert from points to stops?
     
  13. whumber

    whumber Mu-43 Regular

    38
    Jun 13, 2017
    It's a base-2 log scale, so a one stop increase from 806 would be 1612. The difference between the E-M1ii and GX8, as indicated by the sports score, is log2(1312/806) = 0.7 stops.
     
    • Agree Agree x 1
    • Informative Informative x 1
    • Appreciate Appreciate x 1
  14. PeHa

    PeHa Mu-43 Veteran Subscribing Member

    371
    Nov 12, 2013
    Sweden
    Last edited: Nov 13, 2017 at 5:42 PM
  15. Gregory

    Gregory Mu-43 Regular

    97
    Mar 4, 2017
    I've not noticed any ads on the dxomark site. Can you point some out?
     
  16. Machi

    Machi Mu-43 Veteran

    324
    May 23, 2015
    DXOMark results are in general valid but one must look at their measurements and not their dumbed down numbers.
    D500 is better or at least the same as the E-M1II in all measured parameters. So your experiences are not surprising.
    But those parameters are for sensor only! In practice interaction between camera (sensor) and lens also plays important role.
     
  17. Clint

    Clint Mu-43 All-Pro Subscribing Member

    Apr 22, 2013
    San Diego area, CA
    Clint
    Develop like photos for your D500 and EM-1Mark II to the best you can for each of them. Send the files to a printer, say maybe for 16 x 20 prints. Now what is the difference?

    Looking at the numbers doesn't do much good - who else than some photographers are going to say this is better than that base on numbers? Looking on the screen and pixel peeping tells you nothing about how the photos will actually look in real life. Who is going to be looking at your photos at 1:1 on a screen.
     
    • Winner Winner x 1
  18. whumber

    whumber Mu-43 Regular

    38
    Jun 13, 2017
    But looking at photos on a screen is by far the most common way of viewing photos now. Most people very much like the option to view photos at 100% to view all the fine detail which is made much more accessible when viewing on a screen vs making prints where you would need to use a loupe to achieve the same effect.
     
  19. ijm5012

    ijm5012 Mu-43 Hall of Famer

    Oct 2, 2013
    Pittsburgh, PA
    Ian
    Sorry mate, but last time I checked, people weren't viewing 1:1 crops of my photos on Facebook and Instagram.

    Plus, if that's your true destination for the image, then go buy a cheap bridge camera to shoot with, because at the max 2048 pixel resolution that Facebook supports, you don't need anything more than a 4MP camera, and all of the 100% goodness of your 40+ MP FF cameras is going to be lost on your viewers.

    For those who actually print photos (selling prints from a wedding, selling landscape shots, displaying images in a showcase), all of that is still done via print. If you're just going to be posting stuff on social media, then there's no need for an ILC.
     
    • Agree Agree x 1
  20. whumber

    whumber Mu-43 Regular

    38
    Jun 13, 2017
    I actually find it kind of hilarious that you posted this with a link to your Flickr page in your signature.