In perhaps the most anticipated and overdue sensor test of all time, DxOmark has finally tested the Olympus OM-D E-M5 sensor. We may never know why it took this long, but I'm thankful they decided to publish it after all this time.
Link: DxOMark - Olympus OM-D E-M5: The best of the micro 4:3 cameras
Those of us who have been using the OM-D E-M5 for a while won't be surprised by the DxOmark test results. Keep in mind that DxOmark reports have some limitations. Most importantly, they don't give you a sense of how the numbers and charts translate into real life results. For example, they tell you that the E-M5 has 1.7EV more dynamic range than a Panasonic G3 or GX1 (at base ISO), but you need additional experience or examples to show you how a 1.7EV difference affects pictures
Another thing I'd like to clarify is that when DxOmark talks about "manufacturer ISO" (aka nominal ISO) vs "measured ISO", people often think DxOmark's measured ISO is the "real" ISO while any deviation from this by the manufacturer is "overrating" (manufacturer ISO > measured) or "underrated" (manufacturer ISO < measured). Sometimes people go as far as to accuse manufacturers of "cheating", ie purposely misrating their ISO values to do better in tests. All of this is misunderstanding.
To simplify things, manufacturers all specify an ISO based on the brightness that a JPEG will display for a given exposure (determined by shutter speed, aperture, and lighting conditions). RAW processors like Adobe Lightroom will automatically apply a similar tone curve to the in-camera processor in order to get the default conversion to the same overall brightness level as the in-camera JPEG. So for a given lighting condition, a Panasonic picture at f/2.8, 1/100s, ISO 400 will look as bright as an Olympus picture taken at f/2.8, 1/100s, ISO 400.
Olympus tends to want more highlight headroom in the files at the expense of shadow latitude, so what they call ISO 400 usually corresponds to a lower level of analog gain (sensitivity) than what Panasonic calls ISO 400. This is why DxOmark says that Olympus ISO 400 is measured ISO 214. In order to give you the same final brightness, the Olympus file gets "pushed" by the in-camera tone curve (or Lightroom, etc) more than the Panasonic file gets "pushed" by its tone curve. In other words, an Olympus camera will choose a faster shutter speed in order to preserve the highlights, push the file enough to get the same overall brightness (as a Panasonic file with a longer exposure), and as a result the Olympus RAW file will have noisier shadows and more more highlight detail than a Panasonic using the same sensor.
Depending on the testing methods used, this may result in the Olympus camera being disadvantaged during noise testing, exactly the opposite of what people usually say when they accuse Olympus of "cheating" by "overrating" their ISO
. It may also result in a false advantage for highlight testing, and will certainly result in a false advantage for most DR measurements based on in-camera JPEG testing (most review outfits).
DxOmark uses an arbitrary definition of ISO which relates the camera sensitivity to the amount of highlight headroom for that given sensitivity. It's a great way for them to standardize results when testing many different cameras from different manufacturers. However, DxOmark's definition of ISO (measured ISO) is non-standard and actually technically less correct than the definition which the manufacturers use (which corresponds to one of the standard definitions of ISO
So what can we take away from the DxOmark results?
First, Micro 4/3 as a system has closed the dynamic range gap with APS-C cameras.
The latest Sony NEX cameras maintain a slim margin advantage in this aspect (less than half a stop for the NEX-5N), while the E-M5 (and by extension the E-PL5, E-PM2, and probably GH3) sensor's DR now exceeds that of any previously measured APS-C sensor from a Canon or Samsung camera.
Second, the Micro 4/3 system is now punching above its weight in terms of low light high ISO performance
. The E-M5 sensor tested only 0.3 stops worse than the NEX cameras in this category, whereas one would expect a 0.7 stop disadvantage based on format size. In other words, it would take a stabilized 33mm f/1.6 lens on a Sony NEX body to match the low light capability of the Panasonic Leica 25mm f/1.4 lens on a current Olympus body, or a stabilized f/3.2 zoom to match the low light capability of the Panasonic 12-35mm zoom on the GH3 (assuming that camera uses the same sensor).
These are small differences. The bottom line is that the E-M5 sensor is a step forward for the system. Whether these sensor advances help us make substantially better photos is an entirely separate discussion. I'd argue that for most people under most circumstances, the answer is no.