Our new Dynamic Range measurement system involves shooting a calibrated Stouffer Step Wedge (13 stops total range) which is backlit using a daylight balanced lamp (98 CRI). A single shot of this produces a gray scale wedge from (the cameras) black to clipped white (example below). Each step of the scale is equivalent to 1/3 EV (a third of a stop), we select one step as 'middle gray' and measure outwards to define the dynamic range. Hence there are 'two sides' to our results, the amount of shadow range (below middle gray) and the amount of highlight range (above middle gray).
To most people highlight range is the first thing they think about when talking about dynamic range, that is the amount of highlight detail the camera can capture before it clips to white. Shadow range is more complicated, in our test we stop measuring values below middle gray as soon as the luminance value drops below our defined 'black point' (about 2% luminance) or the signal-to-noise ratio drops below a predefined value (where shadow detail would be swamped by noise), whichever comes first.
The E-3's gradation setting produces images in low or high keys, it does so by modifying the tone curve as well as the exposure (around a third of a stop under for Low Key and a third of a stop over for High Key). To show the difference in tone curve more clearly for this graph we adjusted the exposure manually to match the middle grey as closely as possible. As you can see the Low key pulls down the shadows a little, whereas the high key pushes up the highlights a little. Overall dynamic range remains the same whichever setting you use. The auto mode does some serious lifting of the shadows and introduces fairly unpleasant noise issues (we'd advise shooting raw if you want to do this kind of tonal adjustment).
Here we have tested the extremes of the contrast setting from -2 to +2. As you can see the -2 setting does indeed deliver quite a bit more shadow range (because it lifts steps above our cut-off point) but does not extend highlight range (which clips at around +3.0 EV, the same as the default contrast setting).
ISO Sensitivity and Dynamic Range
Our biggest complaint about the last few Olympus E-system SLRs has been the narrower than average dynamic range, especially (though not exclusively) in JPEGs. We were pleased to see that the E-3 does a far better job, producing JPEGs with up to 9 stops of dynamic range (at ISO 200, which may indicate that, like many other cameras, this is actually the 'optimum' setting). But it's not all good news. The extra DR lies almost exclusively in the shadow area; you're getting maybe a third of a stop more highlight range with the E-3 than you'd get with, say, the E-510. This is an important improvement, but it's far from class-leading.
|Sensitivity||Shadow range||Highlight range||Usable range|
|ISO 100||-5.8 EV||3.0 EV||8.8 EV|
|ISO 200||-5.9 EV||3.1 EV||9.0 EV|
|ISO 400||-5.9 EV||2.9 EV||8.8 EV|
|ISO 800||-5.1 EV||2.8 EV||7.9 EV|
|ISO 1600||-4.3 EV||2.7 EV||7.0 EV|
|ISO 3200||-3.7 EV||2.8 EV||6.5 EV|
Dynamic Range compared
The graphs below show that although the E-3 does well with shadow detail and offers a similar overall dynamic range to its nearest competitors, it falls short - up to 1.0EV short - of the best in class when it comes to the all-important highlight region (using the default, metered exposure). As mentioned above, the E-3 is a distinct improvement on the E-510 (and narrowed the gap considerably), there is still a way to go before it can match the best cameras in this class when it comes to capturing the brightest highlights.
The wedges below are created by our measurement system from the values read from the step wedge, the red lines indicate approximate shadow and highlight range (the dotted line indicating middle gray).
Experience tells us that there is typically around 1 EV (one stop) of extra information available at the highlight end in RAW files and that a negative digital exposure compensation when converting such files can recover detail lost to over-exposure.
The E-3 is unusual in that there isn't a great deal of headroom in the raw files; you can pull back the highlights (by applying negative exposure compensation) but in reality the results are rarely worth the effort; you still get around 9 EV of dynamic range, and there's very little extra information at the highlight end (we'd guess that maybe a third of a stop would be a fair assessment).
Whilst this limits the usefulness of raw for dealing with exposure (specifically over-exposure) issues it does at least mean that Olympus is putting almost all the dynamic range the sensor is capturing into its JPEG files, which is unusual.
- ACR Default: Exp. 0.0 EV, Blacks 5, Contrast +25, Curve Medium
- ACR Best: Exp. -2.0 EV, Blacks 0, Contrast -50, Curve Linear
WARNING: When looking at the graph above it's important to bear in mind is that although ACR was able to retrieve the 'luminance' (brightness) of wedge steps which were previously clipped there's no guarantee of color accuracy as individual channels may clip before others (see below).
As can be seen in the examples below it is possible to pull back a little of the highlight detail that is clipped in JPEGs (and default raw conversions), but doing so tends to result in posterization and color issues. There is virtually no color information in this extra highlight range. As the lower example shows, if an E-3 JPEG looks like the highlights are clipped, they probably are clipped.
|ACR default conversion||ACR with -2.0 EV digital exp. comp.|
|ACR default conversion||ACR with -3.0 EV digital exp. comp.|
Using exposure to retain highlights
There's been a huge amount of discussion on our forums about the E-3's dynamic range and how the results above show no disadvantage because you can, theoretically, just expose a little 'to the left' (under expose) and make full use of the sensor's dynamic range.
Whilst this is indeed true in theory, there are some limits to the practicality of such an approach. The tone curve used by Olympus in the E-3 appears to be tailored not only towards getting the maximum possible total range out of the sensor (measuring total dynamic range in the E-3 raw files using a linear development indicates it's not that different to most of its competitors), but also to controlling noise.
In essence this means that E-3 is, slightly, tending to expose for the shadows, rather than using the tone curve to lift them (which would mean noisier results even at ISO 100). You can override this behaviour as long as you're prepared to do a little work in the raw conversion (and change your exposure settings).
The graphs below tell the story perfectly. First we have the default JPEG output of the E-3. This matches our experience of the usable range in 'real world' shots, with a slightly compressed range in the highlights and a surprisingly extended one in the shadows.
Then we have the Nikon D300's default JPEG curve (this is included simply for comparison as one of the better APS-C cameras). Obviously you can get a little more out of the 14-bit RAW files than the standard JPEGs shown here.
Finally we shot the E-3 in raw mode, under exposing by around 0.7 EV (2/3 stop). During the raw conversion process we included a +0.7 EV exposure increase (to bring the exposure back) and made a slight adjustment to the tone curve to lift the shadows and mid-tones to match as closely as possible the default tone curve (which is very similar for all cameras). Note that we tried both ACR and Olympus Studio 2.0 for the raw conversion and for the purposes of this test (dynamic range) the results were identical, though they will obviously depend on how much noise reduction you apply (which will also have an effect on the amount of detail you see in shadow areas).
This is the result:
*1: Exposure reduced by 0.7 EV, ACR raw development, +0.7 EV and curve adjustment to restore midtones.
As you can see this process does indeed 'bring back' the highlights and reduce clipping (as would be expected). There is, however, an overall reduction in dynamic range (to around 7.5 EV total) - mainly because the shadows end up very noisy (visibly so), even at ISO 100. With careful processing of raw files you can fine tune the output to close the gap between the E-3 and the other cameras in this class.
The difference between the E-3 and its competitors isn't stark (and of course you may be happy to live with a little extra noise in the shadows), but a difference there is. These tests prove that you can overcome E-3 highlight clipping by careful use of exposure (and raw mode), but doing so doesn't quite produce results that match, for example, the D300's default JPEGs. Note however that this is mainly limited by noise, and it's up to the user to decide how acceptable a compromise this represents.
We also tried to get a better result using the E-3's shadow adjustment technology (SAT). There is no dynamic range benefit to using SAT on a correctly exposed shot (the highlights are still at risk of clipping). Underexposing the shot with SAT turned on then using Photoshop's curves to sort out the midtones, does increase dynamic range and is the best option for getting the maximum range when shooting JPEGs. Obviously all the techniques discussed here for extending dynamic range are equally applicable to some extent or another with all SLR cameras, particularly when shooting raw.