Raw Dynamic Range

Editor's note: This content was originally published before we had access to Canon's EOS-1D X Mark II. We are currently working on an updated comparison.

After running our Exposure Latitude and ISO Invariance tests we've found that while the D5's sensor is clearly optimized for low-light image making, it represents a rather surprising - and disappointing - step backwards in terms of low ISO dynamic range.

Let's take a closer look, starting with our Exposure Latitude test scene.

Exposure Latitude

In this test we look to see how tolerant of pushing exposure the D5's Raw files are. We've done this by exposing our scene with increasingly lower exposures, then pushed them back to the correct brightness using Adobe Camera Raw. Examining what happens in the shadows allows you to assess the exposure latitude (essentially the dynamic range) of the Raw files.

Because the changes in this test noise are primarily caused by shot noise and this is mainly determined by the amount of light the camera has had access to, the results are only directly comparable between cameras of the same sensor size. However, this will also be the case in real-world shooting if you're limited by what shutter speed you can keep steady, so this test gives you an idea of the amount of processing latitude different formats give.

Even at a modest 3 stop push, we see the D5 has already contributed a considerable amount of noise to its low ISO images compared to the 24MP D750. After a 5 stop push the gap widens, with the D750's 5 stop file looking more like the 3 stop file from the D5. In fact, the D5's performance closely resembles the performance seen from the Canon 6D, with just a bit more color noise showing in the 6D's file.

After very aggressive pushes, an interesting series of bands appears at the top and bottom of the image. We don't know what's causing this additional noise (which appears to be confined to specific read-out rows) but it's interesting to note.

ISO Invariance

A camera with a very low noise floor is able to capture a large amount of dynamic range, since it adds very little noise to the detail captured in the shadow regions of the image. This has an interesting implication: it minimizes the need to amplify the sensor's signal in order to keep it above that noise floor (which is what ISO amplification conventionally does). This provides an alternate way of working in situations that would traditionally demand higher ISO settings.

Here we've done something that may seem counter-intuitive: we've used the same aperture and shutter speed at different ISO settings to see how much difference there is between shooting at a particular ISO setting (and using hardware amplification) vs. digitally correcting the brightness, later. This has the advantage that all the shots should exhibit the same shot noise and any differences must have been contributed by the camera's circuitry.

To make a long story short, the D5 (unlike the current Nikon benchmark camera, the D810) is not ISO Invariant. We see a similar performance between ISO 6400 and the pushed ISO 1600 and 800 files, but the ISO 400 image reveals a noise penalty - the darker tones in the image haven't been amplified above the camera's noise floor, so get swamped by noise. Comparing to the D810 gives us a clearer idea of how an ISO invariant camera should perform: the noise levels stay similar regardless of whether an image was shot at ISO 6400, or pushed to that level.

This shows that the D5 is adding some downstream read noise, which limits low ISO dynamic range but doesn't much affect high ISO dynamic range (which our friend Bill Claff's analysis shows to be excellent, if not class-leading). The D5's performance is, once again, closer to what we have seen in Canon cameras, such as the EOS 6D.* The Canon does add a little more noise than the D5 as well as introducing slight banding which can be hard to clean up. This limits the ability of the D5 to shoot at a low ISO setting in low light (to protect highlight information), rather than pushing up the ISO amplification.

A quick note:

Inevitably, some of you will be disappointed by the results that we've shown in this testing. And so are we, a little bit. Our understanding of it is this: The D5 - like the D4 and D3-series before it - is Nikon's flagship camera by virtue of certain aspects of its capability. Specifically things like speed, AF performance, durability, and the more esoteric aspects of performance and specification valued very much by the working professionals it's aimed at. High ISO image quality, especially, has been a point of pride for Nikon in its flagship DSLRs since the D3, and as we'd expect, the D5 raises this already ridiculously high bar even further.

We wonder whether the need to maintain and build on these capabilities has led Nikon to continue the development of the sensors it designed for these previous Dx series cameras, perhaps suggesting that there wasn't a Sony-designed sensor available that would offer the high-speed readout and high ISO performance that Nikon required. Alternatively, it could be a matter of pride that the company's flagship offering be built around a sensor developed in-house.

Either way, in our opinion, we'd try not to over-stress the importance of the fact that the D5 has poorer base ISO dynamic range than its current peers (after all, you can buy multiple D810s for the same price, if low ISO DR is important to you). For its intended audience, the D5's high ISO imaging capabilities, advanced autofocus and durability are likely to be much more important.


* Interestingly, a sinusoidal dynamic range vs. ISO curve is reminiscent of older designs, distinctly different from the performance of more recent cameras we've come to expect (note the D810's rather linear drop with ISO). In some older designs, this sinusoidal nature was due to a dual amplification architecture - which was far from ideal when it came to dynamic range, as it means that intermediate ISOs on the lower end are less than ideal.