We've reported in recent years how Canon's newer sensor designs have started to close the dynamic range gap, compared with chips from the likes of Sony and Toshiba. Dynamic range isn't everything, of course: Canon's Dual Pixel sensors have brought advances in live view and video autofocus that for many people will be every bit as significant as the noticeable shortfall in Raw file malleability.
But it was promising to see Canon getting competitive in an area where it had fallen behind.
Sadly though, it seems the benefits that appeared in the sensors used in the EOS 80D and EOS 5D IV have not been applied to the latest EOS 6D II, and the new camera has less dynamic range than we've become used to. Graphs plotted by regular DPR collaborator Bill Claff illustrate this pretty clearly. In this article, we're taking a look at what this might mean for your images.
Dynamic range assessment
Our exposure latitude test shows what happens if you brighten a series of increasingly dark set of exposures. This illustrates what happens if you try to pull detail out of the shadows of your image.
$(document). ready(function() { ImageComparisonWidget({"containerId":"reviewImageComparisonWidget-57046185","widgetId":542,"initialStateId":3691}) })
As you can see, the EOS 6D II begins to look noisy much sooner than the broadly comparable Nikon D750, meaning you have less processing flexibility before noise starts to detract from your images.
The EOS 6D II should have a 1. 3EV image quality advantage over the 80D, when the images are compared at the same size, since its sensor is so much bigger. Despite this, the EOS 80D's$(document). ready(function() { $("#icl-3692--1019644042"). click(function() { ImageComparisonWidgetLink(3692); }); }) images shot with the same exposures look cleaner, when brightened to the same degree. Have a look and you'll see the difference is around 1EV$(document). ready(function() { $("#icl-3693--487818319"). click(function() { ImageComparisonWidgetLink(3693); }); }), despite the head start that the 6D II's chip should have. This corroborates what Bill Claff's data suggests.
ISO Invariance
The downside of our exposure latitude test is that reducing the exposure also increases the noise. Our ISO Invariance test uses the same exposure shot at different ISO settings, such that the shot noise contribution is the same in each image. This way any differences must be a consequence of electronic noise (and how well the camera's amplification overcomes it, at higher ISO settings).
$(document). ready(function() { ImageComparisonWidget({"containerId":"reviewImageComparisonWidget-58725857","widgetId":541,"initialStateId":null}) })
This isn't good, especially not by modern standards. We're used to seeing sensors that add so little noise$(document). ready(function() { $("#icl-3694--1065892121"). click(function() { ImageComparisonWidgetLink(3694); }); }) that there's barely any visual difference between shooting at a high ISO and using a low ISO (retaining additional highlights) then brightening. Instead we see that you have to amplify to around ISO 3200$(document). ready(function() { $("#icl-3695--808510428"). click(function() { ImageComparisonWidgetLink(3695); }); }) before you see no additional impact from the camera's electronics. This suggests a reversion to the level of the original EOS 6D$(document). ready(function() { $("#icl-3690-1001550611"). click(function() { ImageComparisonWidgetLink(3690); }); }), even perhaps a bit worse. That's unfortunate for those shooting high contrast scenes or anyone who suffers from Canon's metering decision to underexpose in backlit situations, as exposure latitude will be limited.
Real world impact
If you shoot JPEG, you'll never notice any of this, since the differences occur beyond the ~8. 3EV or so that tend to be incorporated into a typical image. Similarly, at higher ISO settings, amplification overcomes the electronic noise, so you see the camera begin to out-perform the 80D and then close the gap with the D750, just as Bill's chart suggests.
However, it means if you're processing from Raw at low ISOs, you have much less flexibility in terms of what you can do with the file than we'd expect from a modern camera. Almost as soon as you start to push the image or pull detail out of the shadows, you risk hitting the camera's electronic noise floor and hence you won't see the advantage over the smaller sensor 80D that you might reasonably expect.
Adding to the problem is Canon's metering system, which tends to underexpose images when there's strong back-light. If the metering sensor were high resolution or advanced enough to detect faces, one might expect proper exposures for human subjects even in backlit shots; however, we've found the low resolution metering sensor in the 6D Mark II to be often incapable of detecting - and properly exposing for - faces. That means that backlit shots will be underexposed (unless you intervene), and you'll have limited ability to recover these underexposed shots because of the sensor's poor performance.
Canon EOS 6D Mark II | EF 35mm F2 IS | ISO 100| F9 | 1/200th Shadows lifted, highlights lowered, slight selective brightening to couples' faces. As you'll see if you click to view the full-sized image, noise in the areas of lifted shadow is very apparent.
This is an extreme example but it's a photo I'd expect to be able to shoot on other full frame cameras without revealing so much noise. All of our test results suggest I could have achieved just as good a result from a contemporary APS-C camera, if not better if it were also capable of properly detecting human subjects and exposing for them.
. dpreview.com2017-7-20 07:00