Studio Tests - DX format
The Sigma 70-200mm performs generally well on DX, and is strongest towards the short end of the zoom range; however it can't match the Nikon 70-200mm F2.8 AF-S VR's exceptional sharpness on the smaller sensor format. As usual, it benefits from the 'sweet spot' advantages of low distortion and minimal vignetting which are common to shooting full-frame lenses on DX.
|Sharpness||Sharpness results are best towards the middle of the zoom range (around 100mm), and least impressive at the long end. As expected the optimal apertures tend to be F5.6-F8, however the lens also performs creditably wide open, especially at shorter focal lengths.|
|Chromatic Aberration||Chromatic aberration is minimal at 70mm, but then increases progressively towards the long end. At 200mm it's the highest we've measured on a 70-200mm zoom, and a clear issue for real-world shooting.|
|Falloff||We consider falloff to become perceptible when the corner illumination falls to more than 1 stop less than the centre. As usual for a full-frame lens used on DX, there's nothing to worry about here.|
|Distortion||Distortion levels are low, from 0.48% barrel at 70mm, through neutral at around 85mm to 100mm, to -0.78% pincushion at 200mm. This is highly unlikely to be noticeable in normal use.|
Focus shift, curvature of field, and lens testing
We test lenses by establishing the optimum focus with the aperture wide open, then stopping down progressively without refocusing; this is designed to represent the results you can expect from a lens properly calibrated to give optimum AF results wide open (which we presume most users would prefer, especially with a fast telezoom). However during testing we identified that this lens is subject to a distinct focus shift on stopping down (which technically is a symptom of spherical aberration); this is very obvious at closest focus, but also has some effect on our test chart shots, which are taken at focus distances between 2.3m and 10m. The lens also exhibits curvature of field, which means that the region of optimum sharpness is slightly bowl-shaped, with the corners curving towards the camera.
The effect of this on our chart test data is most visible at the shorter end of the zoom range, with measured central sharpness actually falling somewhat on stopping down - this indicates that the region of sharpest focus has moved away from the camera, and is now slightly behind the plane of the chart. However due to curvature of field, the corners of the chart move into the region of sharp focus on stopping down. At 70mm and F8, for example, the corners therefore measure as being distinctly sharper than the centre. The practical upshot is that our test chart data actually slightly under-represents this lens's potential sharpness at intermediate apertures, although at F11 and smaller, the twin effects of increased depth of field and diffraction softening re-equalise the data.
However this focus shift has deeper implications than just test chart shooting; it means that focusing the lens to give absolute pixel level sharpness is necessarily inconsistent at short distances. Quite simply, if the lens needs to be focused differently according to the shooting aperture, the AF system won't be able to cope, and will appear either to front-focus at wide apertures, or rear-focus when stopped down, depending upon calibration (our sample clearly front-focused at F2.8 and 1m, but focused accurately at longer distances). Indeed this is likely to be the primary reason for reports on our forums of front-focusing at 200mm, as such tests are invariably conducted at short distances; however in this case the results probably give a false impression of the performance of the lens for more general use, as the effects on middle- or long-distance shooting appear to be insignificant in practical terms. But this focus shift genuinely has a major impact on close-up shooting, and this is explored in more detail later in this review.
Extrapolation of test results to Four Thirds Format
The Sigma 70-200mm F2.8 EX DG Macro HSM II will be available in Four Thirds mount, and it's possible to predict its expected performance on this system based upon the MTF50 data for the DX format. The Four Thirds frame (17.3mm x 13mm) has approximately 82% of the height of the DX frame (23.6mm x 15.8 mm), and the diagonal is approximately 75% of its length. Therefore, the expected data for Four Thirds can be attained by multiplying the MTF50 data (y-axis) by 0.82, whilst considering only the data out to 75% of the frame diagonal (i.e. the x-axis of the graph).
Overall this suggests that the 70-200mm F2.8 will be a reasonably competent performer on Four Thirds, although distinctly soft at 200mm; it also seems safe to assume that a lens of similar range but specifically designed for the smaller sensor (as opposed to 35mm full-frame) will generally be a better bet.
Third party lenses have something of a reputation for less-than-neutral colour balance, with Sigmas in particular subject to widespread anecdotal reports of a distinct yellow cast, so in this test we measure any colour cast introduced by the lens in comparison to the camera manufacturer's 50mm lens (generally considered a good standard for neutrality).
In this test, the camera is pointed towards an evenly illuminated white wall, and light entering the lens completely diffused using an 'Expodisc' white balance filter. A custom white balance is taken using the 50mm lens (in this case the Nikon 50mm F1.4D), then exposures made using the 50mm lens and the lens under test. The RGB values from the centre of the frame are reported (measured as an 11x11 average).
|Nikon 50mm F1.4D
(212, 212, 212)
|Sigma 70-200mm F2.8 DG Macro HSM II
(223, 222, 219)
Here we can see that the Sigma 70-200mm shows a slight yellow shift when compared to the Nikon 50mm F1.4D. However whilst this is indeed measurable, it's scarcely visible (indeed in this comparison it's dwarfed by a difference in exposure between the two lenses, even when using the same shutter speed and aperture), and if necessary it could easily be corrected using a custom white balance.
Specific image quality issuesAs always, our studio tests are backed up by taking hundreds of photographs with the lens across a range of subjects, and examining them in detail. This allows us to confirm our studio observations, and identify any other issues which don't show up in the tests.
Softness at longer focal lengths
Our test data predicts that the lens will look a little soft on DX at longer focal lengths, and this is borne out in real-world shooting. The two image samples below represent how this appears in practice; at 135mm, central resolution is high but the corners are soft even at F8, whereas at 200mm the whole image can look slightly soft, especially when shooting with the aperture wide open.
|F8, Nikon D300||F2.8, Nikon D300|
|100% crop, centre of frame||100% crop, centre of frame|
|100% crop, top right corner||100% crop, top right corner|