The Sigma 17-70mm performs generally pretty well in studio tests. It's weakest at wideangle, with somewhat soft edges and corners that never fully sharpen up on stopping down. But aside from that it's an admirably consistent performer, with good cross-frame sharpness and reasonably low levels of chromatic aberration, distortion and vignetting.
|Sharpness||Sharpness results are generally pretty good. The lens is weakest at 17mm, where the edges and corners are rather soft, and don't sharpen up at any aperture. But at longer focal lengths it's very well-behaved, with decent sharpness wide open, and excellent results at its optimum apertures around F5.6-F8.|
|Chromatic Aberration||Chromatic aberration is kept reasonably low. It's worst at wideangle, where there's pretty strong green/magenta fringing towards the edge and corner of the frame. It decreases on zooming in, and is very low around 35-50mm, before red/cyan fringing becomes visible at 70mm.|
|Vignetting||Vignetting is overall rather low, given the lens's maximum aperture and compact size. It's strongest at wideangle, with 1 stop light falloff in the corners wide open, dropping down to 0.6 stop at F4. But in practical use this counts as nothing to worry about.|
|Distortion||Distortion is kept pretty well under control. There's clear barrel distortion at wideangle, but it's no worse than a typical 18-55mm, and much better than most lenses with extended zoom ranges. This disappears at 24mm, before turning to moderate pincushion distortion from 35-70mm.|
As the 'Macro' label suggests, the Sigma 17-70mm offers good close focusing specs, on paper at least. But the reality is a little more complicated, as the working distance from the front of lens to the subject is extremely short. Coupled with the relatively large diameter front of the lens, this causes all sorts of problems with lighting and disturbing your subject. The internal focus design means that the angle of view widens quite dramatically at minimum focus too, to an effective focal length that's much closer to 50mm, so the lens doesn't 'feel' like a moderate telephoto any more at all.
Here we're looking at what we'd consider to be about the closest practical focusing distance, where the shadowing by the front of the lens generally isn't too severe, and the camera can still confirm autofocus in live view. At this point the focus distance is 0.21m, slightly shorter than Sigma's specified 0.22m, and the working distance just a shade over 4cm. You can get even closer in manual focus.
With the lens set to F4 the image is pretty soft across the beard, but central sharpness improves dramatically at F5.6. However the corners are still soft in this flat test chart shot, and only properly sharpen up at F16. There's also quite strong barrel distortion, and strong blue/yellow fringing towards the corners from lateral chromatic aberration. In context, 18-55mm kit zooms generally offer similar coverage, but with lower distortion and a more practical working distance.
Full Frame Coverage
The Canon, Nikon and Sony mount versions of this lens will mount on full-frame DSLRs, and on Nikon cameras DX crop mode will be automatically selected (the camera will therefore shoot at reduced resolution). The lens's image circle doesn't cover the 35mm full frame format at any focal length, giving severe vignetting regardless of focal length, aperture, or focus distance. So this really isn't a lens you can sensibly share across SLRs of different formats.
The 17-70mm features Sigma's own 'Optical Stabilization' system, and the company claims that it allows hand-holding at shutter speeds up to four stop slower than usual without seeing the blurring effects of camera shake. The mechanism is silent when operational, with only the stabilization of the viewfinder image betraying the fact that it's running.
To determine the effectiveness of the OS system we subjected the 17-70mm to our studio image stabilization test, using the wideangle and telephoto settings. The subject distance for these tests was approximately 3m at 18mm, and 4m at 70mm; the test camera was the Canon EOS 700D.
We take 10 shots at each shutter speed and visually rate them for sharpness. Shots considered 'sharp' have no visible blur at the pixel level, and are therefore suitable for viewing or printing at the largest sizes, whereas files with 'mild blur' are only slightly soft, and perfectly usable for all but the most critical applications.
|17mm OS OFF||70mm OS OFF|
|17mm OS ON||70mm OS ON|
The 17-70mm fares respectably well in these tests, although it doesn't quite match the best optical stabilisation we've seen. At wideangle it offers a solid three stops of stabilisation, allowing hand-holding at shutter speeds as low as 1/4 sec, rather than 1/30 sec with OS off. At 70mm things are a bit more complicated; we're able to get a decent proportion of sharp shots at shutter speeds 4 stops slower with OS turned on (1/10 vs 1/160 sec ), but the system is never close to 100% effective, so it pays to take multiple shots in marginal conditions.
Real world examples
The examples below should give you an idea of how well the Sigma's OS system behaves in everyday shooting; in both cases the image would be hopelessly blurred without stabilization. But instead we've been able to take advantage of the ability to hand-hold at slower shutter speeds to get shots that would otherwise be difficult.
In the first example, the camera was shot in live view, held high up to shoot over the top of a crowd of people to capture the floodlit dome of St Paul's just after sunset. This is a notoriously unstable shooting position, but even so, of multiple shots at shutter speeds around 1/15 sec, half came out perfectly sharp. In the second example, the use of a shutter speed about three stops lower that could otherwise be handheld has allowed the use of ISO 1600 rather than 12800, and the image still retains a decent amount of detail even in the camera's JPEG.
37mm, Canon EOS 100D
45mm, Canon EOS 100D
|1/15 sec, F5.6, ISO 400||1/6 sec, F4, ISO 1600|
|100% crop||100% crop|