The Sigma 18-200mm F3.5-6.3 OS HSM gives distinctly mixed results in our studio tests, and more so than other superzooms. A tendency for soft corners at most settings is balanced by relatively well-controlled chromatic aberration and distortion. Sharpness is best at wideangle (18mm) to short telephoto (50mm), and quite acceptable at full tele, but distinctly unimpressive in the intervening range (80mm - 135mm). Overall performance is therefore somewhat less consistent than any of its direct competitors, the Canon EF-S 18-200mm F3.5-5.6 IS, the Nikon AF-S 18-200m F3.5-5.6G DX VR, or the Tamron 18-270mm F3.5-6.3 Di-II VC.
|Sharpness||Results are very much a mixed bag, with an overall signature of good central sharpness but soft corners. However sharpness also varies wildly through the range; it's best in the wide to short-tele range (and particularly at 18mm and 50mm), but then falls apart completely at mid - telephoto (80mm - 135mm), before rallying to a quite respectable performance at 200mm. As expected, best results are generally obtained at apertures around F8-11 (although there's actually little to be gained by stopping down at 200mm).|
|Chromatic Aberration||Chromatic aberration, or rather a comparative lack of it, is one of this lens's strengths. CA is generally kept pretty low, especially in the more visually disturbing red channel, and is only likely to be noticeable at the two extremes of the zoom range. At 18mm there's red/blue fringing at a level fairly average for this type of lens, and at 200mm some green/magenta fringing (although noticeably less than on competing lenses). All in all pretty good for a superzoom.|
|Falloff||We consider falloff to start becoming a potential problem when the corner illumination falls to more than 1 stop below the center. Falloff is generally low and unlikely ever to be photographically problematic; it measures 1 1/3 stops at 18mm F3.5 (essentially disappearing on stopping down to F5.6), and about 1 stop wide-open at the telephoto end of the range.|
|Distortion||Sigma has managed to keep distortion relatively low for this class of lens. Barrel distortion at wideangle measures 1.8%, however this is a complex 'wave' distortion with substantial corner re-correction, of a type that is relatively difficult to correct in software if desired. At longer focal lengths the pattern changes to pincushion, reaching a maximum of about -2% in the 35-50mm range. Overall, this is rather less extreme than we've seen on other superzooms.|
|Maximum magnification is 0.32x, achieved at 200mm and a closest focus distance of about 35.5cm (much closer than Sigma's stated 45cm), giving a slightly tight working distance of 14.5cm from the subject to the front of the lens.
Image quality is pretty good; central sharpness is high even wide open, and only improves slightly on stopping down. The corners of our test chart image are however distinctly soft at F6.3, but sharpen up substantially at F16. There's slight barrel distortion and visible chromatic aberration.
|Macro - 75 x 50 mm coverage
Distortion: Mild barrel
Corner softness: Poor
Focal length: 200mm (300mm equiv)
FX (Full Frame) Coverage
Both the Nikon and Canon mount versions of this lens will mount on full-frame DSLRs; on Nikon cameras (D3, D3X, D700) DX crop mode will be automatically selected (and the camera will therefore shoot at reduced resolution). The lens's image circle doesn't cover the 35mm full frame format properly at any focal length, giving severe vignetting; this effect is very obvious and well-defined at 18mm, but gets increasing diffuse at longer focal lengths. At 200mm the image circle almost covers full frame, but the extreme corners are still completely black; this might be interesting for specific creative purposes, but not really usable for normal everyday shooting.
The bottom line is that if you shoot using a full frame camera, a lens with appropriate coverage really does make far more sense (and for a superzoom, that means something like a 28-300mm).
Specific image quality issues
As 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. The Sigma 18-200mm turned out to be perfectly competent in normal use, with few serious deficiencies.
The 18-200mm is quite resistant to flare, especially towards the wide end of the zoom. Place the sun towards the edge or corner of the frame, and you'll rarely see any obvious flare patterns or loss of contrast. However it can start to suffer in strongly backlit conditions towards the long end of the zoom - in this case contrast can drop significantly due to an overall haze of veiling flare. In summary the Sigma is not too bad in this regard, but perhaps not quite as good as other lenses in its class.
|18mm F16, sun in corner of frame||200mm F6.3, strong backlight|
Background blur ('bokeh')
One genuinely desirable, but difficult to measure aspect of a lens's performance is the ability to deliver smoothly blurred out-of-focus regions when trying to isolate a subject from the background, generally when using a long focal length and large aperture. This lens can allow you to achieve quite substantially blurred backgrounds, especially at longer focal lengths and wide apertures.
In common with superzoom tradition, bokeh is generally reasonably neutral, neither adding to nor detracting from the image as a whole. For closeup shots, backgrounds are quite smoothly blurred and perfectly attractive; however under certain circumstances bokeh can begin to look slightly more distracting with harsh, bright-edged rings to point highlights. But overall the Sigma is probably neither better not worse than any other similar lens.
|200mm F8||50% crop|
|52mm F4.8||50% crop|
Chromatic aberration is notably low with this lens, certainly for a superzoom. It's only likely to have much of a negative impact on real-world images at wideangle, and the strong red-cyan fringing in the sample below is close to the worst-case scenario (dark branches against a bright sky is as near to a pure black/white pattern as you'll see). At the telephoto end there's a little green/magenta fringing, but rather less than we've seen from other superzooms. Of course if you shoot JPEGs with Nikon bodies such as the D300 and D90 which feature automatic CA compensation, the effects of lateral CA will be significantly reduced anyway.
|F6.3, Nikon D300||F6.3, Nikon D300|
|100% crop, left of frame (RAW + ACR)||100% crop, lower right (RAW + ACR)|
|100% crop, Nikon D300 JPEG||100% crop, Nikon D300 JPEG|
Softness at 80mm
One issue impossible to ignore from our studio tests is the significant drop in sharpness across almost the entire frame encountered at 80mm. In order to demonstrate how this appears in real-life images, we've adapted the classic 'brick wall' test to alternative construction materials, and shot the same image wide open (F5.3) and stopped down to F8. The 100% crops below shown that even the center is soft and exhibits low contrast wide open; and while it improves dramatically on stopping down to F8, the lens still struggles to render much in the way of detail in the corners of the frame.
|80mm, Nikon D300||80mm, Nikon D300|
|100% crop, center||100% crop, center|
|100% crop, top left||100% crop, top left|
The 18-200mm features Sigma's own 'Optical Stabilization' system, which first debuted on the 80-400mm F4.5-5.6 EX OS in 2002. Unusually the company makes no specific claims about the capabilities of its technology (in terms of how much lower the shutter speed can be dropped before the effects of camera shake become apparent), but with the likes of Canon, Nikon, Tamron and Panasonic achieving three to four stops with similar systems, we have a reasonable benchmark for to expect. The mechanism is very quiet when operational, but makes noticeable clicking noises when it activates and deactivates, presumably the result of the OS group moving in and out of the 'at rest' position.
We've generally found the stabilization units in SLR lenses to be pretty effective in real-world use, and to quantify this, we subjected the 18-200mm to our studio image stabilization test, using the wideangle and telephoto settings plus one mid-range focal length (50mm). With its effective focal length range of 27-300mm, we'd normally expect to be able to get good results handheld at 1/50 sec at wideangle, and 1/400 sec at telephoto without image stabilization. The subject distance for these tests was approximately 2.5m for 18mm and 50mm focal lengths, and 8m for 200mm; the test camera was the Nikon D300.
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.
|18mm OS OFF||50mm OS OFF||200mm OS OFF|
|18mm OS ON||50mm OS ON||200mm OS ON|
Sigma's optical stabilization technology is clearly doing a useful job of compensating for camera shake; it doesn't seem to be as effective as the very latest in-lens units from the likes of Canon and Tamron, but it's not doing badly at all. This particular implementation appears slightly more effective at wideangle than telephoto, which is unfortunate as it's more valuable at the longer end; on closer examination of the image files it appears that Sigma's auto-panning detection may be a little over-sensitive, and have a tendency to disable stabilization along one axis when it really shouldn't.
At 18mm, we got better results at 1/3 sec with OS turned on than at 1/25 sec with it off. At 50mm, we saw a relatively low percentage of shots which were critically sharp at slower shutter speeds with OS on, but considering also those with just mild blur, we were able to achieve 50% usable shots at 1/6 sec, compared to 70% at 1/50 sec with OS off - a gain of perhaps 2.5 stops. The tests at 200mm told a similar story; critically sharp shots were relatively rare even with OS on, but we were able to obtain a decent percentage of useable results at shutter speeds 2 - 3 stops slower compared to when it was switched off.