First a note about focal length of the Sigma APO 120-300mm F2.8 EX DG OS HSM. The 120-300mm range is measured at infinity focus, but like many other internal focusing zooms, at closer focusing distances there can be small changes in focal length. For example when close focused at around 10ft at the nominal 300mm zoom setting, I measured an effective focal length of about 240mm. What this means is that you get about 80% of the image size you might expect from a 300mm lens focused at that distance.
The focal length change gets smaller as the focus distance is increased, so that at 30ft its up to around 273mm and at 50ft it's up to 281mm. At infinity I measured close to 300mm. As I said, many lenses do this but sometimes it can take people by surprise if they are unaware of the fact. The effect is sometimes referred to as "focus breathing".
In general focusing was quite fast, quiet and accurate. When used with an EOS 5D and EOS 7D I didn't notice any tendency to front or back focus and comparing the results of AF with those using Live View on the 7D for manual focusing there was no consistent difference in sharpness of the images.
As I mentioned in the introduction, there's no focus range limiter switch on the Sigma APO 120-300mm F2.8 EX DG OS HSM so if it (or more likely the camera) gets confused it can search the range from infinity down to around 2m (1.5m at 120mm, 2.5m at 300mm). The time taken to do this is around 0.67 seconds. The "round trip time" from infinity to close focus and back to infinity was around 2 seconds, since there's about a 0.7 second pause when the focus direction reverses. The benefit of a focus range limiter is that the lens can search for focus much faster over a restricted range, assuming you know your subject will be somewhere within that range of distances.
Though AF is quiet, it's not totally silent and that's most noticeable when tracking AF is used and you can hear the AF mechanism moving in small steps. It's not loud, but you can hear it. AF tracking accuracy is notoriously difficult to measure since there's really no standard test and it probably depends as much on the camera as the lens. However I did shoot some sequences of approaching and retreating cars using an EOS 7D body and tracking AF and a good fraction of the images were sharp after the initial lock. I know that's not a very quantitative test but I'm not really setup to do quantitative tests of AF tracking.
The Sigma APO 120-300mm F2.8 EX DG OS HSM has optical stabilization built into the lens. The Sigma OS system works in a similar way to the Canon IS system. Sensors detect lens movement and move one of the internal optical groups in a way to counteract the movement of the lens and try to keep the image projected onto the sensor stationary.
There are two OS modes. The first is the general mode which stabilizes on both the horizontal and vertical axis, the second is used when panning and stabilizes only on the vertical axis. Sigma state that the OS system provides to to 4 stops of stabilization, meaning that the lens can be hand held at shutter speeds up to 4 stops slower than would be possible without OS and sharp images can still be obtained.
The exact degree of stabilization is difficult to quantify because it's all a matter or percentages. Stabilization raises the odds of getting a sharp shot, but it doesn't guarantee it. You also have to define "sharp" in this context. If by "sharp" you mean just as sharp under 100% viewing on a monitor as the image would be with the lens on a tripod, then you get far fewer sharp images than if you use "sharp" to mean that to an average viewer and 11x14 print would appear sharp.
In practical terms I'd say that the Sigma OS system is good for 2-3 stops most of the time if you judge the images based on a criterion that they wouldn't look blurred in an 11x14 print to the average viewer. Most of the time means that you can probably expect maybe 3 out of 4 shots to be "sharp". If you assume that for a 300mm lens shot on a full frame camera you'd normally need at least 1/300s to get a sharp hand-held image, then 2 stops of stabilization would mean you could use 1/75s and 3 stops 1/35s. It's quite possible you may get an occasional sharp image at 1/20s or even 1/15s (Sigma's "4 stop" figure), though in my experience it was unlikely.
All three of these aberrations are well controlled. Vignetting (corner darkening) is just over a stop in the corners of a full 35mm frame wide open (f2.8) at 300mm and just under a stop at 120mm. This is small enough that on most images you wouldn't notice it and if you did it could easily be corrected post-exposure. At smaller aperture and on crop (APS-C) sensors, vignetting is really a non-issue.
Distortion is low. At 120mm it's hard to see any distortion at all and even at 300mm there's less than about 1% pincushion distortion. Unless you're shooting pictures of distortion test charts, this level of distortion is unlikely to be noticed.
Chromatic aberration is also well controlled. Even in the corners of the full frame image it was hard to see any chromatic aberration. Again, not something that's likely to be noticed in the image, even by the most critical pixel peeper.
Overall the Sigma APO 120-300mm F2.8 EX DG OS HSM is a sharp lens across the whole range of focal lengths and at all apertures. In the center of the frame sharpness is excellent with perhaps very slightly lower sharpness at 300mm than 120mm. Across the whole range the center sharpness really doesn't increase noticeably as the lens is stopped down. Center sharpness is so good that there is no need to stop down anyway! On resolution charts you can see that there is a very slight sharpness increase going from f2.8 to f4, but you have to look closely.
The corners of the full frame image are a little softer than the center, especially wide open and more so at 300mm than at 120mm. However corners cannot be described as "soft" even wide open at 300mm. It's just that they a a notch less sharp then the center. As with all good lenses, stopping down past f8 results in slight softening due to diffraction, though normally you'd need to be shooting images of resolution test charts to see this at f8 and f11. At f16 and f22 it's a little more noticeable, but as I said, this is just optics. All lenses show diffraction related softening at small apertures.
In terms of comparison with other lenses, the Sigma APO 120-300mm F2.8 EX DG OS HSM compares very favorably with the Canon EF 70-200/2.8L IS II USM when you're looking in the center of the frame in their overlapping focal length range (120-200mm). However the Canon lens is a little sharper in the corners of the full frame image. This is a severe test though as the Canon lens is probably one of the sharpest telephoto zoom lenses out there right now.
Compared with the Canon 300/4L USM the Sigma APO 120-300mm F2.8 EX DG OS HSM showed about the same center sharpness at the same aperture. Again, this is a high level of performance.
One of the features of f2.8 lenses (apart from their intrinsic speed) is that fact that they can be used with 1.4x and 2x multipliers and still retain AF on all DSLR bodies. With a 1.4x TC the Sigma APO 120-300mm F2.8 EX DG OS HSM becomes a 169-420mm f4 and with a 2x it becomes a 240-600mm f5.6.
The TCs I had available were the original Canon 1.4x and 2x models. Canon now have MkIII versions of both, plus Sigma have their own 1.4x and 2x TCs. The results I will describe here are based on the use of the original Canon 1.4x and 2x TCs. Results may be different when the lens is used with other TCs.
As would be expected, image sharpness drops when multipliers are used, and the drop is more noticeable with the 2x than the 1.4x and with the lens at 300mm vs 120mm. However the image actually holds up pretty well, especially with the 1,4x TC, and unless you want to make large prints or you expect image quality on a par with a prime lens, the Sigma APO 120-300mm F2.8 EX DG OS HSM with a TC may well prove to be a reasonable (and far cheaper) alternative to a longer lens.
I'd caution about thinking of the Sigma APO 120-300mm F2.8 EX DG OS HSM (or indeed almost any other 300mm lens) with a 2x TC as a "cheaper, lighter" substitute for a 600mm lens if what you reallywant is, in fact, a 600mm lens. Yes, it is cheaper and lighter but no, it's not as sharp. However if what you want is a 300mm lens that can occasionally be pressed into service as a 600mm lens with a 2x TC, then that's a reasonable proposition. I did a few test shots comparing the Sigma lens with the 1.4x TC at 420 f4 and with the 2x TC at 600mm f5.6 against Canon EF 500/4.5L and the Canon prime was sharper and had higher overall image quality out to the corners of the frame.
I will note again however that I was using the Canon 1.4x and 2x MkI TCs. Results with different TCs may well be different. I'm sure Sigma would recommend the use of their own Sigma APO 1.4x EX DG and Sigma APO 2x EX DG multipliers with the Sigma APO 120-300mm F2.8 EX DG OS HSM.
NEXT -> Final comments and conclusions