Canon EF-S 17-85mm 1:4-5.6 IS USM review
The 17-85 gives what can only be described as mixed results in our studio tests. Its performance is rather disappointing at wideangle, where it serves up a witch’s brew of corner softness, green/magenta chromatic aberration, barrel distortion and light falloff; however it becomes very much better at telephoto, and in the 35-70mm range is really very good indeed.
|Sharpness||Best results are obtained in the normal to telephoto range, where the lens gives consistently high sharpness readings from 35mm to 85mm. Unfortunately things go downhill towards wide angle, especially in the corners which are distinctly soft at 17mm. The lens appears to be optimized to perform wide open at all focal lengths, and shows no systematic increase in quality on stopping down to smaller apertures; at F11 and beyond the sharpness drops rapidly due to diffraction, as expected. We’d certainly advise against using apertures much smaller than F16, except in those rare cases where extreme depth of field is important.|
|Chromatic Aberration||Chromatic aberration is distinctly worst at 17mm, where the almost-overlapping red and blue lines indicate green-magenta fringing, which is visually the most intrusive (see also the example below). The story is similar at 24mm, but at 35mm and beyond CA becomes much less of an issue.|
|Falloff||We consider falloff to become perceptible when the corner illumination falls to more than 1 stop less than the centre. Once again, the lens gives a poor showing at 17mm, with 1.7 stops of falloff wide open; you’ll need to stop down to around F8-11 to reduce it completely below perceptible levels. However aside from that, there’s nothing too much to worry about, with just a hint of falloff wide open at 24mm.|
|Distortion||The 17-85mm also shows unusually high barrel distortion at wide angles, measuring a whopping 2.4% at 17mm. This is also not 'pure' barrel distortion but a more complex 'wave'-type, with the barrel effect most pronounced towards the centre, then pinched in again towards the edges, which makes it relatively difficult to correct in software. At 24mm and longer the pattern changes to pincushion, and at its worst reaches -1.5% at 50mm; again pretty pronounced, and readily visible if the image should contain straight lines.|
Specific image quality issues
By far the most problematic aspect of this lens's performance in real-life shooting is chromatic aberration at wide angle. Strong green/magenta fringing is visible in 17mm shots even at modest magnifications, and stopping down the aperture has no effect. It remains a problem at 24mm, but is much less of an issue at longer focal lengths. This type of chromatic aberration can be 'corrected' in software, and particularly efficiently if you’re prepared to shoot RAW. For instance, using our standard converter, Adobe Camera RAW, processing with -30 Red/Cyan and +30 Blue/Yellow effectively eliminated CA.
|100% Crop, left of frame||17mm F8.0, EOS 400D|
Aberration correction using Digital Photo Pro (updated July 2008)
With version 3.4 of their free RAW conversion software Digital Photo Pro (DPP), Canon have added lens aberration correction support for the 17-85mm IS. This allows RAW shooters to eliminate (or at least reduce) the effects of chromatic aberration, falloff (which Canon refer to as 'peripheral illumination'), geometric distortion, and 'color blur' (which appears to be Canon's term for axial chromatic aberration).
DPP's aberration correction works by using profiles specific for each lens, making the software fairly straightforward to use, and with no major need to play with the slider settings (except perhaps for the occasional small tweak). Here's an example of chromatic aberration correction using default settings for the image above; DPP has effectively removed the ugly green and magenta fringing completely.
|100% Crop, left of frame||17mm F8.0, EOS 400D|
The corrections for peripheral illumination and geometric distortion are equally impressive; in the example below we can see clear correction for falloff even at F8 (however in more extreme cases, this will come at some cost to image noise). More importantly, the strong barrel distortion at 17mm is also well corrected; this is most obvious in the bowed line along the lower edge of the frame, but verticals also become properly parallel. Of course there's no such thing as a free lunch, and this kind of software distortion correction will require some localized 'stretching' of the image data resulting in a loss of resolution in affected regions, but overall this is a small price to pay when distortion is otherwise objectionable.
EOS 400D, 17mm F8
|Original||Peripheral Illumination||PI + Distortion|
Optical Image Stabilization
The 17-85mm features Canon's second generation optical image stabilization, which claims to allow handholding at shutter speeds 3 stops lower than usual before blur from camera shake becomes apparent. It's near-silent in use, with just a quiet whirring noise when operational, and with distinct clicks when it activates and deactivates, presumably due to the lens elements moving in and out of the 'at rest' position.
We've generally found the units in SLR lenses to be pretty effective in real-world use, and to quantify this, we subjected the 17-85mm to our studio image stabilization test at both wideangle and telephoto. With its effective focal length range of 27-136mm, we'd normally expect to be able to get good results handheld at 1/30 sec at wideangle, and 1/160 sec at telephoto without image stabilization. The subject distance for these tests was approximately 2.5m.
|17mm IS OFF||85mm IS OFF|
|17mm IS ON||85mm IS ON|
Here we can see clearly the effectiveness of Canon's in-lens IS unit. At the wideangle end and shutter speeds of 1/6 sec, IS is delivering 70% critically sharp shots, compared to just 20% without. Even at very slow shutter speeds of 0.3 sec, IS is giving usable images with no more than mild blur 80% of the time, compared to just 10% without.
Performance is even better at the telephoto end; at 1/20 sec, IS gives 100% sharp images, in contrast to just 10% without IS. Even at 1/10 sec, 90% of images are usable (just mild blur) with IS on, compared to 10% with it switched off. Impressive stuff, the image stabilization is clearly delivering the goods here.