Studio Tests (Full frame)
The Canon EF 24-70mm f/2.8 L II USM delivers truly excellent results on full frame cameras - again it's at least as good as anything else in its class, and a match for most primes. Sharpness is pretty impressive even wide open, and superb across the frame at F5.6. However distortion and vignetting are higher than on APS-C, as is usually the case. The lens outperforms its predecessor in terms of overall sharpness, most notably wide open at the telephoto end.
|Sharpness||At F2.8, central sharpness is extremely high at all focal lengths. The corners may not quite reach the same dizzying heights, but they're still very good indeed. As we'd expect optimal cross-frame sharpness is achieved around F5.6 - F11, beyond which diffraction starts to blur the image.|
|Chromatic Aberration||Lateral chromatic aberration is very low. You will see a little fringing towards the edges and corners if you go looking for it, but it will rarely have much practical impact.|
|Vignetting||Vignetting is, as usual, much stronger on full frame compared to APS-C. It's most pronounced at wideangle, with a 2 stop drop in brightness in the extreme corners at 24mm and F2.8; this drops progressively on stopping down but never quite goes away. At the telephoto end, there's some vignetting of the extreme corners of the frame at F2.8, but this disappears at F4.|
|Distortion||Distortion becomes more pronounced on full frame compared to APS-C. There's quite strong barrel distortion at wideangle (1.7%), changing to visible pincushion distortion at telephoto (-1.6%). This is strong enough to be potentially problematic in geometric compositions.|
As you'd expect, at minimum focus the story is much the same as we saw on APS-C. Image quality at F2.8 is pretty awful (click here for the test chart shot), but stop down to F5.6 and the centre sharpens up nicely. The corners appear soft in our flat-field chart test due to curvature of field, and only really sharpen up at F11 - F16. There's a little barrel distortion, and quite strong chromatic aberration towards the corners of the frame.
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 Canon EF 24-70mm f/2.8 L II USM performs with the consistent excellence you'd expect from this type of lens; its main failings come when shooting at or close to its minimum focus distance.
Many fast lenses have problems with flare when pointed at bright light sources - the sheer amount of glass makes suppression of internal reflections relatively difficult. With this in mind, the 24-70mm does very well - we've stressed it by shooting directly into the low winter sun, and it's barely batted an eyelid.
The examples below give an idea of how the lens behaves. With the sun in the frame at wideangle, and the aperture stopped right down to F22 (which makes flare patterns most defined), there's a diagonal array of spots, but minimal overall loss of contrast. There's a large red dot diagonally opposite the sun itself, but even against a near-black background it's not especially intense. At larger apertures, or against brighter backgrounds, is essentially disappears.
|Canon EOS 6D, 24mm, F22||Canon EOS 650D, 70mm, F6.4|
The second example, shot at 70mm on the APS-C EOS 650D with the sun at the edge of the frame, is a about the worst we ever managed to coax out of the lens in extensive real-world shooting. Even then, the magenta flare spot disappeared with a small shift of framing.
The studio test suggest the 24-70mm has relatively low lateral chromatic aberration, and this is confirmed in real-world use. As usual, CA is at its worst at wideangle, but even with in-camera compensation disabled you have to look hard to see any fringing. Perhaps even more impressive is the almost complete lack of longitudinal CA - in all of our real world shots, there was minimal fringing around out-of-focus elements of the image.
In the examples below we're looking at colour fringing in the corners of the frame at 24mm. Here we've used DxO Optics Pro to provide firstly an uncorrected image, and secondly one with lateral CA removed. Even in the first version there's only a little red/cyan fringing in the corner of the frame, and you'd have to print really large for it to have any negative impact.
Lateral Chromatic Aberration
|24mm F8, Canon EOS 6D, CA uncorrected||100% crop, top left: green/magenta fringing|
|24mm F8, Canon EOS 6D, CA corrected||100% crop: fringing removed|
The tests show that the 24-70mm F2.8 exhibits about 2 stops vignetting wide open at 24mm, which is pretty typical for its class. In the 50-70mm range it also shows quite abrupt vignetting in the extreme corners, which can be visually quite intrusive in some some situations. Naturally it can be corrected in post-processing, using either Canon's own Digital Photo Pro or third party software such as Adobe Lightroom or DxO Optics Pro. Most Canon SLRs since the EOS 50D can also automatically correct vignetting in their JPEG processing, and the most recent high-end models such as the EOS 5D Mark III and 6D can also do so during post-capture in-camera RAW conversion.
The rollover below compares image shot at F2.8 using the Canon EOS 6D with versions that have had vignetting corrected using in-camera RAW conversion. In the wideangle shot, the vignetting is visible but nor especially ugly - in many cases it will simply add an element of framing to the shot. But at 53mm the appearance of the vignetting is much less pleasant, and in shots with evenly-toned areas in the corners you'll probably want to get rid of it..
|Canon EOS 6D, 24mm F2.8, uncorrected||24mm F2.8, in-camera compensation|
|Canon EOS 6D, 53mm F2.8, uncorrected||53mm F2.8, in-camera compensation|
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/or a large aperture. The 24-70mm's fast maximum aperture means it can provide nicely-blurred backgrounds at the telephoto end.
The degree of blur isn't all that matters, of course, but also its aesthetic quality (which is what the word 'bokeh' refers to). The lens does beautifully here - backgrounds are smoothly blurred at pretty well any combination of focal length and subject distance. There's also practically no unsightly colour fringing due to chromatic aberration. This is really desirable in this type of lens, and doesn't always go hand-in-hand with the kind of sharpness the Canon delivers.
|EOS 6D, 70mm F2.8||Background detail|
|EOS 650D, 70mm F2.8||Background detail|
Here we're looking at examples both shot at 70mm F2.8 - one on the EOS 6D with a focus distance of about 2m and relatively distant background, the other a closeup on the EOS 650D. The 24-70mm has rendered both attractively, with no hint of CA.
Close focusing and macro
The one area where the EF 24-70mm f/2.8L II USM does fall a bit short is when used close to its minimum focus distance. Here image quality suffers dramatically at larger apertures, with properly-sharp results only at F5.6 or smaller. Because the lens is focusing through lots of aberrations, both auto and manual focusing can be rather hit and miss - in AF the camera tends to backfocus, and in manual it can be almost impossible to judge when the image is properly focused. If you're after critical focus accuracy and have the camera on a tripod, your best bet is therefore to use magnified live view, and adjust focus while holding down the depth of field preview button with the aperture set to F4 (this is just as awkward as it sounds).
The examples below illustrate this, using the EOS 6D set on a tripod and focused using magnified live view. The 100% crops are taken from the critically-focused region towards the centre of the frame. At F2.8 the image almost looks out-of-focus, but the fine delineation of the point highlights proves otherwise. The F4 shot is still a little soft, but F5.6 is properly sharp. Diffraction slightly softens the image at F16, and has a clear impact at F22, but in practice this is offset by the increased depth of field.
|EOS 6D, 1/50 sec F2.8 ISO100, ~0.39m||F2.8, 100% crop|
|F4, 100% crop||F5.6, 100% crop|
|F8, 100% crop||F11, 100% crop|
|F16, 100% crop||F22, 100% crop|
In the crops below, we compare the manually-focused shot with the best of several AF attempts, using a focus point placed on the high-contrast edge of the petal. This demonstrates why AF is inaccurate at close range - it actually maximizes the overall contrast of this sharp transition, but this means the lens ands up back-focused by a couple of millimetres. Because of this, the top of the petal (which is fractionally further from the camera) is closer to being in-focus in the AF shot, as can be seen in the lower crops.
|F2.8, magnified live view manual focus||F2.8, autofocus|
|100% crop, top of petal||100% crop, top of petal|
This only seriously affects a relatively narrow range of distances - perhaps 0.7m down to the minimum of 0.36m - so it does seem a bit a churlish to complain. But it's definitely worth knowing about when using the lens for closeups.