Tamron AF 24-70mm f/2.8 SP Di USD VC Review - Part II

Focus

The Tamron AF 24-70mm f/2.8 SP Di USD VC uses Tamron's Ultrasonic Drive motor (similar to Canon's USM), focus is fast and quiet. I measured the time taken to focus from infinity to MFD (minimum focus distance) at just under 0.3 seconds. You can just hear the lens focus, but if you were more than a couple of feet from the camera, it would be silent. On my EOS 7D I found focus to be positive (no hunting) and accurate. Comparing AF with Live view MF at 10x magnification, I found no difference. A rotation of about 120 degrees of the manual focusing ring was required to cover the full focus range. The filter does not rotate during focusing (or zooming).

Zoom

The zoom action was smooth with no zoom creep. Focal lengths of 24mm, 35mm, 50mm and 70mm are marked. The zoom ring turns though about 90 degrees to go from 24 to 70mm. Rotation is clockwise (viewed from the rear) to go from 24mm to 70mm. This is the opposite direction from Canon's own lenses, but is the convention for Tamron lenses. It's also the convention for Nikon lenses, so Nikon users will feel at home with it!

The lens extends by about 30 mm when zooming from 24 to 70mm.

Optical Performance

Unless otherwise indicated, comments here refer to the performance of the Tamron AF 24-70mm f/2.8 SP Di USD VC mounted on a full frame DSLR (Canon EOS 5D, 13MP). I did not have a 5D MkII or III available at the time of this test, so the original 5D was used.

Distortion

Distortion at 24mm measures around -3.5% (barrel). This amount of distortion is quite noticeable on subjects with horizontal and vertical lines near the edges of the frame. It can of course be corrected in software post-exposure.

Distortion at 24mm. Full frame on an EOS 5D

Distortion drops rapidly as the lens is zoomed out. By 35mm is down around -0.5% (barrel) which is rarely noticeable on images. At 50mm there's virtually no distortion and at 70mm there is mild pincushion distortion (less than 1%).

I'd say that only at the widest setting (24mm) is distortion likely to be noticeable in most images. It's a little higher than you'd see on the comparable Nikon and Canon 24-70/2.8 zooms, but not much higher. However it's better than the Canon 24-105/4L is at 24mm.

Vignetting

Vignetting at 24mm is quite noticeable, with the corners being a little over 2 stops darker than the center of the frame (full frame). Stopping down does reduce vignetting, but even at f8 the corners are almost a stop darker than the center.

Vignetting at 24mm and f2.9 on a full frame EOS 5D. The red frame shows the area covered by an APS-C crop sensor which would show much lower vignetting

One you zoom away from 24mm things get better, with corners being around 1.5 stops darker than the center from 35-70mm wide open. again stopping down helps and by around f5.6 the darkening drops to around 0.5 stops.

These numbers are not way out of line for a fast 24-70 zoom. Both the Nikon and Canon 24-70/2.8 lenses show close to 2 stops of corner vignetting wide open at 24mm. The Canon 24-105/4L shows even higher vignetting at 24mm than the Tamron AF 24-70mm f/2.8 SP Di USD VC. Vignetting from the Nikon 24-70/2.8 is shown below. Actually it looks a little more decentered then the vignetting pattern from the Tamron AF 24-70mm f/2.8 SP Di USD VC.

Nikon 24-70/2.8 vignetting at 24mm f2.8 (full frame)

Things are much better if you are shooting with a crop sensor of course since the corners of the full frame are cropped away.

Chromatic Aberration

The shows very good control of lateral chromatic aberration at all focal lengths. It's unlikely that chromatic aberration will be a problem at any setting.

100% crop from corner of a full frame image taken at 24mm, f2.8 using an EOS 5D

Magnification of the above image reveals about 1 pixel of CA, red on the outside, cyan on the inside. When compared with the Canon EF 24-105/4L IS USM, at the same focal length and aperture the Tamron AF 24-70mm f/2.8 SP Di USD VC showed lower levels of CA.

Sharpness

Center sharpness is high at all focal lengths, even wide open. At 24 and 35mm center sharpness remains pretty constant from f2.8 to f5.6, then starts to drop a little due to diffraction effects. At 50 and 70mm center sharpness does improve slightly on stopping down, peaking at f4-f5.6. However as far as center sharpness goes, there's no problem shooting the Tamron AF 24-70mm f/2.8 SP Di USD VC wide open at any focal length.

For comparison here are crops from a full frame image shot with a Nikon D800 and Nikon 24-70/2.8 at 24mm and f2.8 under the same conditions. These are not 100% crops. To equalize the size between the 13MP EOS 5D and the 35MP D800, the D800 image was downsized and so does not clearly show the extent of the higher resolution in the center. Closely examined, the D800 image showed the expected 1.6x higher linear resolution due to the smaller pixel spacing (higher pixel count).

Center (left) and corner (right) from an image shot with a Nikon D800 and Nikon 24-70/2.8 at 24mm and f2.8

Here are comparison 100% crops from the center of the image shot with the Tamron AF 24-70mm f/2.8 SP Di USD VC at 70mm on an EOS 5D. The upper crop was shot at f8, the lower shot at f2.8.

Here are similar 100% crops but this time taken from the lower left corner of the image. You can see that stopping down to f8 brings up the corner sharpness. In the center, stopping down from 2.8 to f8 makes little difference.

When I shot with the Tamron AF 24-70mm f/2.8 SP Di USD VC on an EOS 7D and the Nikon 24-70/2.8 on a Nikon D800, with both lenses at 24mm and f2.8 and both lenses the same distance from the subject, the Tamron lens equaled the Nikon lens in the center and beat it in terms of resolution off axis, measured at the same distance from the center. Said another way, the 7D off axis (away from center) image with the Tamron lens (18MP) was better than an APS-C crop from the Nikon D800 with the Nikon lens (14MP) and it was clear that the difference was lens based rather than sensor based, since center resolution in both cases was similar.

Corner sharpness of the Tamron AF 24-70mm f/2.8 SP Di USD VC is generally lower than center sharpness at all focal lengths and apertures, as would be expected. Corner sharpness improves as the lens is stopped own peaks at around f8 for all focal lengths (at which point it's only slightly less sharp than the center). Overall this means that and aperture of f5.6-f8 gives the best overall image sharpness from center to corner (and also minimizes vignetting).

When compared with the Canon 24-105/4L IS USM, the Tamron AF 24-70mm f/2.8 SP Di USD VC showed significantly better corner performance at 24mm on both full frame and APS-C images when both lenses were shot at f4. Even at f2.8 the Tamron lens was better than the Canon at f4. Center performance of both lenses was similar across the focal length range. Corner sharpness was also similar in the 35-70mm range in an APS-C crop image, though the Canon 24-105/4L IS USM showed higher levels of chromatic aberration.

Image Stabilization

Tamron's VC (vibration control) system is similar to Canon's IS (image stabilization). Sensors detect lens movement and a small group of lens elements are moved so as to counteract the lens movement and keep the image on the sensor from moving. Tamron claim up to 4 stops of stabilization, but stabilization is difficult to measure. It's all a matter of probability. At any shutter speed there's a possibility the image will be sharp, but the probability becomes much less than 1 when the shutter speed drops below about 1/focal length expressed in seconds (when shooting full frame). So for a 70mm lens sharpness will be expected to drop when shutter speeds drop below 1/70s. Let's say 1/60s to make things harder! 1/30s would be one stop slower, 1/15s would be two stops slower and 1/8s would be 3 stops slower.

When shooting at 70mm and 1/8s I got no sharp images with the VC turned off. 50% were very blurred while the remaining 50% were blurred to a greater of lesser extent but probably usable for small prints. With IS turned on about 25% of the images were sharp, another 50% were slightly blurred but usable and the remaining 25% were quite blurred but maybe OK for small prints.

With lens based stabilization you can see the effect in the viewfinder. In the case of the Tamron AF 24-70mm f/2.8 SP Di USD VC stabilization takes a fraction of a second to lock in. If you just press the shutter as fast as you can, AF can lock and the shot be taken before stabilization is effective. Better to 1/2 press the shutter and once you see the image lock, press it the rest of the way.

My general conclusions about image stabilization apply here as they do to most stabilized lenses. Stabilization can help when shutter speeds become slow, but stabilization is no substitute for a good tripod! However a tripod isn't always convenient or even possible, in which case a stabilized lens has a much better chance of getting you a sharp shot than an unstabilized lens.

NEXT -> Part III Conclusions