Tamron SP AF28-75/2.8 XR Di LD Aspherical (IF) Macro
Optical TestingI used Norman Koren's Imatest software to look at lens sharpness. Imatest is a useful program which allows you to make sophisticated plots of the system response function, which is simply the MTF of the lens multiplied by the MTF of the sensor.
The problem with doing such tests and publishing the data is that while there is one way to get the right answer there are an infinite number of ways to get answers which are wrong! Absolute focus accuracy is very, very critical for example. Exposure can have some effect. Getting the sensor in the camera exactly parallel to the sensor is critical too. The target and target distance may also have a small effect.
In order to publish data from Imatest, I'd want to be in a situation where I could run the same test on a lens many times and come up with exactly the same numerical results and I'm not yet at that point. I am identifying exactly what I need to do that, and once I have a reliable methodology I'll probably start publishing numbers, but right now I'm not quite there.
Note that this isn't a criticism of Imatest, it's a realization of the axiom GIGO ("garbage in, garbage out"). Unless you feed the program good data, you can't expect good data to come out!
So rather than publish numbers I might latter have to revise or retract, I'm simply going to qualitatively describe the results of the testing and in section IV I'll present some crops from real world images. Real world images also give a much better "feel" for the differences between lenses. If one lens "scores" a 45 and another lens "scores" a 52, that really doesn't tell you a lot about how the images of real world subjects will look.
I chose to compare this lens with the Canon EF28-135/3.5-5.6 IS USM for two reasons. First their price is similar (around $370 for the Tamron, $415 for the Canon) and second, I have an EF28-135 IS USM available for comparison. You could also argue that since IS gives the Canon lens a 2 stop advantage in "hand-holdability", the low light performance of these two lenses might be similar (for static subjects anyway). Some people might want a comparison with the Canon 24-70/2.8L USM, but I don't own one. It's a $1150 lens, 3x the price of the Tamron, and so probably not a direct competitor. I also used a 50/1.8 for comparison, since I had one of those too!
I did a large number of tests, running these lenses through all apertures at 28mm, 50mm and 75mm, shooting several types of test target as well as "real world images" and evaluating images for sharpness, contrast, flare, distortion and chromatic aberration. Note that all test were done using a Canon EOS 20D digital SLR which has an APS-C sized sensor (ca. 22.5 x 15mm). I did not test the full frame performance of the lens using film or a full frame DSLR. Center performance would be expected to be the same, but performance at the edges of a 36mm wide frame would be expected to be slightly worse (lower sharpness and greater chromatic aberration) than that seen at the edges of a 22.5mm frame.
Technical assessment of performance
28mmAt 28mm center sharpness of the Tamron lens was good at all apertures. Sharpness peaked around f4 to 5.6 at which point edge shapness was only slightly lower than center sharpness. Wide open edge sharpness was down a little on center sharpness, as might be expected, but stopping down to f4 brought it up nicely.
The Tamron was somewhat better than the Canon 28-135 at all apertures down to about f11. At smaller apertures both lenses showed similar performance, with maybe a slight advantage to the Tamron in terms of edge sharpness. This is pretty much as expected since at apertures of f16 and smaller diffraction is generally the limiting factor on sharpness in the center of the frame and is the same for all lenses.
50mmAt 50mm the Tamron was actually very slightly sharper then the Canon 50/1.8 prime at equal aperture. The difference was small and in real world images might be hard to detect. Both lenses gave peak performance from about f4 to f8. At f11 and smaller apertures the effects of diffraction lowered sharpness.
Just for the hell of it, here's a system response function plot for the Tamron 28-75/2.8L and the Canon 50/1.8. The vertical axis is the spatial frequency (in cycles/mm at the sensor) at which the system MTF is equal to 0.5. As I cautioned above, don't put too much faith in the absolute numbers presented here. The Canon 28-135 trailed in sharpness. It wasn't bad but it was a notch down on the Canon prime and Tamron 28-75.
75mmAgain the Tamron 28-75 was somewhat sharper than the Canon 28-135 at all apertures. Wide open at f2.8 there was some softness but at f4 things were sharp and performance from f4 to f8 remained pretty constant. Edge sharpness increased as the lens was stopped down and peaked around f8.
In the final section (Part IV) of this review there are some real world shots which show comparative edge and center sharpness of these lenses.
© Copyright Bob Atkins All Rights Reserved