I descend for internet service once per day at best but not always. Please be patient when subscribing.
Autofocus Accuracy—Nikon D200, 17-35/f2.8 AFS
Background material should be read first in Autofocus Accuracy to understand context and test methodology.
This test was shot using the Nikon D200 and 17-35/f2.8 AFS. Four shots were taken, forcing the camera to refocus each time by first moving the focus point past infinity. All shots were taken at f2.8 (wide open). The subject was approximately 80 feet away.
Staring intently at a small portion is difficult, let your eyes scan the crops as a whole.
There is clear variation in sharpness in the left crop. Observe the smearing on the vase at the bottom of the crop, and the reduction in contrast and fine detail edges throughout. A sharpened version can be seen below the unsharpened version.
|Left, no sharpening|
|Left, Unsharp Mask 500,0.3,0|
There is clear variation in sharpness in the left crop. O
As with the left crop, there is clear variation between shots. It is heartening to see that left and right crops are consistent. If the lens were optically misaligned, or the lens and sensor were not aligned, we might draw different conclusions, depending on whether we examined the left or right crops.
|Right, no sharpening|
|Right, Unsharp Mask 500,0.3,0|
Little, if any difference is apparent, consistent with the right of center results.
|Left of center, no sharpening|
|Left of center, Unsharp Mask 500,0.3,0|
It is difficult to see any difference between the crops, either unsharpened or sharpened.
|Right of center, no sharpening|
|Right of center, Unsharp Mask 500,0.3,0|
Focus with within the center 2/3 area produced indistinguishable results for all 4 frames. However, the range of sharpness on the left and right edges was noticeably different from the best frame to the worst. Thus, it is not possible to assess optimal focus for the image as a whole without examining the entire image—not just the center area.
What explains this result? The most likely explanation is that optical aberrations are at their worst wide open, with the center suffering the least. The aberrations are small enough in the center area to produce a sharp looking result within a small range of focus, with depth of field sufficient to retain the sharpness high enough to produce indistinguishable results in the center. As focus moves away from the optimal point, the aberrations increase near the edges, beyond the capability of depth of field to mitigate it.
An alternate explanation is that there is a small curvature of field present, which tolerates little variation before the edges go slightly out of focus.
That’s a theory at least—I’m not an optical expert, but it’s my best guess.
Lens evaluators take note: multiple frames are an absolute requirement for verifying the validity of conclusions about sharpness. The range of differences seen near the edges are more than enough to call one lens “soft” and another “sharp”.