Upload & Sell: On
Received my 2nd copy, which has only a 1/2 stop miscalibration on the aperture lever vs a full stop on my first copy.
* VC "break-in" issue and "fix" *
Both copies exhibited an out-of-box phenomena where the vibration-control doesn't work properly initially. The symptom was two-fold: 1) When the VC first engages the images shifts dramatically, much more than the typical Tamron-VC shift and certainly more than a Nikon VR lens. There is an initial large shift followed a few quick shorter shits. It takes about a second before the shifting stops. 2) The VC doesn't actually yield any stabilization when it exhibits symptom #1. Photos taken with VC on are just as-blurry and many times more blurry than with VC off. It appears the large shift that occurs when VC is engaged is leaves the floating element positioned so that it can't supply any stabilization.
On my first copy the lens exhibited the above symptoms for the first hour and then suddenly the symptom disappeared and the VC started working as advertised. I surmised the VC elements must have needed some nudging to either wear-in or position them correctly. Just a wild guess at this point. When I realized the second copy had the same issue even after 15 minutes of use I expedited this "wear in" by entering Live View on my D800, which has full-time AF and so the VC stays engaged. I let the IS stay on for about two minutes, after which the VC worked properly.
Once the VC is working properly, I'm able to achieve sharp photos down to 1/10 hand-held on both my D800 and D3s about 70% of the time @ 70mm. On my D5100 I can achieve sharp photos at the same percentage at 1/10 as well, even though the effective focal length is .5x more. The lens is rated at 4-stops of VC but that would mean 1/4 second on FF, which becomes too slow without some type of physical support, like leaning against a wall.
* Miscalibrated aperture lever *
My second copy is off by 1/2 stop, so that f/4.0 is actually f/3.3 in terms of light/DOF. You can read more about my experience with this on my first copy here: http://forums.dpreview.com/forums/read.asp?forum=1030&message=42138310
* Decentered *
My second copy exhibits moderate decentering, with the extreme left edge much sharper than the right. The two edges become equal at about f/5.6 (which is really f/4.8 when you account for the aperture miscalibration). Oddly at f/.8 the right side becomes sharper than the left. This decentering was noticed at 70mm. If this copy acts like other mid-range zooms, I expect the decentering may flip to the other side at the wide end, though I haven't checked this yet. The decentering is worst at MFD distances but luckily improves at mid-range distances.
Decentering at near distance:
Decentering at middle distances:
* IQ *
My second copy is extremely sharp at all focal lengths, much more so than my first copy. Even at f/2.8 there is an abundance of detail and contrast. Lens is sharp edge-to-edge at f/5.6 and above. There is some CA/fringing, even up to f/5.6, depending on the scene. IQ is a bit less at MFD vs middle-to-infinity focus distances. Flare control seems decent. Bokeh is a bit nervous and lens suffers from the previously-reported "onion bokeh" effect of concentric circles for OOF highlights.
* AF *
AF speed is very fast, just a click off the Nikon 24-70. Focus appears accurate thus far - my 2nd copy tuned at 0 on both my D800 and D3s.
* Sample photos *
Here are some sample photos, on both the D800 and D3s. All were shot raw and sharpened in two passes (raw + large-radius USM in CS5). Full-sized images are available in the gallery, and EXIF is maintained. All of the D3s images were indoors at very low shutter speeds. I expect to add more images in the coming week.
* Conclusion *
The lens has sample variation/new-model teething problems but if you can find a copy with only a slight aperture calibration error, "break-in" the VC, and get a sharp copy...the hassle IMO is very much worth it.