So I attempted to buy the new Nikon AFS 300/4 PF ED VR today, for the third time. This seems like an odd thing to write, so please hear me out. Every sample of this lens I’ve tried, on every D810 body (now three of each) produces very strange double-image artefacts only with VR on. With VR off, the optics look consistently excellent. With VR on, I couldn’t get a single sharp image regardless of drive mode, shutter speed, EFC on/off, tripod or handheld. The funny thing is that I did not see this on any other body than the D810: the D800E was just fine with VR on, as were the D3, D4 and D750. I initially ruled this out as sample issues or QC, but now that I’ve tried several lenses from different batches and from different country stock with a range of bodies, I think there may be something much more serious afoot. I have reported my findings to Nikon and they are investigating…
I can only hypothesise that there is some very strange interference going on between the VR mechanism and the D810’s shutter unit. It appears plenty stable in the viewfinder – perhaps moreso than any other lens I’ve mounted – but the results are unusable. Is anybody else seeing this with their lenses, or is it just my bad luck with every sample and body?
There is a folder of full size samples here on dropbox.
__________________
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved
Ming Thein | Photographer 
I just got the same thing with the 810 and 300 VR. About 1/2 of the images were double, others weren’t. Was a solution ever found?
Unfortunately not; it appears there’s a shutter shock interaction between the D810’s unit and the 300VR. The 850 or Z7 can use EFC all the time, thus eliminating initial shutter shock. If there’s any induced shake on closing, it isn’t visible in the image.
Hi, I have the same problem with the Tamron 18-270 VC and the D5500 – 1/2 of the pictures are doubled at 1/100-1/200 with VC on. Without VC no problem, with the Nikon 10-20 and VR on 1/100-1/200 no problem. Only the Tamron 18-270 VC on and the Nikon 18-105 VR on gives this double-pictures at a shutter-speed between 1/100 and 1/200.
This sounds very much like classic symptoms of shutter shock…
Why it is so bad with VC/VR and better without?
You’ve got an extra moving element which is probably having some electronic interference with the shutter (also magnetic, like the lens element suspension/ motion system) – switching it off likely removes this interference.
I fixed the minimum shutter speed in the ISO senitive settings to 1/250. I think it is the best way to prevent these shutter shocks. After that I had nearly no shutter shock anymore, because the camera trys to get to 1/250 and then move to 1/50. Frome 1/100 to 1/200 I will always get these double pictures and I’m using the 18-105 and the 18-270 very often.
It is an issue Nikon partially addressed in later serial numbers as they replaced my early version of the lens with a slightly later version at no charge. Regardless, the problem was still present with my D750 but a bit improved with the “updated” version. I did however find that the problem was much less of an issue with later cameras like the D500 or D850, so optimal camera pairing does definitely help.
I am having the same problem with the 810 and the Nikorr AF-S 24-85. About 20% were double images.
I wonder if there are good/bad pairings of lens and body due to interference between VR and shutter mechanisms. The 810 was a lot better than the 800/800E though, in my experience. The 24-85 is an older lens and VR mechanism, too – so it’s hard to rule out contributions from that side of things…
I just received my copy of the new Nikon 300mm f/4 PF two days ago and I have to say at first I was extremely disappointed in the VR performance. After hearing so much about the VR issues with this lens I decided to go straight to testing the lens at those critical shutter speeds that result in blurred images (1/60-1/200 Sec.). Right away there was a noticeable difference when hand holding the camera (I could not get a single usable image, by far the worst shutter speed was 1/160 sec, followed by 1/125 sec. Even some of the tripod mounted images showed this flaw, the worse again being 1/160 sec. With the VR turned OFF I was able to get sharp results at that same shutter speed when mounted on a sturdy tripod. I was also able to get usable images hand holding at much slower shutter speeds (1/20 Sec-1/40 sec) with VR turned on, than I was at 1/160 sec. I am using a D300s for these tests, and this was an issue that could be easily repeated, or so I thought…
After doing some more “real world” shooting this evening, I’m unsure where I stand on this issue. This evening I was able to make very sharp images at 1/160 Sec. @ f/4, ISO 1000-1600 on my D300s, the very same shutter speed I could not make a sharp image the previous day. I should also mention my serial # is in the 208XXX range, so the firmware “fix” should have been updated on my lens. I’ve uploaded some 100% crops over on my website and you can have a look for yourselves:
http://www.ZackClothierPhotography.com/-/galleries/nikon-300mm-pf-samples
It does appear that there is some marginal loss of sharpness when hand holding with VR ON at those shutter speeds, see first photo comparison, (approx. 1/80-1/160 Sec) but not enough for me to be concerned about.
Anyone have any ideas other than user error, which at the moment, I am chalking it up to. I am really trying to love this new lens.
There may be a technical explanation because the firmware fix partially corrected the problem but I suspect only a Nikon insider would have this information.
I have seen double images with this lens using my D810.
Did the firmware update fix it for you?
After thoroughly testing a post firmware 300/4E PF VR at 1/160 sec with/without VR engaged, I found true as claimed by Nikon that VR images were generally a it sharper than most but not all images shot without VR activated. In fact, as shutter speeds decrease to as low as 1/20 sec, VR engaged sharpness was even better compared to higher speed like 1/160 sec. The reason I suspect is because it takes a few milliseconds (ms) for VR to stabilize any additional image motion after the shutter button is released. This will cause the “image doubling” for speed like 1/160 sec (6 ms) but little perceptible doubling at speed like 1/20 sec (50 ms). Hopefully in the future, firmware fixes of the lens or camera will be developed to reduced further the response times for the VR feature.
I agree all VR implementations require some ‘settling time’ for the gyros to lock in; however the problem with the 300/4 occurs regardless of the amount of ‘settling time’ – which is not normal.
I agree because I was careful to not fully depress the shutter button until I was sure the viewfinder image settled down. I wonder what’s happening in brief blackout time while the mirror goes up and the shutter fires? Is there some “fuzzy logic” that needs to anticipate image position?
Alan, thanks for your findings to bring some more light into the issue. Just for curiosity, how did you figure out the 6 ms and 50 ms for the shutter, I assume you can read off from some specs or white papers ? Anyway, it makes sense and what I´m hearing is, even the post firmware doesn´t completely fix the issue. I was desperately waiting for a Nikon 300/f4 with VR but there still seems to be no good alternative in that segment.
It’s not based on any data only widespread reports of vibration issues with 1/160 sec (6 ms) and far better vibration suppression at 1/20 sec (20 ms). However after more thought, It can’t be true because all of my shots were taken after the VR had time to engage.
It’s based on any data only widespread reports of vibration issues with 1/160 sec (6 ms) and far better vibration suppression at 1/20 sec (20 ms). However after more thought, It can’t be true because all of my shots were taken after the VR had time to engage.
It’s based on widespread reports plus my experience of vibration issues with 1/160 sec (6 ms) and my experience of far better vibration suppression at 1/20 sec (20 ms). However after more thought, It can’t be true because all of my shots were taken after the VR had time to engage.
oh ok I see, thanks for the update anyway. Seems like we have to live with that limitation even after post firmware, and hope for future updates.
To Nikon’s credit, the VR works as advertised in that image blur from vibration is definitely reduced (but not eliminated) compared to no VR. What I found interesting was comparing shots with VR compared to no VR at 1/160 sec as seen here: http://photo.net/nikon-camera-forum/00dIJq
On average with speeds of 1/100-1/200 sec, images are sharper with VR compared to no VR but not as tack sharp as VR switched off under very steady hand held shots. That’s because the VR induces a slight baseline doubling effect when activated at these marginally fast speeds. All VR shots were acceptably sharp at 1/160 compared to only about 25% that were as sharp or sharper without VR. Speeds like 1/250 and higher work fine. I agree that eventually there may be another firmware fix but for now I wouldn’t let the less than perfect VR issue stop consideration if you want a compact sharp 300/4 because the VR is still quite beneficial at longer speeds that could never be made without use of a tripod or other form of support.
Thanks again for your your insights. The test shots are impressive, especially because you´ve even used a D750 as the issue has been officially reported only for the D8xx by Nikon. I agree this may give potentially another thought for reconsideration..
Thanks.
I’ve had the exact same results, glad to see I’m not alone.
Shooting with the D750… I can’t get absolute critical sharpness below 1/200. Better than VR off. Viewed at normal sizes, the images look pretty nice. But if doing any big crop, large print, or pixel peeping, you see a little bit of image-doubling blurriness.
My testing:
http://www.dpreview.com/forums/post/56026416
http://www.dpreview.com/forums/post/56026416
It’s slightly disappointing in a lens that is otherwise just so impressive. My take-home lesson is to avoid relying on the VR below 1/250 if you’re going to need critical perfect sharpness.
From what I’ve seen testing my post-firmware fix lens, absolute vibration free images are difficult if not impossible to achieve with VR switched on at speeds of about 1/100-1/200 sec due to an unexplained phenomenon but not significantly present in slower speeds like 1/20-1/80 sec. Thus, I feel VR can be used with little or no reservation at these slower speeds in obtaining keepers that otherwise would require some form of support like a tripod or beanbag. In summary, VR should be used at any speed where only a single frame opportunity exists for a subject (assuming no tripod support). OTOH, if shooting 1/100-200 sec and you can squeeze off at least 5 or 6 frames in continuous shooting mode with VR turned off, chances are excellent that one of these will be critically sharper than with VR switched on.
The biggest question is, why does it only occur on this lens?
Apparently Nikon has acknowledged the problem and there is a fix. Unfortunately the lenses have to be sent to Nikon for firmware repair. http://nikonrumors.com/2015/04/16/nikon-offers-firmware-update-fix-for-the-vr-issues-found-in-the-nikkor-300mm-f4e-pf-ed-vr-lens.aspx/
This says that production was stopped so hopefully when stocks are replenished all are free of this problem.
They also say there’s no guarantee the firmware patch fixes anything either…
So here comes Nikon with there response:
https://nikoneurope-en.custhelp.com/app/answers/detail/a_id/64211
Send it for an update.
“Firmware for lenses with a serial number of 205101” already updated.
Not very satisfying as usual. It will probably bring another class action suit too.
There’s also a rather worrying line which states that it doesn’t always solve the problem…
What bothers me is that Nikon will not pay for shipping, they will not pay or compensate their customers for lost income and the fact that those customers will be without their rather expensive lens for a number of days/weeks. They might as well spit on us as well. 200 000 batteries recalled in 2013, oil and dust on D600, flare on D750, problems with D7000 and now this. Do they have QC department any more ? I am Nikon shooter but not for long any more. I am just tired of their attitude towards customers. And yes I do know that cameras today are complicated, I just don’t care nor should I. My job and my life are “complicated” but Nikon still wants my dollars without any consideration so I don’t see why should I show them any.
Sadly, I don’t think there are any companies that haven’t released something with some major QC issues in recent memory. Again, almost none of them will fix it, either – Olympus is still in denial about the E-M1 and E-P5, and wants you to buy a new camera instead or live with a half-baked patch that cripples as much as it solves the problem. Leica ships products with major flaws (who else do you know of with cracked sensors in several model lines?) and then when they run out of parts, makes you pay to upgrade – even if you don’t want to. Perhaps the only good guys left are Zeiss, but sadly they don’t make camera bodies.
From DPREVIEW in response to the Nikon advisory and firmware fix:
“We are very curious about this news, since in our testing (see below) we have found this issue with several of Nikon’s VR-equipped lenses, not just the new 300mm F4. And it’s not only Nikon – preliminary tests of Canon’s new 50MP EOS 5DS cameras also show issues at certain shutter speeds when using lenses with optical image stabilization without any sort of Mirror Up pre-delay. It is unclear whether this particular problem is unique in some way to the 300mm F4 VR, but we’re speaking to Nikon and will update this story with more information as and when we receive it.”
Indeed. More concerning is the fine print on the Nikon site that the FW update doesn’t always solve the problem, and in their own words, “Depending upon the way the camera is held when shooting, or the shooting conditions, images may be blurred even after this lens firmware is updated.”
I finally received my lens yesterday and immediately mounted it first to my D810 and then to my D750. I was sorry to find the very same VR problem that has been reported by others since this lens’ first day of release! At moderate shutter speeds (1/80 sec., 1/160 sec., 1/200 sec.) my handheld shots with the VR OFF were much sharper than the shots taken with the VR ON! The VR actually introduced significant blur and ruined EVERY shot at these shutter speeds. This was not just an occasional fluke … this was a very repeatable and startling condition! At slower or higher shutter speeds the VR functioned normally.
I took shots using these moderate shutter speeds of many subjects … taking a series of 4 shots with VR off and then 4 shots with the VR on (and vice versa) then viewed the results at 100% using both Nikon View NX2 and Lightroom 5. In every series the VR ON shots were very noticeably blurred (unusable) and the VR OFF shots were, without exception, very noticeably much sharper. Incidentally, most of the shots taken handheld with the VR OFF at 1/200 sec (and many at 1/160 sec.) were tack sharp! The lens is so short, light, well balanced and comfortable on both the D810 and the D750 that I was able to capture tack sharp images handheld with the VR OFF at these shutter speeds. I can’t say that I could do that with my 300mm F2.8G ED VRll!
I called Nikon and reported the problem. Later I received an email from Nikon saying that the Nikon Tokyo factory was “investigating this very closely.” I really wanted to love this lens but sadly, I packed it up and returned it to B&H (who were very professional and accommodating). If Nikon finds a cure for this VR problem (and I sincerely hope they will) I will be the first in line to buy it again.
Ok, good to hear that they are “investigating” the problem. Next step is that they quietly start repairing the lenses. Or (even better) publicly admit the problem (as the did with D750).
As a further note, with V1 and FT1 adapter 300mm PF VR is super effecient. I gues this is one of the more lucky combos… The lens also resolves well enough to work with this small sensor. I gues it would be ok even with the niw higher resolution Nikon 1 cameras.
For the time being there’s a workaround. Just put a grip under your D810 or an L-bracket.
My copy has the VR issue as well. With grip or bracket it’s gone. It has to do with the vibrance of the body during shutter release or so, I’m not a technician. I’m told to expect a solution within weeks.
Neither grip nor L bracket made any difference for me.
Sorry to hear. Both options are working for me (and a couple of others as well).
I was using a RRS bracket on my D810 when I experienced the VR problems.
Looks like shutter shock. Have you tried if it the ghosting double picture turns 90 degress if you rotate to portrait mode and takes a picture?
Agreed, except it doesn’t occur with any other VR lenses or bodies. Looks like some strange EM interaction between VR components and shutter components. Didn’t try rotating to portrait, and no longer have access to a lens – interesting thought…
I have similar problems with Df & D600 (I use mostly Df). The problem area is between 1/100-1/250 sec speeds, 1/160 sec being the worst one.
I almost sent the lens to service but hesitated because I was not sure that they have the cure. I have my fingers crossed that firmware change is enough. I am perfectly happy with the optics of my lens, would not like the service to physically repair my lens.
I don’t have this exact same issue as I can get terrific shots 1/250s and above, handheld, VR ON, with and without TC1.4eII when mounted on a D810. No problems at all with a D300s. I have emailed NIkon with my issues and received a swift response. They said they were aware of the issue as some users had reported similiar findings. They are currently examing the issue and it’s cause and will provide further information once all the facts have been established. All I can say is that if you are unlucky enough to have this issue with your lens, email Nikon support with as much information as possible. Despite this issue I still believe it’s a fantastic lens, I hope this issue can be quickly solved so it can reach it’s full potential.
Sorry, your message was a bit confusing: what issue did you have with yours?
Sorry, just to clarify, I can’t get acceptly sharp images below 1/200s when mounted to a D810 with VR set to NORM. 1/250s and above no issues at all. The difference is stark. VR OFF my results are as one would expect with a lens of this focal length. With my D300s I have managed to get good results down to 1/60s VR on. I also own a 200-400f4 VR and the VR is far more effective on my D810 when shooting at 300mm, 1/200s and below.
Thanks for the clarification. That sounds much more like the problem I’m seeing…
Ming,
I was on photo trip with a rented 300 PF and another person had his own PF. We saw no problems like this on the D810, D800e, D600, and D7200. As an aside, the D7200 seemed to have abnormal inconsistencies with focusing at long distances using both copies of the lens but that could have been a camera unit specific issue.
Louis
Sounds like there are definitely good and bad batches out there, which is too bad, because it appears to be pretty amazing if it works properly.
Did you use a tripod for samples with VR ON? I can shoot at 300mm with 1/15 and VR ON without problems, handheld and the imagine is sharp at 100% crop.
With and without tripod. Same result with all three lenses and bodies.
http://cameralabs.com/reviews/Nikon_Nikkor_AF-S_300mm_f4E_PF_ED_VR/
https://photographylife.com/reviews/nikon-300mm-f4e-pf-ed-vr
I’d like to have this lens work + TC-14. It seems everything Nikon does is a PITA.
great set of links:
Gordon over at Cameralabs is clearly seeing an issue, though not as severe as Ming. While Nasim seems to have no problem with the VR. Both reputable reviews so perhaps there are good and bad copies out there.
It really is amazing how bad Nikon can be when they are bad, it seems like they never learn.
BTW, I’m using the TC-1.4 with it and it really makes for a winning combination … if your lens works right.
you may want to look at this dpr thread http://www.dpreview.com/forums/post/55548469 could be the same issue
Grip made no difference.
Just took delivery on this lens and tried it out on my D750. There was definitely image shake degradation when hand held with VR turned on if shooting at 1/125-1/250. Interestingly at 1/20 hand held, the image was tack sharp with VR turned on It’s almost as if VR becomes progressively less effective as the speeds go up, or worse, maybe it’s inducing image shake on it’s own at medium speeds.
Mine has the problem — just as shown in your images — when shot at 1/60 or 1/125 second on my D810. Images shot below or above that shutter speed range are sharp. I bought mine in the US and received it about a month ago. I informed Nikon repair facility about the problem a couple weeks ago. They weren’t aware of it. Otherwise a great lens, but this is obviously a serious problem that Nikon must fix.
Thanks for the update. Did they offer a solution?
They were not aware of a solution, but asked me to check back. Also, to be more specific about the problematic shutter speeds, I only checked sharpness at 1/30 (sharp), 1/60 (double images), 1/125 (double images), and 1/250 and faster (all sharp).
Hmm, this is sounding more and more like a resonance issue along the lines of the E-M1…
As another data point, a local store received one of these lenses from Nikon in the last few days. I gave it a try with my D810. Same double image problem at same shutter speeds as with my copy of the lens.
😦 Looks like we’d better wait…
For what it’s worth, there was a blog discussion which claimed that attaching a MB-D12 grip to the D810 “solved” some of the VR out-of-focus issues. The attached grip apparently reduces the vibration.
Tried that, no difference. It is NOT a vibration problem because I had no problem getting consistently sharp images with VR off at the same shutter speed. There is some negative interaction between VR and shutter…
Yes same here with my new D810. On a a tripod with a 70-200 and shutter delay with VR on I get this ghosting double image. With VR off the image is unbelievably sharp.
i have the D810 plus the lens, have used all focus modes and with VR on and off, my images are all fantastic with non of this. It might be the lens as this would have showed up with other VR lenses
And the same problem with three different lenses on three different bodies, at least two of each from a different batch? At least we know there are some ‘good’ combinations out there…
Saw the exact same thing on a friend’s D810 image of product photography. Our teacher insisted he use Mirror lock up mode, and compared the two images. With it, a very sharp image. Without it, same image as the two images on the left on the top of your post: a strange vertical-shift double image. He used a zoom lens of smaller focal length than the one you mention, and it is a problem of the specific camera, and of many other full frames (as per our teacher).
No, it isn’t a mirror lockup problem. You are probably looking at very low shutter speeds. I know how to get a sharp image with and without mirror lockup. This problem is specific to this lens and camera: there’s is no way otherwise that I can get a sharp image WITHOUT VR at the same shutter speed, well above the handholding threshold (1/500+).
I have this setup, got the 810 because of your review. Got my 300 mm PF ED VR f4 preordered, as soon as I heard about this lens and got it right away. It doesn’t exhibit this problem. Just went out and tested it again. Since I usually shoot birds and use Nikon 1.4 teleconverter, I took the converter off and took shots. Viewed raw files at 100% in Lightroom and Capture. I can’t tell the difference center focus point in good light.
Sorry you haven’t gotten a good copy yet, seems strange.
My AFS 300mm f2.8 collects dust now. I love this lens. Just sayin …
Now suggests sample variation since there’s at least one good copy out there. I assume you have VR on all the time? Any chance you could try it without the TC and see if the results are still consistent VR on-off?
Sorry, I didn’t make that more clear. I went out today to double check. Hand held, with no TC on, in good light, I get pics that a pretty close to identical VR on or off. Like you said, VR snaps on like no other, so you really know when you have it on. Sport also works well.
I haven’t really done much low light stuff with this lens, so can’t comment about below 1/125 stuff, I’ll have to test it.
Thanks. You should see it even above 1/125; I was having issues from 1/20 to 1/750. Looks like you got a good one 🙂
Hi Ming, I’m betting I’m not the only one looking forward to hearing what Nikon have to say. Lets hope they don’t weasel out of an honest answer! (I love your posts by the way; always fresh and always interesting.)
Thanks. Well…they initially claimed there was no problem with their cameras, and have now gone silent after I sent them samples.
I suppose you’ve let the VR settle before triggering the shutter. I have seen similar results from other VR lenses when attemting a quick grab shot by quickly pushing the shutter through all the way. VR still moved the lens elements when the shutter opened.
But that’s just a guess. I am pretty sure you DID let VR settle. Plus at very high shutter speeds even VR element movement should not be captured.
Yes, I let VR settle for several seconds. I have owned enough VR lenses to know that this is an absolute must otherwise jumping will occur (and don’t have issues with any of my other VR lenses).
I’ve had this EXACT issue with my 24-85vr and have sent it in twice and have never been able to get Nikon to acknowledge there is an issue. The new 300 purchasers are more experienced, and there is maybe a higher percentage of defective lenses but at least Nikon will not be able to get away with saying it is a user error like they have been telling me about my 24-85. But so far they have not acknowledged anything.
I’ve had this issue on my D700 and D600. Not sure about my D750 (since I always use the 24-85 with VR off nose just to be safe) but since people are mentioning interference, for a while I though it was only malfunctioning when I was using a flash on camera but it could be that the double image was more obvious with a flash
Hmm, there are a number of components inside a flash that could create RF interference and possibly affect the gyros…
I had a similar problem with the 24-85 and my D800 as well. At medium shutter speeds with VR turned on, the images looked smeared, like the VR was doing too much. Turning VR off solved the problem.
After getting no satisfaction from Nikon, I sold it to someone with a D600, who was perfectly happy with it.
On the other hand, I’ve never had any of the well known problems with my E-M1. Go figure.
We win some, we lose some.
Is it possible there is an interaction between the VR, circles sometimes produced by the Fresnel element and the anti moire algorithm in the firmware of the D810?
Highly unlikely, since it just appears that the VR has made the image ‘jump’ in one direction rather than any concentric/radial pattern which we’d expect from something fresnel-related.
I hope you find a fix soon.
A search will give you plenty of similar findings.
A search gives all sorts of results, I wanted first hand information from readers, not forum speculation…
Nikon was an synonym for first class quality for decades! I seems to be, that this time is over now! I guess, NIKON gets pushed very hard with this new mirrorless technology! Even an increasing number of your pro colleagues, are switching officially their camera systems, for good reasons!
I had mirrorless and landed up going back because none of the products are mature.
VR is like a complicated setup of gyros and motors trying to detect and counter movements. Often VR will be inferior to a tripod alone.
Agreed, but this is not ‘slightly worse than tripod’ – this is CLEARLY worse than handheld at the same shutter speed, and with a distinct double image rather than just motion blur resulting from camera shake. It’s also much worse than results from other VR lenses and doesn’t happen on the D800E or D3…
Yes this is a bad model. They could fix in firmware probably. I had a Nikon 300mm AF-S f4 impecable. Just big. Been having a lot of fun the 14 oz Olympus Stylus 1 300mm 2.8 with nice EVF. Does not project outwards when open. Fast AF. Similar to Nikon 70-300 VR.
A guess, electro-magnetic interference?
That would be my hypothesis too.
Reading the specs I see it has an ‘electromagnetic diaphragm mechanism’ for aperture control.
My binary brain wants to think it should cause problems at all shutter speeds, or be clean across the board… but I’m not an engineer. I don’t design lenses, I just use them for photography. 🙂 (maybe the diaphragm has some sort of variable E-M output depending on aperture, thus affecting a particular shutter range?)
Regardless, this is a minor bummer, as I planned on buying this as my first “you paid **how much** for that lens?” prime.
The diaphragm action should not affect the shutter – they’re independent variables.
But yes, for anything this expensive – it leaves me scratching my head.
Is the problem still there with electronic shutter or video stills?
Yes.
I should have written lensrentals.com
Most of the negative reviews have mentioned poor VR results in the 1/80th to 1/160th of second shutter speeds. Slower speeds giving better performance, as well as higher speeds. But apparently not everyone is getting bad results, so reviews are mixed.
Tried that; shutter speed doesn’t seem to make much of a difference.
I have seen reviews that mention the problem, such as one done by lensrental.com
No, LensRental didn’t talk about a VR problem. They said their one and only sample was slightly decentered. Not the same thing at all. Even slightly decentered, the results were still excellent (as good as the older 300/4 AF-S, which is optically excellent). They have now reported that a second sample they received was perfectly centered and that testing showed comparable center performance and slightly better edge performance compared to the first sample.
This is how misleading information gets spread. People report rumors or misunderstanding as fact. It pays to go back to the source.