Turn off VR when using a tripod and exposures longer than about a second - initial data

<p>I have heard the dictum about turning off your VR when tripod mounted many times. To be honest, it sounded odd that turning *off* VR would produce sharper pix. Yesterday, I finally got around to testing it. Conditions:</p>

<p>- d700 @ very low ISO, using full frame<br>

- 24-120 VR @ 120 mm<br>

- f11<br>

- 10 seconds<br>

- Lights = ambient tungsten<br>

- color temp setting 2500K<br>

- Gitzo medium tripod<br>

- distance ~ 15 feet<br>

- Both images shown are moderate crops ( 1/3rd linear, 1/9th in area) of the full frame. The lower lock in both crops is just about the dead center of the full frame.<br>

- Processing = absolutely identical for both, ie, a bit of contrast, downsizing to 700 pixels wide for photo.net, and my usual sharpening after downsizing.</p>

<p>The image attached to this post is with the VR off. The image attached to the next post is with the VR on.</p>

<p>Tom M</p>

<p>PS - Before posting this, I searched PN to see if someone had previously conducted such a test and posted the results. I couldn't find anything. Hopefully, I'm not reinventing an obvious wheel. ;-)</p>

<p> </p><div></div>

<p>This post shows the results with the VR on. Everything else is identical.</p>

<p>The effect is obvious: There is a secondary image displaced to the left and down from the primary image. It looks somewhere during the 10 sec exposure, the lenses in the VR system of this lens finally decided that they really didn't know what to do, so they just drifted off towards the maximum compensation values in the SW direction, and then stayed there for the remainder of the exposure.</p>

<p>Obviously, other VR I lenses, and especially the newer VR II system will likely behave differently, so this test doesn't speak to them. However, it would be very interesting to know if the benefits of turning off the VR system are present at shorter exposures, say 1/10th and shorter, ie, exposures you might consider taking without using a tripod. If I get around to it, I'll do some more tests at shorter exposures.</p>

<p>The results of my little quick and dirty test caused me to wonder about the reports of absolutely terrible optical quality of the 24-120 VR lens. It seems like the reports about this lens are much more polarized than I would expect, some people love it, some think it's cr*p, with few people in the middle. Since the people concerned about lens sharpness are most likely to use a tripod, I started wondering if at least some of the very negative reports about sharpness with this lens might have arisen because some of the folks doing tripod mounted tests with this lens may have simply forgotten to turn off the VR system during their tests.</p>

<p>In any case, the bottom line is: Don't forget to turn off VR when tripod mounted ... or at least take a couple of exposures, one with it on, and one with it off to get a feeling for when it stops helping and starts degrading the performance of your lens.</p>

<p>Cheers,</p>

<p>Tom M</p>

<p>PS - Don't worry, the tripod wasn't bumped during the second exposure. I repeated the test several times, getting the same result each time.</p><div></div>

<p>I don't see the second image.</p>

<p>Tom, may I ask how many test samples did you try for VR on and VR off? For example, if you tried 5 times with VR on, did you consistently get such blurry results all 5 times?</p>

<p>While I don't have the 24-120mm VR lens, intentionally or unintentionally, I have used the 200-400mm/f4 AF-S VR on a tripod with VR on, and I have never had such serious problems.</p>

<p>Shun, I realized this was an obvious question that I did not address. I repeated the test three times (and inserted a PS to my 2nd post to address this).</p>

<p>Tom M</p>

<p>Shun, were you looking for the effect in exposures in the 1- 10 second range? I'm forming a suspicion that the VR system (especially in early VR lenses like the 24-120 VR) can hang in there for a good fraction of a second, but then gives up and rapidly drifts off to its maximum compensation values in each direction. Thus, if your exposure is shorter than say, a half second, you won't see the effect because it hasn't drifted off yet. OTOH, if the total exposure time is much longer than say 10 seconds, the fraction of the exposure time that the VR system is at one of its limits will also decrease. You will note that my secondary image is considerably weaker than the primary image, and there is not a continuous blur or streaking of the image between the two positions.</p>

<p>Obviously, it would be great if someone else could confirm my test results on this lens (... or others).</p>

<p>Cheers,</p>

<p>Tom M</p>

<p>One more reason to trade in my 24-120, but, just for laughs 'n giggles, I'm going out to take a few h/h with the VR O-F-F and O-N. I improved it a little bit with manual focus, but didn't think to flip off the VR.<br>

This is good info. Thanks for posting!!!!</p>

 

<p>VR is not a target-sensing feedback system, rather a single axis tilt sensing system regardless of target location. As such, it can only make moment-to-moment corrections based on its previously sensed position about the lens' rotational axis.</p>

<p>A more sophisticated VR system would be target-sensing with dual axis compensation which I imagine will be the future.</p>

<p>Tom, plenty of Canon IS and Nikon VR super tele users intentionally switch on VR (or IS) to help stabilize their long lens on a tripod at 1/60 sec, 1/30 sec. They report that it makes a noticable improvment.</p>

<p>If you are only talking about 1 to 10 seconds time exposure, I think you should clarify that in the subject. I have no idea about how VR should behave under such long exposures. But the current title for this thread does not specify time exposure. As a moderator, I can update the subject.</p>

<p>Assuming that Tom had his tripod's head locked, then both his and Shun's experiences are consistent with Nikon's instructions:</p>

<p>http://support.nikontech.com/app/answers/detail/a_id/7676/kw/tripod/p/81/r_id/116678</p>

<p>My own experience of once unintentionally leaving VR enabled on my 70-200 while on a locked tripod is consistent too (blurry photos until I noticed and then disabled VR).</p>

 

<p>Hi Sam - I didn't realize you were one of the moderators. Thanks for the offer to update the title of the thread. I would suggest changing the title from:<br>

----> "Turning off your VR when using a tripod is not a joke!" to<br>

----> "Turn off VR when using a tripod and exposures longer than about a second - initial data"</p>

<p>I initially thought about such a specific title, but was hesitant to do so because thusfar, I have no real data to support specific time limits. However, you are correct in that the current title sounds too general. I think that if, in the future, anyone is attracted to the thread with the new title, and then reads the thread (including this post), they will realize that this is a work in progress and not the final recommendation on the subject, so nothing will be lost by making the title more specific.</p>

<p>Thanks again for your offer & services.</p>

<p>Sincerely,</p>

<p>Tom M.</p>

<p>Subject updated.</p>

<p>Bill C - Yes, my tripod head was firmly locked.</p>

<p>Michael Chang - You are absolutely correct in your statement that the current Nikon VR system is not a feedback system, but is rather little more than an estimator for the change in angular position of the camera over some time period. Suitably scaled in amplitude (ie, depending on the FL of the lens), this estimated value (the change in angle) is then applied to the image shifting optics in the lens. Thus, it is more akin to a feedforward system.</p>

<p>If my hypothesis is correct, Nikon set the lower frequency bound for the system too low in VR-I. Because of (the well known) noise and drift in the accelerometer outputs, over periods of several seconds, the system's estimate for the new angular position of the camera drifts away from zero even when there is no real input to the accelerometers (ie, no real change in angle). I suspect that the required tweak in the transfer function was extremely simple (in either hardware or software) and probably forms a good bit of the difference between VR-I and VR-II.</p>

<p>Tom M</p>

<p>PS - Hopefully, there are photo.net folks around world mounting their 24-120's (and other VR lenses) to tripods and clicking away right now to verify this. :-)</p>

<p>VR is quite able to destroy image sharpness at speeds lower than say 1/60 sec. So is the infamous "long lens technique".</p>

<p>Tested on various long Nikkors including the 400/2.8 VR, all lenses mounted on really sturdy tripods (Sachtler ENG 2 CF and CF HD, Burzynski or Video20 head).</p>

<p>A point in case is that Nikon in the paperwork for the new 70-200/2.8 VR Mk.2 explicitly states that VR should be OFF when a tripod is used.</p>

<p>While empirical data is what most users are concerned with, understanding Nikon's technology will help correlate such findings with its design.</p>

<p><a href="http://www.google.com/patents?q=nikon+vibration+reduction&btnG=Search+Patents">Link to a list of (full-text) patents granted to Nikon's VR design.</a></p>

<p>Hi Michael -</p>

<p>Your point about understanding the underlying technology is right on the mark, and I, for one, do love technology.</p>

<p>However, I have served a 5 year term on the patent review board of a major lab (> 1700 Ph.D.s), and, speaking from experience, I am not jumping for joy at the prospect of having to go through each of the 30 patents in the list that you Googled. By any chance, have you gone through those 30 patents, and determined which were the most relevant, say, to the 24-120 VR? </p>

<p>;-)</p>

<p>Tom M</p>

 

<p>For whatever it's worth, the photos I alluded to earlier, that exhibited motion blur with the 70-200 with VR enabled while on a locked tripod (Gitzo 1325 with Arca-Swiss B1, and MLU and a remote release), had shutter speeds of 1/1.3, 1.6, and 4 seconds.<br>

<br />But, not all the photos at similar shutter speeds, with VR enabled, exhibited blur --- it was hit-or-miss.</p>

 

<p>My results with my 70-200 and 200-400 match Nikon's advice perfectly. The old 70-200 should <em>never</em> be used with VR and a locked down tripod. Sometimes, VR1 makes things worse even on a bean bag or hand held against a solid object.<br>

The 200-400 may not have this problem, but when I am shooting at 400mm (560mm on DX), I lock down, turn off VR, put a heavy bean bag on the lens, and go to MUP or shutter delay. </p>

<p>VR is designed to give you 3 to 4 extra stops of shutter speed. If I'm shooting with a 200 mm lens without VR, I would want to use a minimum shutter speed of 1/250 or an ideal shutter speed of about 1/500. VR gives me 4 stops over this ideal shutter speed.</p>

<p>1/500<br>

1/250 (1 stop)<br>

1/125 (2 stops)<br>

1/60 (3 stops)<br>

1/30 (4 stops)</p>

<p>With VR, I can shoot with a 200 mm lens at 1/30th of a second and camera shake should not be a problem. I've tested this many times, and it works as advertised.</p>

<p>VR isn't designed for 10-second exposures. 3 stops faster than 10 seconds is about 1 second. Can you hand-hold a one-second exposure? I can't. On a tripod with a long exposure time, the VR system is going to lurch around periodically. I'm not sure why, it just does. This isn't the type of shot that VR is meant for, so the results of the test are not surprising.</p>

<p>Ah, HA! So this is why they put the "VR Off" switch on the lenses! Crafty buggers, expecting me to actually read the microscopic printing in the Instructions sheet.</p>

<p>ha. Mark I like sarcasm.</p>

<p>Jose, More a goof on me than anything else. But I was hoping that turning off the VR would make a sufficient difference in my 24-120G's IQ. Even dug out my glasses and read through the constructions. But alas, my particular copy has stayed soft (never approached Tom's crispness) so it gets traded in on Tuesday. I also kick myself for buying the kit (thinking "Nikon glass is NIKON glass") without researching the lens. Between what I get for it in trade and what I paid two months ago, I could have saved myself $250!</p>

<p>Mark -</p>

<p>Fortunately, my particular 24-120 seems quite sharp except at the extreme corners, especially if you stop it down a bit from the maximum aperture at a given FL. To me, it looks about as good as my 18-200 VR DX, which I also hand-selected. I have used my 18-200 on my d700, but I hate throwing that many pixels out the window, and there just aren't that many choices for walk-around zooms with VR in the FX format.</p>

<p>To decide on the 24-120, I took test shots with three different 24-120's at my local camera store before I settled on this particular unit. There were considerable differences in corner sharpness between the different units. On the first unit I tried, the VR just plain didn't work! On the second unit, the upper LH corner was wildly more blurred than the opposite corner.</p>

<p>I hope that these unit-to-unit differences were due to QC variations when the lenses were being built, but once built, won't fall or get blurry after a year or two. We'll see.</p>

<p>Normally, I buy most of my gear from B&H, but in this case, knowing the reputation of the lens, I bought locally and paid a significant premium above the B&H price, but was able to decide on a particular lens on the spot, without a bunch of returns by mail. That's one of the benefits of supporting your local store.</p>

<p>Tom M</p>

<p>Could it be that the slap of the mirror causes the VR to go berserk when the camera is on tripod? Was MLU used or not?<br>

Is this effect noticeable at shutter speeds that are not "off limits" for VR, like 1/60 or 1/30 s?<br>

I think I recall some of the longer Nikon VR lenses are supposed to work OK on tripod with VR on.</p>

<p>Hi Sem -</p>

<p>Mirror slap could not have caused the effect seen in the images I posted. I did not use MLU, but since my exposure was for 10 seconds, and the camera was mounted on a good solid tripod with a seriously large ball head, any vibrations from mirror slap would have (a) been initially minimal, and (b) would have been damped out much more quickly than the 10 seconds of my exposure. In addition, mirror slap would have caused an oscillatory vibration of the camera, resulting in a smoothly blurred image, not a clear secondary image such as can be seen in my test shot.</p>

<p>However, as you point out, at speeds in the (roughly) 1 sec to 1/60th sec range, mirror slap surely could cause problems, especially with long lenses.</p>

<p>Finally, yes, for some of the longer Nikon VR lenses, you are indeed supposed to leave the VR on, even when on a tripod. For a list of them see the link posted earlier in this thread.</p>

<p>Cheers,</p>

<p>Tom M</p>