Nikkor 300/4 PF VR firmware fix: sample image comparisons to pre-firmware fix version

<p>Having returned my first 300/4 PF VR for refund that demonstrated insufficient VR compensation at speeds like 1/160 sec, I recently took delivery on a later version with the firmware fix (after ser. # after 205101). The problem has improved by about 50% but not completely corrected. In fairness to Nikon, they stated: "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". As with the first lens, VR was a better performer at speeds like 1/25 or 1/40 than say 1/100-1/180. Sample VR images as follows. Here is the first lens @1/100 before the "firmware fix".</p><div></div>

<p>Post firmware fix @1/100</p><div></div>

<p>Pre-firmware fix @1/160</p><div></div>

<p>Post firmware fix @1/160</p><div></div>

<p>Pre-firmware fix @ 1/200</p><div></div>

<p>Post-firmware fix @ 1/200</p><div></div>

<p>Lastly, no VR at 1/100</p><div></div>

<p>Could you remind me which body you're using this on?</p>

<p>Out of curiosity, if you took the with Fixed VR @ 1/160 a couple of times do the movement 'smears' have the same angle, ie your movement, or is there a more even 'fuzz' of VR prism shake? I'm wondering if it goes smear, smear, fuzz, fuzz, smear, smear as you pass through the danger zone speed-wise? Kinda if the vibration harmonic is body weight/lens weight dependant. Not brilliantly explained, but I hope you get what I mean!</p>

<p>Shot with a D750 and camera was held horizontally on all shots. Fairly consistent vertical double image effect rather than random movement blur as when the VR is turned off. More shooting down the road will determine if the same occurs when the camera is held vertically.</p>

<p>thank you for this.</p>

<p>does not look good, at all...</p>

Alan, thanks for the samples. So the firmware update affects the D750 also, not only the D800/D810 as Nikon suggests. Are your images pixel-level crops, or what percentage of the frames are they?

 

My typical application for that lens would be birds in flight or outdoor sports. Shutter speed is probably 1/1000 sec or

faster with VR off. 300mm maybe a bit short for those purposes. For indoors, f4 is on the slow side.

<p>Now I have a headache. This is not impressive at all. Honestly, I get better results using my old MF 300/4.5. I have found VR to be helpful at times with slower shutter speeds and shorter lenses but it isn't my favorite feature.<br /><br />Rick H.</p>

<p>Thanks for posting these images. It seems the improvement due to the firmware fix is clearly visible, though from such a small number of images it is not possible to get a complete picture as hand-held shots with or without VR tend to vary in sharpness from shot to shot. I assume you posted "typical" results that you've observed.</p>

<p>With VR the goal of the designers was to reduce vibration due to hand shake to allow slower speeds to be used hand-held than otherwise possible. However, the lens VR performance is evaluated with CIPA standard tests, which involve evaluation of fairly low resolution images.</p>

<p>http://www.cipa.jp/image-stabilization/documents_e/DC-X011-2012_E.pdf</p>

<p>If the circle of confusion used to determine acceptable sharpness in testing the performance of the VR system is set to 63 um as in the CIPA draft, that's about 10 pixels in a D750 image. Thus an image that has blur circle of confusion smaller than 10 pixels is regarded as acceptable in terms of its sharpness for the purposes of the CIPA testing of VR performance. With this in mind, it is not surprising why a lot of photographers report less benefit from the VR system than the CIPA rating in the lens specifications, as they expect the images to be sharp to the level of individual pixels, which may not necessarily be the goal of the VR system designers, or at least the standard that was devised to evaluate VR performance. (If someone is intimately familiar with the testing method please comment. I just had a quick look at the document.) I think the actual designers of the lenses do aim to make the vibration reduction as effective as possible, but measurement of acceleration / angular velocity has finite precision and compensation is imperfect as well, so in practice VR is closer to achieving the "good enough for small prints" goal than "get the very best sharpness out of a given lens". Personally I can accept that, and use VR where it contributes positively, but otherwise I typically keep it off.</p>

<p>For me, most of the applications for a 300mm lens call for the use of shutter speeds in the ball park of 1/500s or faster (in case the subject is clearly active, such as a walking person) or for relatively stationary, but still living subjects (e.g. someone playing or performing at a concert) in which case 1/320s or 1/250s should still give a good results with some consistency, and occasionally 1/200s may be needed because of low light conditions, but 1/125s or 1/160s are clearly within the range where the subject movement will blur the images with a high probability, and VR can't really cure that. With the AF-S 300mm f/4D, I typically hand hold at 1/1000s or faster, as even 1/500s would result in clearly reduced sharpness in the case of human subjects walking; they don't even have to be running. Thus if I'm able to hand hold at 1/500s and 1/250s and get good results, as I can with the 300mm PF, with VR on, VR does increase the practical usefulness of the lens, even without the firmware fix. However, of course, if I buy the lens, I want it to have the latest firmware so that it performs as well as it can, if only for those occasional 1/200s shots.</p>

<p>I'm wondering if any of the forum contributors here have the 300mm PF along with the TC-14E (any version)? I think in many outdoor stage situations the 300/4D by itself was easy enough to use, but with the TC-14E II, I could not get good results as there was simply not enough light in the stage that is in shade, especially towards the evening, and with the TC it was better to stop down to f/8 which was the final nail in the coffin. I believe from e.g. the cameralabs.com tests, that the PF version should give quite good center sharpness with the TC-14E III (which is what I currently use with other lenses, though not with the 300mm f/4D since they're not really compatible), but how is the VR performance with the TC attached? Does the presence of the TC shift the window of shutter speeds where the VR works well? With the 200/2 + TC-20E III, I got good results at f/5.6, 400mm, using shutter speeds of about 1/320s with VR on, and that setup is usable, but after 20 min of shooting that rig hand-held, my hands will shake like crazy. ;-) (Yes, I know it screams for a monopod, but I need to do it hand-held in some situations). Even though I felt the results at 1/320s were acceptable, the sharpness did visibly improve when the setup was mounted on a tripod, so there is clearly room for improvement, and hand-holding a heavy 400mm setup is not the way to get the best possible sharpness. At 200mm and 280mm it is much easier to use that lens hand-held, and nowadays I've had enough practice with it that the bare lens is easy enough. Now, what the 300mm PF brings to the table for me, is that it is likely to give better image quality at 420mm, and 300mm, and is extremely portable and easy to travel with. However, I don't expect hand-held results at medium speeds to be "tripod sharp" at these focal lengths, no matter what the lens is and however good its VR system is. It just isn't a realistic expectation, especially at 400mm.</p>

<p>My recommendation is not to judge the 300mm PF based on its VR performance alone, but consider the whole package. </p>

<p>Thanks for posting Alan. I agree with Norbert, this doesn't look good at all. You don't have a single shot up there that looks acceptable. Glad I held on to my AFS 300 f4.</p>

<p>It may look bad to some because cropping was around 100% on a subject about 50' away on single shot mode in order to demonstrate image detail with VR engaged. This is what you'd likely at 100% with the camera's LCD screen on playback. Uncropped, the image is far more acceptable especially if making a print. I don't know if shooting vertical camera orientation reduces the effect or more importantly for me, when shooting at continuous high for BIF. I wonder how their 300/2.8 VR does under similar circumstances, i.e. at 100% crop? Are these images rock steady as if mounted on a tripod or is the a little image degradation handheld on VR? All things considered, I'd certainly take this lens over my previous 300/4 AFS for size reduction alone. The VR feature would be invaluable shooting a concert handheld at speeds like 1/50 sec. because it delivers tack sharp images at these low shutter speeds.</p>

<p>To properly evaluate VR, you need to shoot a series of images, at least 10 but perhaps 20, 50 ... to see how well VR performs overall. Here we are focusing on a few shutter speeds that are known to cause ineffective VR.</p>

<p>As I said, to me, VR on a 300mm/f4 is not a critical issue. For action photography, the main issue is subject motion, where you are going to use a much higher shutter speed anyway. f4 is too small an aperture for serious indoor photography.</p>

<p>To me, the main "problem" for the 300mm/f4 PF is that it is difficult to justify its cost. Clearly it is small and light for a long tele, but even the old 300mm/f4 AF-S is very easy to hand hold and optically excellent. The main drawback for the old version is slow AF and a poor tripod collar, but at half the cost, it is far more cost effective.</p>

<p>If you want to use a 300mm under dim light, most likely you'll need a 300mm/f2.8 AF-S on a monopod or tripod. Otherwise, a 400mm/f4 PF lens similar to the DO lens Canon makes will really show the size and weight reduction of DO/PF, but the price is likely going to be drastically higher than $2000, also similar to the Canon lens.</p>

<p>I certainly agree with Shun that a 300/2.8 VR II would be preferable in low light to minimize blur due to subject motion as obviously VR does nothing to mitigate this issue. Additionally the f/2.8 version's VR goes down to 4 stops compared to the f/4 version of 4.5 stop gain so in this issue, the f/2.8 is still slightly ahead. I'm curious if the 300/2.8 images at for example 1/160 show a similar slight "double image" phenomenon as the f/4 version.</p>

<p>Interestingly, some users are now reporting on other forums that at least some of the VR PF 300/4E's require a printed circuit board to be replaced in addition to the firmware install, and this is all part of the service.</p>

<p>I think the VR 300/2.8 II would give a bit better results at medium speeds (i.e. 1/60s to 1/200s) because it is a much heavier lens and as such the rig is stabilized by the mass. As long as the photographer can handle the lens comfortably and don't start shaking a lot because of the weight. :-) However, I think most of the time with a 300/2.8 you'd want to use a monopod or tripod to hold the weight of the lens. This in turn imposes some limitations on how the lens can be used.</p>

<p>I'm surprised that firmware can affect the VR capabilities of a lens at all. That implies that there's some CPU control in the servo loop between motion sensing and prism activation. That, to me, ain't good. By the time the motion sensor output has been digitised and a small CPU has calculated the feedback needed, via some lookup table or whatever, then turned it back into a sequence of stepper signals, it's all too late and never going to be as good as a real-time analogue servo-loop. Or even a non-electronic gimbal system.</p>

<p>I also agree that a small sample of handheld shots is no true test of VR efficacy. By its very nature the shake produced by handholding is random in direction and variable in intensity, so there's naturally going to be a difference from shot to shot if the VR isn't 100% effective. Without a repeatable mechanical "shake rig" of some sort it's difficult to tell if this so-called firmware fix has been effective, and by how much.</p>

<p>It's not (in all cases) just firmware, several users have reported that Nikon told them that a PCB inside the lens was exchanged as well. A microcontroller or -computer runs the VR algorithm. The measured acceleration data has noise and the control feedback needs to filter this noise and estimate a (statistically) optimal value for the adjustment of the VR lens group position. Stepper motors are not used in VR systems, rather they are voice coil motors. Since it is digital, adjustments can be made to the algorithm by changing parameters of the filter and control system. There are also different algorithms used for the viewing / composition, and during the actual exposure. I think this kind of a system would be difficult to implement using analog electronics and it would be sensitive to slight variations in component parameters, temperature, etc. With an analog control system, every lens would be different in its behaviour, which is probably what the manufacturers try to avoid like the plague. My experience with an analog PID control circuit was very discouraging; it is much easier to adjust and optimize the functioning of a digital system and get it stable. If it uses an optimal filter, the control system algorithm does compensate for the lag between the measurement and control operations, to avoid systematically being "behind" in the control operations, but it cannot do this perfectly since the movement of the rig in the immediate future is impossible to predict with certainty (it can be estimated based on prior acceleration data and some statistical criteria).</p>

<p>VR systems have finite precision and they do introduce some slight blur of their own, but on the spatial scale of the corrected vibrations at slow shutter speeds, the effect is small. For sharpness on the spatial scale of individual pixels, use a tripod and/or fast shutter speed and turn VR off.</p>

<p>firmware.<br /> it got better, alright.<br /> with all that in consonsideration.</p>

<p>yesterday i was out taking photos of a grey heron at dusk.<br /> let me check my stats.<br /> iso 2.5k, f2.8, 1/100, -2.3 EV.<br /> switching to f4 would give me 1/60<br /> <br />half an hour before that i am at 1/160..<br /> so basically for one hour a day you are in the sweet spot of this lens. ha ha ha.</p>

<p>considering the entire package would mean to me, as it is now, this lens is not a worth consideration...at all!<br /> sorry..</p>

<p>a used 200-400, which might have its issues too and comming at a higher price and weight is a different topic,it however still seems to be a better choice overall. at least to me.</p>

<p>and no, i am not saying what lens it was as it is not mine and i now got reallllly long teeth...really...jeez.</p>

<p><em>so basically for one hour a day you are in the sweet spot of this lens.</em></p>

<p>I never could understand what the purpose of limiting photography to some specific time of day is. Direct midday sun can be harsh, but if there are clouds / fog / tree cover / building shade / rain / snow, the light is no longer harsh. If you do it to get sunset / sunrise colours, what about some variety? I think there can be such a thing as "too much of a sweet thing". Maybe there is some location specific factor which I'm unaware of. ;-)</p>

<p><em>sorry..</em></p>

<p>I don't see how anyone could be offended by another photographer's lens choices. ;-)</p>

 

<blockquote>

<p>yesterday i was out taking photos of a grey heron at dusk.<br />let me check my stats.<br />iso 2.5k, f2.8, 1/100, -2.3 EV.<br />switching to f4 would give me 1/60<br /><br />half an hour before that i am at 1/160..<br />so basically for one hour a day you are in the sweet spot of this lens. ha ha ha.</p>

</blockquote>

<p>ISO 2500?<br /> Sunny 16 tells us that at ISO 2500, you can shoot @ f16 and 1/2500 sec under the full sun. f4 is 4 stops brighter than f16, and 1/160 sec is another 4 stops slower than 1/2500 sec. 4+4 you are talking about 8 to 9 stops darker than sunny 16. I wouldn't bother to shoot when it is so dim unless you are shooting nocturnal animals, because the quality of light is simply too poor, or I would add some external flash. Needless to say, when you are down to 1/160 sec and 1/60 sec, you'll have all sorts of issues with animal subject motions, for one full hour. It is not a realistic setting I would use.</p>

<p>I captured the attached image last Sunday (May 3) at 7:15pm (19:15). We had early dinner that evening at 6pm, and we went back out to shoot after dinner. As the sun was setting, I managed to use auto ISO with the shutter speed limited to no slower than 1/1600 sec to freeze any motion, and it was still f4.5 and 1/1600 sec.</p>

<p>And my lens was a much longer 500mm/f4 AF-S version 1. There is no VR, which wouldn't be necessary anyway @ 1/1600 sec, and the lens on a big tripod.</p><div></div>

<p>dark trees, swampy water, white grey bird, after sunset, blue hour.</p>

<p>i am not talking sun.<br /> my setting was forced by light.<br /> period.<br /> if you have to go to iso 2.5k to get 1/100@-2.3EV @ f2.8 @ 400mm what do you do.<br /> animal stands still and barley moves...and yes..it is dark in between dark trees and there is a white bird.<br /> expose for the bright parts, lose all the detail there is to the dark ones...well..go somewhere in between and go for -2.3 EV instead of -4, which would have been good rendering of the white feathers, as the historgam suggested.</p>

<p>i do know what i am doing...<br /> and i do know about the sunny 16 rule, too</p>

<p>i do not like to shoot full out sun because usually it just does render stuff deifferently and therefore does not look<br /> good enough...i am looking for moody stuff. of course i do shoot fullout sun..i shoot everything..<br>

<br /> quality of light..blablabla whatever.</p>

<p>in any case...with the 300 f3 pf df whatever thingy the vr probably would have f-ed up the imaged.<br /> awesome.</p>

<p>considering a 70-200 instead would be my advice.get a 1.4 and images will be better.<br /> almost<br /> ..sharper</p>

<p> </p>

<p>Norbert, do you use tripods?</p>

<p>And if the animal is standing so still and you have time to prepare, do you have to use 1/60 sec while you know very well that VR on that lens actually works better at 1/30 sec? Somehow you just have to use between 1/60 and 1/160 so that VR gives you maximum problems (and you have something to complain about) but never 1/30 sec or 1/250 sec. And you also cannot adjust your ISO higher or lower to stay away from that "sweet spot" shutter speed range??</p>

<p>When I tested the 70-200mm/f4 AF-S VR and 80-400mm AF-S VR, I was amazed how well VR works at 1/20 sec on those lenses. (And now I own both of those lenses but not the 300mm/f4 PF.) Unfortunately, to me such great VR capability is not particularly useful as my subjects are rarely so still, especially my typical subjects for the 80-400.</p>

<p>In any case, Nikon does provide an optional tripod collar on the three lenses I mention here for good reasons. When you can use a tripod, it usually beats VR. Sometimes the environment doesn't allow tripods or monopods; sometimes we are just too lazy to use one.</p>