IS/VR adversely affecting image IQ

sk66

sk66

Advertiser
Messages
8,671
Name
Steven
Edit My Images
Yes
There have been many times where the use of VR has been discussed... and there has been debate over whether VR can "hurt" an image. My personal ROT is to turn it off if I don't need it because it can adversely affect an image. When VR affects an image it looks like anything from soft focus or slight motion blur up to a "double exposure." I decided to post some images showing the effect. All images were taken using the 400mm f/2.8 VRII.

In this test shot I had VR turned on in normal mode while on a tripod. Obviously this is "wrong"... I did it intentionally in order to magnify the issue. I used timed release so the only potential motion is from mirror/shutter slap. I did not use flash... there is no other possible cause for the "double exposure." SS was 1/4.

View attachment 33348

To show the difference here is a second test shot taken exactly the same as the first, but taken with VR turned off. You can see a little motion blur due to mirror/shutter slap, but it is much less of a double exposure type of look.

View attachment 33349

BTW, the center circle is 8mm in diameter on the test target...

For this very hard crop image I was at 800mm and 1/100 SS. The camera was *handheld* using my SharpShooter Camera Mount (rifle stock) in a kneeling position (i.e. supported by my front knee). Obviously 1/100 is way slow for 800mm, especially on the D810. One would think this is a situation where VR would be a good idea... I did, so I turned it on. Four stops of VR "gain" should have put me in a reasonably usable range. Again, I didn't use flash (I probably should have though).

View attachment 33350

The image looks soft, not surprising given how hard I was pushing my luck. But look at the lower edge of the beak... there's that tell-tale "double exposure" that tells me VR contributed to the softness.

Neither example is of using VR at higher SS's (above ~ 1/500). IME it looks similar but less distinct... it's harder to identify and it doesn't always negatively impact the IQ. When it does impact IQ it makes you think the camera slightly missed focus, or the SS was a little too slow.

The point of this is not to avoid using VR; the picture of the bird might have been even softer/blurrier if I had turned VR off. The point of this is just to use VR "correctly" and to turn it off when it shouldn't be on.

This is the full shot of the American Pipit

American Pipit by skersting66, on Flickr
 
My mind is two camps over Image Stabilisation (whatever 'flavour' is deployed). When I first owned a Nikkor 80-400mm (which I believe is Nikon's first use of the technology) I took a photo of the sole of shoe worn by a Bridegroom - he still had the price on it :) 400mm on a D100 indoors. 1/60 sec - worth a try from the rear of the building. You could read the price of the shoes. When I looked at the EXIF I'd misread the viewfinder info - it was actually handheld @ 1/6 sec! So, yes, it worked for me in this instance but I've had images which have not worked - as I say, don't know which way to fall with this.

Your tight crop is showing more in terms of chromatic aberration than a 'double image'. The purple fringing on the upper edge of the peak being a trifle disturbing.
 
I ballsed up a couple of times recently when shooting reindeer after a night shoot on a tripod - forgot to switch VR back on and was shooting at 300mm and 1/125th. VR back on and sharpness returns (handheld). Leaving the body and lens in MF was easier to spot!
 
It's a good point questioning CA. But I don't really think it's CA... there's not that much contrast/backlighting, I don't typically get much CA with that lens, it's not of a different color (although arguably pushing purple), and the rest of the image is also soft. I've also not seen CA to be distinctly offset with the same border sharpness.
In the tripod test image there's also probably 3-4 frozen moments/overlaps (it's too "busy" to be just a double image).

Like I said, when VR does adversely affect an image it can be very hard to isolate it/attribute it distinctly... But based upon my experience with this lens/combination I'm pretty certain it's VR causing it.

Same lens/TC/body, no VR. Harder backlight (sky) showing a little CA more characteristic of the setup.

View attachment 33353

My position on VR is: if you need it, use it; the results can't be any worse. But if you don't need it, don't use it...
 
I don't think there's any doubt over VR being off when on a tripod. I recently tried some shots of the constellation Orion using a 105mm Micro Nikkor but forgot to turn off the VR... ended up with some very interesting little doughnuts.

I'm wondering whether that Pipit is actually a mix of CA and subject blur?

Off on a slight tangent... I've always wondered what happens if you were able to hand-hold to perfection with VR on.... in other words your hands were rock-steady as if mounted on a tripod? Does VR go 'walk-about' effectively negating steady shooting. Don't get me wrong, I love VR; it's got me out of more situations than I think it put me in, but I've always pondered.
 
interesting discussion ... please keep it going .....I sometimes get disappointing results with VR and am never sure why ........ I do turn it on sometimes on a tripod and compare results ....but in the end I always feel that it is my ability to keep the lens still that is at question .....
 
I've found once you above the 1/focal length (remembering crop factors!) speed then it is best turned off all together. Below that it depends what I am doing, panning then mode 2 anything else mode 1.
 
BillN_33 said:
interesting discussion ... please keep it going .....I sometimes get disappointing results with VR and am never sure why ........ I do turn it on sometimes on a tripod and compare results ....but in the end I always feel that it is my ability to keep the lens still that is at question .....
Click to expand...

Yip, it is one of those things I never thought to question. Now I will operate for a while without VR and see what impression I gain.
 
With out saying which one we are running a discussion on another forum about VR at fast shutter speeds with long lenses,at the moment we can find no evidence against it.As is often the case the discussion has taken a few side roads one of them is VR with birds in flight,IE if Vr is on and continues focus is selected will it slow down the cameras ability to keep refocusing,this looks like it could.
 
Hazza said:
I've found once you above the 1/focal length (remembering crop factors!) speed then it is best turned off all together. Below that it depends what I am doing, panning then mode 2 anything else mode 1.
Click to expand...

i do the same- turn off VR/IS when shooting fast shutter speeds
 
I generally turn off VR when I don't need it (which is most of the time) but Nikon does seem to provide the most rudimentary stabilisation system and I wonder what your results would been with something else.
 
Interesting examples from the charts in post #1 :thinking:
 
Unconvincing. To contend that IS/VR is not beneficial at longer shutter speeds doesn't deserve even a moment's thought. At higher shutter speeds though, the theory is more interesting, and your first picture obviously shows something going on, but it could be anything! See this article from Thom Hogan published some years ago http://www.bythom.com/nikon-vr.htm but he gave no pictorial evidence, either then or later that I'm aware of, and despite my best efforts at the time, I couldn't find anything conclusive in support. Neither could I find any problems in my own tests, with Canon equipment.

It's actually very hard to test for this, as it's almost impossible to distinguish anything that might be IS/VR induced from normal camera-shake or subject movement, and this applies to your pipit image. Basically you have to shoot hundreds of images, with IS/VR on and then off, side by side, then examine them all and see what kind of averages emerge. To draw broader conclusions, you also need to do that on a few different lenses, including newer models with greatly improved stabilisation technology. I did a fair amount of that, but since I couldn't detect anything at all after the first hundred or so - apart from the very obvious fact that image stabilisation is massively beneficial at longer shutter speeds - I gave up.

If there was some kind of general problem relating to IS/VR, apart from things like tripod use and panning etc that we all know about, then it would have been headline news right from the start.
 
And after shooting hundreds of test shots, do it again with teleconverters? I have a feeling that stabilisation is less effective when using them, more so than you'd expect from the increase in focal length.
 
I did not say it is not beneficial at longer SS's... but I think there's something else to it even at longer SS's.
Here's a tight crop from another Pipit image without the obvious effect. Same bird, same everything... (I hadn't wiped the card)

View attachment 33379

The only conclusion here is that IS/VR is not a miracle worker. It is probably worth noting that using the Sharpshooter Camera Mount kneeling might be more akin to using an unlocked gimbal head...it's definitely a lot more stable than trying to freehand a 400/2.8.

As far as the test target goes, I can duplicate that at will. If I put my camera/lens on a tripod with VR turned on I can just watch the image move around in live view. It's essentially recording that VR shift while also trying to compensate for mirror slap. I've seen it as a distinct double exposure (w/ tripod mode VR) and the busy results I got here (normal mode VR). It does require a slower SS in order to be clearly obvious.

The reason for posting the images is due to the lack of anyone else having done so before (that I am aware of). I do ~95% of my photography using this same setup/lens... I am very familiar with how it performs. It wasn't until I recently did AFMA w/ the D810 (and tried tripod mode VR due to crappy lighting) that I ever saw anything I could clearly/obviously state was due to VR ("double exposures"). I turned VR off and the problem immediately went away. And then the other day I got these results with the Pipit. Usually I don't shoot in conditions that require/benefit from VR... normally I quit before then; and if I don't give up, I usually forget to turn VR on. Typically at anywhere near 1/100 and 800mm on the D810 the results look *very* different without VR. ;)

Unfortunately, I can't post an image of minimal VR degradation and state for a fact that that is obviously the problem. The 100% crop second example isn't fantastic either, and I would be equally inclined to say the "bloom" around the bill is CA or "softness" (high ISO and 2x TC). Thom calls the result "busy bokeh, or bokeh that doesn't have that simple shape and regularity." That seems to correlate to the "busy/multiple exposure" examples I'm posting.

I don't know that there's a universal conclusion to be made from this... other than don't use VR "wrong." My 400/2.8 is a slightly older version (replaced w/ the new FL model) and has a "tripod mode" selection. But I can say that even tripod mode can be "negative" (not "will be") when on a tripod... and there must be a reason that even the latest versions still have an "off" position...

My rule is: if I don't *need it* I don't use it... and I try to stay well away from "needing it."
Unfortunately, even the fairly significant result of the Pipit doesn't really show up on the camera screen zoomed in... so chimping doesn't help unless it's stupid bad (these are very hard crops). And even when it is affecting some images, it won't affect all of them equally (IME).
 
Last edited:
eccles said:
And after shooting hundreds of test shots, do it again with teleconverters? I have a feeling that stabilisation is less effective when using them, more so than you'd expect from the increase in focal length.
Click to expand...
Using long FL's is akin to shooting 1:1 macro... it's just a lot more demanding in every aspect. Especially on a high resolution/MP sensor.
 
mikew said:
With out saying which one we are running a discussion on another forum about VR at fast shutter speeds with long lenses,at the moment we can find no evidence against it.As is often the case the discussion has taken a few side roads one of them is VR with birds in flight,IE if Vr is on and continues focus is selected will it slow down the cameras ability to keep refocusing,this looks like it could.
Click to expand...
No idea on that... the two signals (focus/stabilizations) are generated separately. There's no obvious reason for one to be "ignored" even if they input to the same element(s).
 
zorki said:
i do the same- turn off VR/IS when shooting fast shutter speeds
Click to expand...
But I was using slow SS's... honestly though, I don't usually see anything as "obvious" as those results....especially with shorter lenses/lower resolution sensors (i.e. D4/70-200).

Interestingly though, I almost never turn VR off on my Nikon 1 V2 (because it's buried in the menus) and I've never seen anything so apparent... maybe I should...
 
Last edited:
The pipit images just degrade your argument IMHO. You're attempting to show some kind of VR induced movement when shooting hand-held at very long shutter speeds, with a subject known for rapid and jerky movements...!

If the problem with the test target image is typical and can be repeated at will (with mirror lock-up and preferably electronic first curtain, too) my first thought is the lens needs checking over. But if it is some kind of common VR artefact, then you should be able to repeat it with other VR lenses.
 
I don't know if there are any significant differences between Nikon's VR and Canon's IS systems but doesn't it behave differently for different lenses?

For example, I am sure I read somewhere that Canon's new 100-400mm L IS II doesn't need to have IS switched off when on a tripod. With this in mind I was recently shooting a Little Owl from a fixed position on a tripod with gimbal head - I took about 150 photos over a couple of hours and randomly had IS switched on and off and also tried Stabiliser Mode 3. There are absolutely no differences in sharpness between any of the images - They are all sharp.
 
HoppyUK said:
The pipit images just degrade your argument IMHO. You're attempting to show some kind of VR induced movement when shooting hand-held at very long shutter speeds, with a subject known for rapid and jerky movements...!
Click to expand...
Motion blur (subject/camera) doesn't have 2 or 3 distinct outlines.... sometimes two (start/stop) if it's a picture of a light source on dark. It would be a very weird confluence of situations to occur at 1/100 with the subject/situation presented. But I have 6 images with the result and two without.

I could try to duplicate the result with a couple other longer lenses (70-200/120-300)... I don't believe the lens needs serviced, there's no other "symptoms" and VR stabilizes the veiwfinder/image as expected.
 
Last edited:
I have the issue with the VC of my Tamron 24 - 70 2.8, hand held at 'moderate' shutter speeds (1/125 @ 70mm, f2.8), VC on is a softer image with the introduction of a hint of blur vs VC off which is pin sharp. Repeatable results across any number of frames, three different copies of the lens (with serial numbers significantly different), all new direct from Amazon UK.

I found another case on DP review with virtually identical results from 2013 : http://www.dpreview.com/forums/post/52316394
 
RedRobin said:
I don't know if there are any significant differences between Nikon's VR and Canon's IS systems but doesn't it behave differently for different lenses?

For example, I am sure I read somewhere that Canon's new 100-400mm L IS II doesn't need to have IS switched off when on a tripod. With this in mind I was recently shooting a Little Owl from a fixed position on a tripod with gimbal head - I took about 150 photos over a couple of hours and randomly had IS switched on and off and also tried Stabiliser Mode 3. There are absolutely no differences in sharpness between any of the images - They are all sharp.
Click to expand...
Yes, definitely differences between VR versions... w/ early versions of VR I would turn it off for panning as it always messed it up.

I guess the other conclusion one could draw is that it is lens/situation dependent... don't let an experience w/ an early version of VR dictate your use of a newer version.
 
Very interesting. I've never seen anything to suggest image stabilisation on a tripod is bad, apart from the warning in the camera manual not to do it. I have in the past chased down mirror and shutter slap with a 500mm lens on a tripod. When I remember I turn off the image stabilisation when doing test runs. But I often forget. Long lens sharpness is a very tricky area, and it's not surprising that I've often got inconclusive or inconsistent results. Some of which looked like your examples of stabilisation gone wrong. So I checked back on a few. Every one of them happened when I'd forgotten to turn of the stabilisation.

Hm. I shall start paying more attention to this...
 
Sharky said:
I have the issue with the VC of my Tamron 24 - 70 2.8, hand held at 'moderate' shutter speeds (1/125 @ 70mm, f2.8), VC on is a softer image with the introduction of a hint of blur vs VC off which is pin sharp. Repeatable results across any number of frames, three different copies of the lens (with serial numbers significantly different), all new direct from Amazon UK.

I found another case on DP review with virtually identical results from 2013 : http://www.dpreview.com/forums/post/52316394
Click to expand...
In my attempts to duplicate the results there seems to be "a range" of SS's where it works well... and other ranges where it doesn't. I never see blatant issues with higher SS's... just very questionable results. It becomes more definitive (IMO) at slower SS's, but the results could be worse w/o IS/VR at those speeds. There's a crossover point somewhere in-between.
 
if you put the camera in Live View on the tripod you always see movement no matter how you lock the lens down ....... VR or no VR ........ I'm not saying how it affects anything I have no idea ..... just saying there is always movement seen ..............in my experience anyway

Listening to the cricket ...... rain stopped play when AB and du Plessis are on a roll
 
Last edited:
i dont know if canikon vr is always active, but with the sigma one your ment to half press shutter button to get it to settle or spin up

i think the shutter has a effect too, ive got silly sharp shots with my sigma dps at low speed, leaf shutter :D
 
BillN_33 said:
if you put the camera in Live View on the tripod you always see movement no matter how you lock the lens down ....... VR or no VR ........ I'm not saying how it affects anything I have no idea ..... just saying there is always movement seen ..............in my experience anyway
Click to expand...


....That has been my experience too, Bill, on Canon Image Stabilisation (IS) when using LiveView (LV). I also notice alarming camera body movement as the shutter sounds but I guess that must be on the shutter mechanism's return after the actual shot is taken? However, the results are all sharp on that session regardless of my IS settings when on a Gitzo Systematic tripod nearly flat and low on the ground so rock solid.
 
RedRobin said:

....That has been my experience too, Bill, on Canon Image Stabilisation (IS) when using LiveView (LV). I also notice alarming camera body movement as the shutter sounds but I guess that must be on the shutter mechanism's return after the actual shot is taken? However, the results are all sharp on that session regardless of my IS settings when on a Gitzo Systematic tripod nearly flat and low on the ground so rock solid.
Click to expand...

Morning Robin,

Be interesting if someone with a Sony A7 i.e. a mirror-less DSLR, could comment on this

same sensor as Nikon, in fact Nikon use the Sony sensor
 
I'm going to ask about the IS settings when tripod mounted on the Canon 100-400mm L IS II in that lens thread and report back here in due course.

Don't some of the Sony cameras have VR/IS onboard their body rather than the lenses?
 
pingu666 said:
i dont know if canikon vr is always active, but with the sigma one your ment to half press shutter button to get it to settle or spin up

i think the shutter has a effect too, ive got silly sharp shots with my sigma dps at low speed, leaf shutter :D
Click to expand...
Yeah the Tamron is the same, the VC takes a moment settle before firing off a shot as it snaps hold. The conclusion i came to with mine was aggressive mirror slap causing the issue in the VC. Certainly after three different copies of the same lens with two of the serial numbers being 10,000 units apart (and the third in the middle) it reasonable to say it probably wasn't a batch run issue.
 
combat squirrel said:
I dont get the point of this thread VR is for when your not on a tripod - OBVIOUSLY it will mess up on a tripod its designed to compensate with movement in mind? lol
Click to expand...

In addition to the one that you mention, I think that several other issues are being discussed, e.g. do you use VR if you are shooting at high shutters speeds hand held, or with a monopod or similar ...... looking at long heavy lens as well as other lens

There is also a view out there that in certain conditions VR can be used when the (long) lens is tripod mounted
 
Last edited:
I've not read through all the responses to this post but your test and examples are flawed from the outset as you used VR whilst secured to a tripod. This will cause problems all on its own.

Your "hand held" shot (with a rifle stock mount) isn't really hand held for this sort of purpose, it would need to be hand held with literally just a hand. It will vibrate against the hard stock in an almost similar way it will with a tripod. But 800mm at only 1/100? Even the best VR wont cure camera shake from that shutter speed at such a length.

Also looking at the odd bokeh (its horrible!) did you have a UV or "protection" filter attached?

I've never had a problem with IS/OS adversely affecting an image, its only ever helped me out when I've needed it.
 
Last edited:
odd jim said:
I've not read through all the responses to this post but your test and examples are flawed from the outset as you used VR whilst secured to a tripod. This will cause problems all on its own.

I've never had a problem with IS/OS adversely affecting an image, its only ever helped me out when I've needed it.
Click to expand...

The whole point of the tripod mounted shot was to establish exactly what the error looks like (almost like a 'control' shot). This shot has been used then as a comparison for the rest of the discussion. I think everybody is agreed VR/IS on a tripod is not a done thing. We're trying to establish if there are errors and whether there is any point/benefit with faster shutter speeds.

You will also see points raised about hand-holding with VR/IS.... so the whole thing is NOT flawed from the outset. You need to read it all to get the gist of it. :)
 
Last edited:
has anyone said exactly what VR does, both technically and in layman terms - I read somewhere that the latest VR on Sony I think, looks at movements in many directions not just the usual vertical and horizontal
 
chuckles said:
The whole point of the tripod mounted shot was to establish exactly what the error looks like (almost like a 'control' shot). This shot has been used then as a comparison for the rest of the discussion. I think everybody is agreed VR/IS on a tripod is not a done thing. We're trying to establish if there are errors and whether there is any point/benefit with faster shutter speeds.

You will also points raised about hand-holding with VR/IS.... so the whole is NOT flawed from the outset. YOu need to read it all to get to the gist of it. :)
Click to expand...
You're probably right, though I did read all of the OP and I based my comment on that but I misread the purpose of the tripod base shot :)

I think we might just be expecting too much from our equipment if we are picking holes in images shot at 800 mm with a SS of 1/100 but I see the point of the overall discussion.

That said, using IS/VR/OS is a fairly basic thing to get right is it not?
 
You must log in or register to reply here.
Back
Top