IS/VR adversely affecting image IQ
- Thread starter sk66
- Start date
- Messages
- 12,653
- Edit My Images
- No
Nikon have/are putting VR on all their lens ... I would think that the 24 70 and 12 24 will have VR in the next version ....... and it is now on all their long, (I could say only suitable for tripod mounting), lens ......... is anyone going to hand hold a 500mm, 600mm, 800mm much of the time
Last edited:
odd jim
Flimsiest Lambresta
- Messages
- 9,209
- Name
- Jim
- Edit My Images
- Yes
How were these shot? Presumably hand held?
- Messages
- 4,463
- Name
- Barry
- Edit My Images
- Yes
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?
Yep, pretty much all of that! Simply enough to deploy but there are some doubts. I've been using VR for about 12 years and it's saved more opportunities than destroyed them
Sounds a silly thing but I need slow(ish) shutter speeds for prop blur in my aerial shots - so there is a fine balance in all of this. I had a 80-400mm VR Nikkor which performed beautifully in this regard. I had to sell it but have bought another since which, sadly, doesn't perform as well. Hence my interest... having said I think my problem is CA, not technique or VR.
- Messages
- 1,872
- Edit My Images
- No
odd jim
Flimsiest Lambresta
- Messages
- 9,209
- Name
- Jim
- Edit My Images
- Yes
Ok, these images were shot with a Sigma 105mm OS handheld at 1/50th (iso 1600) at f/4, raws but converted straight to jpeg.
1 - with OS
os test a by Odd Jim, on Flickr
os test a2 by Odd Jim, on Flickr
2 - without OS
os test b by Odd Jim, on Flickr
os test b2 by Odd Jim, on Flickr
I did a few test shots and each time, same results.
Not scientific (and yes the crops are slightly different but the image is unaffected) but a real world application. I'd say OS certainly worked well here.
1 - with OS
os test a by Odd Jim, on Flickros test a2 by Odd Jim, on Flickr2 - without OS
os test b by Odd Jim, on Flickros test b2 by Odd Jim, on FlickrI did a few test shots and each time, same results.
Not scientific (and yes the crops are slightly different but the image is unaffected) but a real world application. I'd say OS certainly worked well here.
Last edited:
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
All lens based IS/VR has to be "activated" with a button press... otherwise it would just eat batteries. Sensor based IS just tends to be on (if enabled) and eats batteries.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
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
The movement seen in live view on a tripod is "different" than with IS/VR off... You will (should) see a stable image that occasionally "jumps" to a new stable position.
If you see movement from mirror slap then you tripod/head/technique probably are not up to demanding/critical work. A good test for this is to tape a laser pointer to the "floating end" of the lens/camera and point it at a wall a good distance away (at least "subject distance," longer distances makes the test more demanding). Then trigger the camera in timer mode so you can watch the laser dot closely... if it moves around during the exposure you have an issue. Using a longer SS will separate the movements between mirror/shutter up and return. You can try M-up, M-up with delay, weighting the tripod, etc, until you find a technique that works with that combination.
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
Similar to what I find at higher SS's... nothing one would say is obviously VR induced... could easily be attributed to missed focus.
- Messages
- 1,872
- Edit My Images
- No
In my case not missed focus, i can hit 10 shots vc off perfect and 10 shots vc on with the induced blur. I also performed controlled tests (all still hand held) with a focus chart and the focus didn't shift but the quality of the in focus elements did.
- Messages
- 12,653
- Edit My Images
- No
Thanks Steven
is this a hands off exercise on a very good tripod locked down or normal shooting?
I still see movement on the LV screen if I use delayed release ........... holding the lens as steady as possible, with full weight on the middle part ..................or hands off or with any other technique that I have tried ...... there always seems to be some movement there with every lens that I have and every camera body
Wimberly head and Gitzo tripod
you mention issue what would that be ......... I need to still improve my technique or something else?
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
No-one is saying VR doesn't/can't work and be of a benefit... There's just a question of when it may be potentially detrimental. This will certainly be VR/IS version dependent...it will also be FL/body/technique dependent I think.
And what is the "four stops gain" based on? The ROT SS? (which I think is too slow for critical work, particularly with todays high MP sensors). Which Body? Critical sharpness or the ROT's "acceptable sharpness?"
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
My point was not that it *was* missed focus, but that VR could be affecting some images and you would be likely to attribute it to missed focus instead.
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
This is a "hands off" test (timer or remote release)
Your problem is likely the Wimberly gimbal head. It has too many pivot points and too much leverage to be rock steady... just putting the camera out there on the platform is similar to raising the center column. It's not really meant to be "rock steady," that's not what a gimbal head is for...It could also be leg/column extension etc...
On my biggest Gitzo w/ no column the Benro GH-2 gimbal can't pass the test no matter how hard I lock it down... I can get it to pass the test if I use M-up and a slight delay before releasing the shutter. If I switch to a different head (UniqBall-45 in my case) the tripod will pass the test w/o mirror up. Granted, this can be a very demanding test (i.e. from 20ft) and you may not actually *need* that much stability.
Last edited:
- Messages
- 1,872
- Edit My Images
- No
Ah, it's the way i read it sorry
- Messages
- 12,653
- Edit My Images
- No
I have a heavy Arca Swiss head ...... I try it with that, although with a 600mm it gets a bit balance critical
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
At 600mm everything becomes a bit more critical...
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
I did tests with a couple other lenses. My 70-200/2.8 does not have a "tripod" switch position which means it is supposed to recognize being on a tripod. And it seems to work well; I wasn't readily able to get VR issues (I tried a few settings).
The Sigma 120-300/2.8 was another story... VR mode 1 consistently looked very similar to VR off, close enough that I am willing to attribute any differences to technique/setup. But VR mode 2 was a different beast. These are all tripod mounted at 1/4 sec.
View attachment 33431
I then went and tried to demonstrate handheld at higher SS's...because there are so many variables it's hard to say anything definitive. At 140mm and 1/320 only the active VR mode (more) consistently produced different results. But at 140mm and 1/2000 both normal and active VR modes consistently produced slightly softer results... Where the "results" were not more consistently worse I grouped them as a single example image. Again, all are very hard D810 crops.
View attachment 33432
Interestingly, this set of images were taken at f/2.8; and driving the ISO up to 3200 from 1000 resulted in better/sharper images (1/2000 SS) than using VR at the lower SS did...
The Sigma 120-300/2.8 was another story... VR mode 1 consistently looked very similar to VR off, close enough that I am willing to attribute any differences to technique/setup. But VR mode 2 was a different beast. These are all tripod mounted at 1/4 sec.
View attachment 33431
I then went and tried to demonstrate handheld at higher SS's...because there are so many variables it's hard to say anything definitive. At 140mm and 1/320 only the active VR mode (more) consistently produced different results. But at 140mm and 1/2000 both normal and active VR modes consistently produced slightly softer results... Where the "results" were not more consistently worse I grouped them as a single example image. Again, all are very hard D810 crops.
View attachment 33432
Interestingly, this set of images were taken at f/2.8; and driving the ISO up to 3200 from 1000 resulted in better/sharper images (1/2000 SS) than using VR at the lower SS did...
Last edited:
- Messages
- 1,369
- Name
- Chris
- Edit My Images
- Yes
Looks very like mirror or shutter vibration to me. Of course that doesn't mean that it is, just that it's a possibility that needs to be eliminated.
- Messages
- 1,165
- Edit My Images
- No
sk66 mate, your lenses or camera is broken or something, your getting odd results - hand held with IS/OS/VR on = clear shots for the rest of us.
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
Nah. With shorter FL's and lower SS's (the kind of stuff you would normally handhold and use VR) it works pretty well, certainly not any worse than without VR. ("active mode" is meant for when something is moving you)
But combined w/ a tripod (unless tripod aware) or higher SS's, not so much. And I don't think it's as effective at long FL's and high resolution as it's advertised to be.
odd jim
Flimsiest Lambresta
- Messages
- 9,209
- Name
- Jim
- Edit My Images
- Yes
IS / VR is most effective at long FLs. At short FLs you can get away with longer shutter speeds anyway (UWAs anyone?)
And why would you use IS/VR when on a tripod? Most people know not to do that for obvious reasons, and secondly, it's on a tripod, so why would you need it anyway?
Last edited:
- Messages
- 1,165
- Edit My Images
- No
YEAH OBVIOUSLY ? Don't use VR on a tripod!! This is known by everyone!? lol
Yes in general, but I recently got the canon 16-35 IS - and MY GOD ITS AMAZING, even here, I can't live without IS now, its so so useful for many creative techniques
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
It would be better if you start at the beginning and read the thread...
The purpose of using it while on a tripod was to show what VR issues can look like while eliminating the other (confusing) potential causes for the image degradation. I know I've had VR *cause* IQ issues when used "correctly," and also when used at higher SS's. Other's in this thread have noted similar results.
If you don't know what VR caused issues might look like you would probably just blame it on soft focus/motion blur... (at higher SS's it *is* motion blur; of the moving lens elements)
The purpose of using it while on a tripod was to show what VR issues can look like while eliminating the other (confusing) potential causes for the image degradation. I know I've had VR *cause* IQ issues when used "correctly," and also when used at higher SS's. Other's in this thread have noted similar results.
If you don't know what VR caused issues might look like you would probably just blame it on soft focus/motion blur... (at higher SS's it *is* motion blur; of the moving lens elements)
- Messages
- 23,200
- Name
- Richard
- Edit My Images
- No
You can't draw any conclusions about IS/VR when on a tripod. The reason it's advised to turn IS/VR off on a tripod is because the lack of movement creates a feedback loop in the system that can then produce movement of its own.
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
I wasn't drawing any "conclusions" about anything... except my own personal conclusion which is to turn VR of if it's not *needed.*
VR and tripod use is also "variable." It is recommended (Nikon) that VR be used if the tripod head is unlocked (i.e. a gimbal/panning). And my 400/2.8 has a special tripod mode that is meant to be used when mounted on a locked tripod head (supposedly optimized for mirror slap). They got rid of that mode on the new FL version and replaced it with VR "sports mode".... whatever that means.
I've seen VR affected images using VR on a gimbal, as well as handheld/monopod... typically with higher SS's but not always. It does seem that somewhere around the 1/500 suggested by Thom is the limit; but that limit doesn't mean VR *will* cause an issue consistently. Usually I suspect VR when an image, or images, in a series looks worse than the others when nothing has changed... such as the pipit examples, those two images are in sequence.
It's also VR version/brand dependent. i.e. panning with Nikon VR 1st version was not so good, but much better with the current version. And Sigma VR is somewhat different from either of them, and also has multiple versions.
And I question the "up to four stops".... Do they detune the capability/technology for shorter lenses? Is it when combined with a monopod/tripod and the recommended mode?
Nikon does say that the rated "improvement" for their FF lenses is only applicable when used with a FF sensor.... which suggests it's an "acceptably sharp" rating and not necessarily a "critically sharp" rating/result (i.e. not up to hard cropping/D810 pixel peeping).
VR and tripod use is also "variable." It is recommended (Nikon) that VR be used if the tripod head is unlocked (i.e. a gimbal/panning). And my 400/2.8 has a special tripod mode that is meant to be used when mounted on a locked tripod head (supposedly optimized for mirror slap). They got rid of that mode on the new FL version and replaced it with VR "sports mode".... whatever that means.
I've seen VR affected images using VR on a gimbal, as well as handheld/monopod... typically with higher SS's but not always. It does seem that somewhere around the 1/500 suggested by Thom is the limit; but that limit doesn't mean VR *will* cause an issue consistently. Usually I suspect VR when an image, or images, in a series looks worse than the others when nothing has changed... such as the pipit examples, those two images are in sequence.
It's also VR version/brand dependent. i.e. panning with Nikon VR 1st version was not so good, but much better with the current version. And Sigma VR is somewhat different from either of them, and also has multiple versions.
And I question the "up to four stops".... Do they detune the capability/technology for shorter lenses? Is it when combined with a monopod/tripod and the recommended mode?
Nikon does say that the rated "improvement" for their FF lenses is only applicable when used with a FF sensor.... which suggests it's an "acceptably sharp" rating and not necessarily a "critically sharp" rating/result (i.e. not up to hard cropping/D810 pixel peeping).
Last edited:
- Messages
- 23,200
- Name
- Richard
- Edit My Images
- No
Lens manufacturers these days usually use the CIPA standards for things like this, here http://www.cipa.jp/image-stabilization/index_e.html The test method I use for magazine reviews usually ties in with that pretty closely.
odd jim
Flimsiest Lambresta
- Messages
- 9,209
- Name
- Jim
- Edit My Images
- Yes
TBH I thought the iq issues caused in your bird shot looked like they were caused by the use of a UV or "protective" filter on a long lens. I asked if one was used and it wasn't answered (unless I've missed it). It's still a point I'm curious about. Particularly because of the nasty bokeh which I'm sure isn't typical of that lens.
This part of your post is why I and others have questioned the tripod issue again as you stated its not of benefit on a tripod or with high SS, both occasions where you wouldn't use it anyway. So a rather moot point and this comes across differently to simply using your tripod shot as a control image.
Last edited:
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
I must have missed your question... Only the required "plain" drop in filter was in place (400/2.8 VRII).
That kind of depends on what you consider a "high SS." If you use the ROT 1/FL you might have issues if you frequently work at long FL's like I do (400mm+). And if you think of it as "tripod" instead of "stability" you might run into issues as well (i.e. used with a pad). Or if you use it with a gimbal head as recommended... And many never turn VR off because they have never seen/recognized an issue, or if they do it's only when mounted on a tripod and only because they're "not supposed to use VR."
And if you use longer FL's with high resolution sensors (most of them these days) my experience says you should use much faster SS's (i.e. 1/2xFL min) which puts it outside of VR use.
I thought the first post where I stated I misused VR on a tripod in order to exasperate the issue made it clear...
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
I was aware of the standard, but I had never read it...
I found this interesting:
For a D8xx with 4.88um pixels that would equate to an ~13 pixel offset; certainly acceptable for the postcard size output print it was based on...
It is also interesting to note that the rating is based upon the average of ~200 images and that outliers are discounted from the average... that means "up to x-stops" means you can expect occasional results that are significantly worse, and occasional results that are better than advertised.
Just more evidence that you should try to avoid needing it as opposed to readily relying on it.
Last edited:
- Messages
- 23,200
- Name
- Richard
- Edit My Images
- No
Not sure what you mean by this. Is it the 'Thom Hogan effect'?
The main point about the CIPA standard is that it's consistent, between lenses and brands. If it says a lens gives three stops of improvement vs another lens with four stops, you can be fairly sure the four stops lens will be noticeably better.
My own test regime also has a fairly 'loose' standard (it's what I call 'acceptable' sharpness rather than optimum) but again, the point is it's consistent. The CIPA standard is certainly generous; I am usually 0.5 to 1.0 stops less optimistic at the end of the day, but not a million miles different. And my tests are not related to the 1/focal length rule of thumb - the reference point I use varies a little (I can usually beat the RoT by about one stop anyway) but I set that by trial and error with each new test and the starting point is the shutter speed where I can get consistent nine out of ten acceptably sharp images hand-held.
Sorry that I'm unmoved so far by your results, but I will say it's an intriguing area and I'd be interested in some more conclusive testing. But that must not involve a tripod or support of any kind, and only static subjects! And it must be repeatable, not just one iffy looking image that could have any number of causes.
Last edited:
odd jim
Flimsiest Lambresta
- Messages
- 9,209
- Name
- Jim
- Edit My Images
- Yes
It did. But then the post I quoted kind of contradicted that, as why state the obvious as it was a control shot?
Last edited:
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
Definitely very generous... Their "acceptable COC" of 63um is based upon viewing a "postcard" from an average distance of 72cm/~2.5ft. I'm guessing the "most popular postcard size" is A6/4x6 (they show it as being smaller than A5). Compared to the (arguably low) "COC standard" of viewing an A4/8x10 from 25cm/1ft, it's not even in the same ballpark...
I'm surprised your results do not differ more greatly.
If you mean the 1/.5x sampling rate (~1/500) Thom notes, then yes... but I can't state with any certainty that 1/500 is applicable across the board. IS/VR systems vary/improve. And even if it *can't* "technically" help, that doesn't mean it *will* hurt.
There are so many variables, and it can be hard to discern the cause. And this CIPA standard all just reinforces my opinion which is:
Do not use VR "incorrectly." Do not enable VR if you do not *need* VR. Try to avoid needing VR if possible (use higher min SS's with more demanding sensors). And don't count on VR fixing your problems when you do need it (i.e. maybe take more images).
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
In response to an earlier question as to my stupidity...
- Messages
- 94
- Edit My Images
- Yes
So Just a few thoughts on why IS might fail.--
IS has to wind up after the focus is established- it is not always on to save battery life. If you press the shutter before IS has settled the sensor might register movement.
In a high pressure situation ie photographing birds, you take the photo as soon as you see it, no waiting for the kit to catch up.
Also working at the edge of the performance envelope ie 4 stop lower than 1/FL, is going to be a challenge for all but the steadiest hands. I can barely hold at 1/FL.
IS on a tripod might fail because the laser gyros (if that is what is used) are more sensitive than the stepper motor's (if that is also what moves the IS element) minimum movement.
and a final thought from the Science Lab- be careful when experimenting that you don't fall into Confirmation Bias. This can be difficult. You need to be disciplined do all the tests needed to be thorough.
IS has to wind up after the focus is established- it is not always on to save battery life. If you press the shutter before IS has settled the sensor might register movement.
In a high pressure situation ie photographing birds, you take the photo as soon as you see it, no waiting for the kit to catch up.
Also working at the edge of the performance envelope ie 4 stop lower than 1/FL, is going to be a challenge for all but the steadiest hands. I can barely hold at 1/FL.
IS on a tripod might fail because the laser gyros (if that is what is used) are more sensitive than the stepper motor's (if that is also what moves the IS element) minimum movement.
and a final thought from the Science Lab- be careful when experimenting that you don't fall into Confirmation Bias. This can be difficult. You need to be disciplined do all the tests needed to be thorough.
- Messages
- 23,200
- Name
- Richard
- Edit My Images
- No
I know where you're coming from and to some extent I agree. But I use a fairly generous standard of 'acceptable' sharpness partly so that my results don't vary too far from the CIPA standard - it would only serve to confuse. The other thing is, it doesn't matter what standard of sharpness is used (within reason). If my tests, to my standard, come out with a three stops improvement over normal hand-holding, then your tests to your standard should show a similar level of improvement.
So when do you not need IS/VR? At what point does this extraordinarily effective technology cease to be a benefit, and actually become a liability? Using IS/VR on a needs basis assumes that there is some kind of rule we can follow, such as Thom's technically-based theory. But that is a long way from proven, and in fact the opposite might be the case and maybe the newest-tech IS/VR systems are still beneficial at higher shutter speeds above the 1/focal length ROT. Not showing big improvements because the scope is reduced at higher shutter speeds, but improvements none the less. How do you know? Camera-shake doesn't just go away, it's always present, just reduced to acceptable levels. Even if IS/VR systems get caught out now and then and misbehave (worst case scenario IMHO, on current evidence) you'd be throwing away the baby with the bath water.
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
I understand the "a standard is better than no standard" approach... and I agree that it is. At least it's "relative."
*For me* (understanding I primarily shoot wildlife/birds with very long FL's w/ D810) my minimum desired SS is 1/2xFL. Or 1/FL if using monopod/rest/aid. Using those guidelines (at any FL really) I have never seen an improvement from IS/VR, only the occasional result I believe VR degraded. I'll push these rules if using a lower resolution sensor, but not using VR to do so. I can't say I've ever seen an improvement from VR above 1/500, but that kind of goes with the previous rule. I have seen VR degrade an image below 1/500 when combined with a stability aid (not a locked tripod).
The area *I think* VR is beneficial and more likely to help than hurt is sub 300mm, sub 1/2xFL, and sub ~1/500 SS's (and if you are doing hard cropping/using a high resolution APS sensor expect less). It's not until I'm in those realms that I might have an issue with sharpness/IQ to begin with. If you want "rules" you'll have to make your own based on testing your own gear... I can't do better than that. My only point is not to assume that using VR is not a potential issue if you are getting less than optimal results.
(The reason I suggest ~300mm is because longer FL lenses are typically used with some type of aid)
- Messages
- 23,200
- Name
- Richard
- Edit My Images
- No
You can't be serious. Try hand-holding an unwieldy monster like the huge and front-heavy Sigma 150-600 S on a blustery day, as I did. Very real, significant and consistently better sharpness with OS on than off at speeds a lot higher than that. Up to 1/2000sec from memory, on Canon 7D.
OP
sk66
Advertiser
- Messages
- 8,674
- Name
- Steven
- Edit My Images
- Yes
Yes, I am serious. What were you shooting?
I *handhold* my 400/2.8 +2x frequently... and my Sigma 120-300/2.8 +2x occasionally... both are more unwieldy than the 150-600's. But I'm not adverse to doing a test to verify.... I need to do some focus testing anyways. (I'm a bit torn between TC use VS cropping)
- Messages
- 23,200
- Name
- Richard
- Edit My Images
- No
I was servo-AF testing at the time with a car driving up and down and just did a few image stabilisation checks inbetween runs. Unfortunately the whole session was ruined by the wind, just not good enough for testing, and I wiped the card. The lens was on loan so I no longer have it. Bear in mind that on a 7D the effective focal length is 960mm so a 1/000sec shutter speed is only borderline according the hand-holding ROT anyway.