IS or not IS.

Just stumbled upon interesting test.

Proof that Image Stabilization Should First be Stabilized

Actually IS reducing sharpness, even when used correctly.

Off course it is not applicable to new Sony E mount cameras, Sony simply doesn't have 300/2.8 E mount lenses:)

The problem that I see in his testing is that you are supposed to turn the VR OFF when on a tripod.

I do not know about this "tripod detection" mechanism that he talks about.

Gary,

 

I think he is saying the lens can detect when it is mounted on a tripod.

I've had two Pentax K100D DSLR's, the first I bought new, the second one used and both have on board IS that works. The first one would make a soft rustling sound in long exposures under 1/30's letting me know it is on and working. Turned off, no sound.

 

The used one I now have makes this same soft rustling sound when IS is turned on, but when I switch it off, the sound sometimes sputters intermittently off and on. Of course it shows the ghosting edge effect as if using it while the camera was on a tripod. Sometimes I can switch it off and on several times and it stays off going by that there's no sound.

 

These small circuitry electronic devices act up in the most unexpected and not so obvious ways even within the same camera model.

Off course it is not applicable to new Sony E mount cameras, Sony simply doesn't have 300/2.8 E mount lenses:)

Sony doesn't, yet, but Nikon does. I don't have the f/2.8 version, but I got this result with the AIS 300/4, hand-held, at 1/5 second. The photo is a 100% crop, taken with a Sony A9 using IBIS only.

 

[ATTACH=full]1223673[/ATTACH]

I don't have any long lenses with VR, nor did I try any formal testing, but in my experience, the premise that VR works better if given a bit of time "to settle down" rings true. This is, of course, hand-held. Never used VR with a camera mounted on my tripod.

 

Possibly the "tripod detection" is the Active Mode, where VR corrects only for vertical movements (to enable panning and shooting from moving objects, etc.)?

Sony doesn't, yet, but Nikon does. I don't have the f/2.8 version, but I got this result with the AIS 300/4, hand-held, at 1/5 second. The photo is a 100% crop, taken with a Sony A9 using IBIS only.

 

[ATTACH=full]1223673[/ATTACH]

Quote from article:

"Here is a quick summary of the above findings:

1. Do not turn Vibration Reduction (Image Stabilization) on unless working at shutter speeds lower than inverse of the focal length of the lens – it does hurt the sharpness a little. For example, if you are shooting at 300mm, VR should be turned on if you are shooting under 1/300 of a second (general guideline, might need faster shutter speed for APS-C and high-resolution full-frame cameras)."
   

Can you guys actually read provided material before jumping to conclusion, 1/5 of sec is not above 1/300 :)

1. Do not turn Vibration Reduction (Image Stabilization) on unless working at shutter speeds lower than inverse of the focal length of the lens – it does hurt the sharpness a little. For example, if you are shooting at 300mm, VR should be turned on if you are shooting under 1/300 of a second (general guideline, might need faster shutter speed for APS-C and high-resolution full-frame cameras)."
   

 

Reading does not confer acceptance.

 

I have noticed a slight reduction in sharpness, to the extent of a pixel or two uncertainty with my A7Rii when used on a tripod. However if the image I provided are taken without IS, the street sign sign would be unreadable, iif not unrecognizable. I haven't tested the A9 for critical sharpness, but it has less resolution and better IS than the A7Rii In my experience, it is always better to use IS, regardless of shutter speed, except on a tripod. Even then, the artifacts are negligible.

 

Hand-held camera shake is approximately 2 deg/sec, which creates an uncertainty equivalent to the limits of DOF (0.02 mm in an 8x10 print, viewed from 10") when the shutter speed is 1/f. Expressed more graphically, that means the effective resolution of a FF sensor is reduced to about 6 MP.

Edited December 13, 2017 by Ed_Ingold

Ed, do you think image quality will benefit from VR , with shutter speed at 1/4000?

Tripod detection VR is present in a handful of Nikon lenses; most lenses should be set to VR OFF when using a tripod, and in my experience this also applies to lenses which have automatic tripod detection mode (I don’t trust it). An explicitly activated TRIPOD mode is present in some superteles such as the VR 500/4 G; from little experience have with that, it seemed to work well. Here is the list of lenses with tripod mode switch as well as thise with auto tripod detection.

 

 

VR mode options | Nikon’s exclusive VR technologies | Empower Your Hand-held Shooting | Nikon

 

 

Active mode is for situations when there are frequent vigorous movements such as might be experienced when shooting from a moving platform such as a helicopter or boat. My experience is that active mode reduces sharpness and is to be avoided. Perhaps my expectations were not low enough.

 

 

SPORT mode is for shooting a moving subject which may maks sudden changes in velocity.

 

 

NORMAL mode is the counterpart to ACTIVE/SPORT/TRIPOD in lenses that have those modes. In those lenses with active mode, normal mode is used when on a stationary platform. It compensates for panning automatically. In those lenses with sport mode, normal mode is used for static or steadily moving subjects. In lenses that have tripod mode, normal mode is for hand held shooting.

 

 

I never use active mode; in lenses that support sport mode, I really like to use it. I turn VR off for tripod shots except for lenses that have the explicit tripod mode switch.

 

Of course, other brands have their own stabilization settings.

Ed, do you think image quality will benefit from VR , with shutter speed at 1/4000?

No, but it's a nuisance to turn it on and off if the setting is in the menu. If it's on the lens, it's too easy to forget to turn it on again. If it does no harm, let it be. Besides to reach 1/4000 you would have to use a high ISO, which itself degrades the image quality.

No, but it's a nuisance to turn it on and off if the setting is in the menu. If it's on the lens, it's too easy to forget to turn it on again. If it does no harm, let it be. Besides to reach 1/4000 you would have to use a high ISO, which itself degrades the image quality.

 

I prefer the switch on the lens, were it is clearly visible.

If it is in the menu, then I don't know if it is on or off without going into the menu. Definitely a PiA if you shoot both on and off a tripod.

With Sony, a switch on the lens and a menu IS selection are mutually exclusive. With Sony A7 and A9 cameras, you can program a button to open the IS on/off menu item. You can also assign it to the quick menu (Fn button). AFIK, Nikon and Canon lenses use a switch on the lens. I suspect most mirrorless cameras have programmable buttons.

Don't see the point of putting an IS on/off switch within a camera's menu.

 

My 2006 Pentax K100D DSLR has the IS switch on the back of the camera body. But as I've previously indicated I can't tell when it's really on or off because electronic imaging devices aren't as dependable as I first thought.

Edited December 20, 2017 by Tim_Lookingbill

Ed, do you think image quality will benefit from VR , with shutter speed at 1/4000?

 

When shooting handheld, IS helps with the framing and keeping the autofocus points on the subject, whether perched/standing or flying/running. Shooting a puffin flying 50-mph and fast-flapping wings, you want both 1/4000-sec. and IS.

 

There seems to be a lot of theories being thrown around here, but those of that actually shoot long lenses leave the IS on, whether Canon or Sony. (Nikon seems to be some special case). The few times I've mounted my 1,000mm rig on my tripod, I could see the image settle down as the IS spun up.

 

With Sony, a switch on the lens and a menu IS selection are mutually exclusive. With Sony A7 and A9 cameras, you can program a button to open the IS on/off menu item. You can also assign it to the quick menu (Fn button). AFIK, Nikon and Canon lenses use a switch on the lens. I suspect most mirrorless cameras have programmable buttons.

 

Ed is correct, some of the new G Master lenses actually have optical stabilization in-lens, which can work in conjunction with the in-camera IS. There's a switch on my 100-400mm G Master to turn in-lens stabilization on or off. In the menu, you can turn in-camera stabilization on and off. The a9 and a7R III allow all kinds of custom functioning.

Just stumbled upon interesting test.

Proof that Image Stabilization Should First be Stabilized

Actually IS reducing sharpness, even when used correctly.

Off course it is not applicable to new Sony E mount cameras, Sony simply doesn't have 300/2.8 E mount lenses:)

 

Too bad he shoots at such a low shutter speed. I'm not sure what subjects that you'll be shooting at 1/250-sec. with a 300mm lens. Human athletes need 1/1000-sec., slow birds need 1/1250-sec to 1/2500-sec. and fast birds need 1/2500-sec to 1/4000-sec. I wouldn't say that he's using the lens "correctly" for most circumstances.

If the lens has an IS switch and you turn it off, it turns IS (IBIS) in the Sony body off as well. IS function in the menu is disabled if the lens has a switch. Batis 85 and 135 lenses have IS (OSS) but no switch. you can only use the menu.

 

The shutter speed of a lens must be 1/3F or faster before camera shake becomes insignificant.

Ed, F= focal length?

Yes, F = Focal Length.

 

Since resolution is inversely proportional to the shutter speed, and MP to the square of resolution, 3F would represent 2^3 = 8 fold improvement, equivalent to 48 MP (or the real sensor size, whichever is less). I'm referring to camera shake, which affects every part of the image. Motion of the subject (e.g., birds in flight) is a separate issue.

Yes, F = Focal Length.

 

Since resolution is inversely proportional to the shutter speed, and MP to the square of resolution, 3F would represent 2^3 = 8 fold improvement, equivalent to 48 MP (or the real sensor size, whichever is less). I'm referring to camera shake, which affects every part of the image. Motion of the subject (e.g., birds in flight) is a separate issue.

 

Understood.

 

The OP referred us to an article that was comparing IQ of a 300mm lens at 1/250-sec., on a tripod, with no IS, with IS spun up and with IS not spun up. While that's interesting, but I think it has very little relevance to how you actually use a 300mm lens in the real world. Shooting a chart at 1/250-sec. has little relevance, for that particular lens. Also, based on my own experience, the longer the lens, the more the need for IS in almost all circumstances and the more important it is to give it time to spin up to full effectiveness.

The term, "spin up," may refer to a time early in the history of IS when Canon and Nikon used gyroscopes as accelerometers. Not only did it take time to bring them to speed, but they tend to precess, rendering the image somewhat unstable in the absence of motion. There were also "inertial" sensors which vibrated like tuning forks and were sensitive to motion outside the plane of vibration. Modern accelerometers are based on inertia of a fixed mass, measured by force transducers. These sensors are inherently stable and free from precession effects.

How I wish for the option to automatically turn off VR when the shutter speed is faster than 1/focal length (or any other option I might choose like 1/2*focal length, 1/3*focal length, or a fixed shutter speed) and then engage again when the shutter speed becomes longer than that set point.

 

I seem to recall having seen shots that had a weird background rendering allegedly caused by the VR working while using a "too fast" shutter speed (might even have been here on PN) but from what I recall it was not necessarily visible in every shot taken with the same set of parameters. More like a "bad luck" kind of thing.

 

Nikon's various VR modes are poorly documented and one gains experience what works and what doesn't by trial and error. With some lenses, especially when panning with a relatively slow shutter speed of 1/125-1/200 when photographing propeller planes flying by, VR operation can cause a lot of images to be just slightly soft/blurry (of course, operator's lack of competence in panning just aggravates the issue).

Edited December 20, 2017 by Dieter Schaefer

The term, "spin up," may refer to a time early in the history of IS when Canon and Nikon used gyroscopes as accelerometers. Not only did it take time to bring them to speed, but they tend to precess, rendering the image somewhat unstable in the absence of motion. There were also "inertial" sensors which vibrated like tuning forks and were sensitive to motion outside the plane of vibration. Modern accelerometers are based on inertia of a fixed mass, measured by force transducers. These sensors are inherently stable and free from precession effects.

 

Ed, "spin up" is not a technical term, but it does accurately describe what I sense as my EF 500mm f/4L IS II USM gets ready for action. It's by no means instantaneous. If you mount the rig on a big, steady tripod (Arca-Swiss/Wimberley head and Induro's heaviest), then turn on the IS, you'll see momentary movement of the image and then stabilization. I'd guess the time from actuation to steadiness is 1/4 to 1/2-sec.

Even if you use an accelerometer, there is noise in the data and the corrective action is affected by this noise. If the desire is for a more accurate reading of acceleration, this probably involves a longer reading time and so the corrective action is also delayed relative to the original movement. This kind of thing will never perfectly compensate all the shake, it involves a noisy measurement and some kind of prediction or estimation of what the optimal corrective action is. So on a tripod if you have a perfectly solid tripod it makes sense to turn VR/IS off. If the focal length is very long (say, 500mm or 800mm) then likely there will be enough vibration so that the correction is more beneficial than harmful. So it makes sense to use VR with such long focal lengths even on tripod. At least for exposure times that are not very long.

Even if you use an accelerometer, there is noise in the data and the corrective action is affected by this noise. If the desire is for a more accurate reading of acceleration, this probably involves a longer reading time and so the corrective action is also delayed relative to the original movement. This kind of thing will never perfectly compensate all the shake, it involves a noisy measurement and some kind of prediction or estimation of what the optimal corrective action is. So on a tripod if you have a perfectly solid tripod it makes sense to turn VR/IS off. If the focal length is very long (say, 500mm or 800mm) then likely there will be enough vibration so that the correction is more beneficial than harmful. So it makes sense to use VR with such long focal lengths even on tripod. At least for exposure times that are not very long.

 

I find this to be true with my Canon 500/f4 II. It's easy to see in the viewfinder.