em_dee_aitch
Senior Member
The question of whether you should turn off VR at higher shutter speeds frequently comes up in this forum and others, due to the fact that VR has a useful limit (probably around 1/500sec shutter speed) and may actually degrade the image at higher shutter speeds.
Personally, I have found it rare to easily see an actual usage case in which this degradation is painfully obvious, but last week I finally did. Note that this is not a VR bash, and I love VR (on long lenses anyway, not on wide angles, as that's a different topic entirely), and I use VR all the time. However, it is important to know when not to use it.
At a wedding last week, there was a great skyline view, so I decided I was going to pan a quick skyline in between shooting reception guests. When I reviewed my shots, what I saw was that the image was strongly degraded on the left side of the frame (when shooting vertically). The focus was set to manual for the panorama, so AF variation was definitely not a factor.
Here are two shots from that panning of the skyline. You can see the same subject matter in different parts of the frame as I pan, with degradation clear:
After I saw that I knew I needed to isolate it to one of the following factors.
1. The 1.4x TC I was using
2. Centering error in the lens
3. VR
4. Camera shake
AF was already eliminated, as previously noted.
To test/eliminate those factors, I went back out the next day and shot one of the same sky scrapers from a distance again. To eliminate the TC, I shot with and without it. To eliminate lens centering, I shot with the camera rotated in 180 degree opposite positions. To eliminate camera shake, I simply searched for any consistent evidence of handshake effecting the VR off shots, and there was none (though in the interest of general knowledge, there was at least 1 VR off frame in which camera shake was apparent, event at very high shutter speed).
In the end, the factor that clearly reproduced the degradation behavior was turning VR on. Turning it of resolved the behavior.
Here are examples showing VR degradation in different parts of the frame, with and without TC 1.4x:
Keep in mind when viewing these that the factor which explains why VR would only effect part of a frame is that at these shutter speeds, the exposure is effectively taken by a scanning slit between the shutter curtains, because only a portion of the frame can be exposed at once at these shutter speeds, as they are stops beyond x-sync. Thus you can conclude that the state of the VR mechanism is not consistent throughout the entire exposure, and that at its very best a VR exposure, within its effective shutter speed range, is probably not as good as the best non-VR exposure, while of course VR can allow you to get certain shots that can never be had in a non-VR method.
The lens used for all shots above was 70-200 VR2. The first two shots show unconverted NEF previews. Due to the limitations of preview files embedded in NEFs, the effect would be even stronger if the shots were converted due to the extra detail revealed in the superior shot (however, since the effect was still quite clear, I did not bother on those two). For the latter three shots, you do see post-converted JPEGs, which were converted with no sharpening whatsoever, as to keep from obscuring the actual effect.
--
David Hill
http://www.sfbayweddingphotographer.com
San Francisco, CA | Austin, TX
Certified Wedding Photography Junky™
Personally, I have found it rare to easily see an actual usage case in which this degradation is painfully obvious, but last week I finally did. Note that this is not a VR bash, and I love VR (on long lenses anyway, not on wide angles, as that's a different topic entirely), and I use VR all the time. However, it is important to know when not to use it.
At a wedding last week, there was a great skyline view, so I decided I was going to pan a quick skyline in between shooting reception guests. When I reviewed my shots, what I saw was that the image was strongly degraded on the left side of the frame (when shooting vertically). The focus was set to manual for the panorama, so AF variation was definitely not a factor.
Here are two shots from that panning of the skyline. You can see the same subject matter in different parts of the frame as I pan, with degradation clear:
After I saw that I knew I needed to isolate it to one of the following factors.
1. The 1.4x TC I was using
2. Centering error in the lens
3. VR
4. Camera shake
AF was already eliminated, as previously noted.
To test/eliminate those factors, I went back out the next day and shot one of the same sky scrapers from a distance again. To eliminate the TC, I shot with and without it. To eliminate lens centering, I shot with the camera rotated in 180 degree opposite positions. To eliminate camera shake, I simply searched for any consistent evidence of handshake effecting the VR off shots, and there was none (though in the interest of general knowledge, there was at least 1 VR off frame in which camera shake was apparent, event at very high shutter speed).
In the end, the factor that clearly reproduced the degradation behavior was turning VR on. Turning it of resolved the behavior.
Here are examples showing VR degradation in different parts of the frame, with and without TC 1.4x:
Keep in mind when viewing these that the factor which explains why VR would only effect part of a frame is that at these shutter speeds, the exposure is effectively taken by a scanning slit between the shutter curtains, because only a portion of the frame can be exposed at once at these shutter speeds, as they are stops beyond x-sync. Thus you can conclude that the state of the VR mechanism is not consistent throughout the entire exposure, and that at its very best a VR exposure, within its effective shutter speed range, is probably not as good as the best non-VR exposure, while of course VR can allow you to get certain shots that can never be had in a non-VR method.
The lens used for all shots above was 70-200 VR2. The first two shots show unconverted NEF previews. Due to the limitations of preview files embedded in NEFs, the effect would be even stronger if the shots were converted due to the extra detail revealed in the superior shot (however, since the effect was still quite clear, I did not bother on those two). For the latter three shots, you do see post-converted JPEGs, which were converted with no sharpening whatsoever, as to keep from obscuring the actual effect.
--
David Hill
http://www.sfbayweddingphotographer.com
San Francisco, CA | Austin, TX
Certified Wedding Photography Junky™
