Find the Maximum Shutter Speed for Vibration Reduction

It’s popular lore that you shouldn’t use shutter speeds beyond 1/500 second when you activate vibration reduction. Is this actually true? Is it the same for all lenses? How about different lens brands?

People aren’t generally aware of how helpful in-lens vibration reduction is, aside from enabling sharper slow-shutter shots. You get to see a steadier subject in your viewfinder, and it actually helps the auto-focus system work better. The AF system needs higher contrast to focus both quicker and more accurately; that’s exactly what the VR system provides. These are the reasons that I don’t want to turn off vibration reduction when I’m not using a tripod.

There have been many times where I forgot to turn off vibration reduction when using a fast shutter, and I was cursing myself for being forgetful. But when I looked at those shots where I forgot to turn it off, they looked really sharp. Was I fooling myself? Was shutter speed that crucial?

Instead of taking somebody’s word for it, I figured that I should find out the answer for myself. The obvious tools for this job are resolution measurement software and a good resolution target. I want to get actual measurement numbers, versus ‘impressions’ or guesses.

The experimental design is pretty simple: hand-hold the camera with vibration reduction active and shoot the resolution target. I’ll shoot with shutter speeds ranging from 1/250 through 1/4000 second.

Since this is probably a game of statistics, I should get a least 10 or 20 shots of the target at each shutter speed, to look for trends in resolution changes. I’m going to try my 105mm Micro Nikkor f/2.8 G VR IF-ED, the Nikkor 24-70mm f/2.8 E ED VR, and the Sigma 150-600 Contemporary.

I’ll make a table of the MTF50 lp/mm resolution versus shutter speed for each lens, and any significant resolution differences should show up. I know from past experience that I need to keep a constant aperture, and I also know that different ISO settings have a negligible effect on resolution until very high ISO values. Therefore, I’m setting my camera on “manual mode”, but with auto-ISO active. I get to control both the shutter and the aperture this way.

Normally, you’d want to be using Live View (with contrast detect) and lock the camera down on a sturdy tripod to get resolution readings, but that isn’t going to work in this scenario. My lenses have been carefully calibrated for phase-detect focus, and I’ll use continuous auto-focus (AF-C). I’ll use my gimbal head on a monopod, so that I won’t have to contend with changing distance while my camera/lens is still free to ‘wobble’ a little. This is my way of creating a “controlled hand-holding” experiment.

Even if the testing environment doesn’t result in optimal resolution, the resolution results should at least be representative relative to each other. The goal of this test is to see if resolution degrades beyond 1/500 second, rather than measure peak lens resolution.

So, how did it work out?

I took a total of 150 shots (10 shots at each shutter setting), with stabilization active in every single shot. I photographed my “A0” resolution target, which is roughly 4 by 5 feet. I used the three mentioned lenses, and everything was shot with the camera mounted on a gimbal and a monopod. I used un-sharpened RAW for all resolution measurements, with the MTFMapper program.

All photos were shot with AF-C, phase-detect focus on my Nikon D500. The camera was in manual-mode with auto-ISO, and I used shutter speeds ranging from 1/250 through 1/4000. The light level resulted in ISOs ranging from about 200 to 2800, depending upon the shutter speed in use. I only start worrying about the ISO affecting resolution beyond about 2400, so it’s just a minor concern.

Shutter 1/1000 with Sigma OS active: awesome resolution on D500

Sigma 150-600 at 150mm and f/5.6

Shutter MTF50 lp/mm average Measurement Range lp/mm

250 72.0 5

500 76.5 10

1000 77.5 5

2000 73.0 5

4000 73.5 5

Micro Nikkor 105mm f/2.8 VR at f/4.0

Shutter MTF50 lp/mm average Measurement Range lp/mm

250 60.6 5

500 60.0 5

1000 60.5 5

2000 60.5 5

4000 61.0 5

Nikkor 24-70 E VR f/2.8 at 70mm and f/4.0

Shutter MTF50 lp/mm average Measurement Range lp/mm

250 64.0 5

500 65.0 0

1000 65.0 0

2000 65.0 0

4000 60.0 0

Conclusion

Micro Nikkor 105 and Nikkor 24-70 lens VR: leave it on!

Sigma 150-600 Contemporary OS: leave it on!

The only lens that seemed to have a minor sensitivity beyond 1/2000 was my Nikkor 24-70 f/2.8 E VR. Even then, it was nearly a “don’t care” at 1/4000 shutter, since the resolution MTF50 lp/mm only dropped from about 65 to 60.

The Micro-Nikkor 105 f/2.8 didn’t register any significant change in resolution with stabilization active at any tested shutter speed.

The Sigma 150-600 resolution dropped about 5% going from 1/1000 to 1/2000 shutter. There wasn’t a significant resolution change going from 1/2000 to 1/4000.

I wouldn’t necessarily generalize these results to represent all lenses with stabilization. Older-technology lenses should be “verified” before assuming you can ignore the lens stabilization status.

All in all, this convinces me that I don’t really have to pay attention to turning off vibration reduction at higher shutter speeds for these lenses. Certainly I wouldn’t bother until shooting beyond 1/4000 before I worried about messing up resolution. Since I virtually never shoot beyond about 1/3000, I intend to keep VR active, unless my camera is mounted on a tripod.

I’d consider this stabilization 1/500 rule to be “myth busted”.

#howto