IS performance test

[xps]

Does anybody know some good tests of image stabilisation on lenses?
What is the maximal usable shutter speed on an lens an IS can stabilize?

An german magazine did one and was not really impressed by the IS of the 70-200 II 2.8
They see the 70-200 is only able to stabilize the lens at shorter than 1/200s on an 80D.
Winner is the Fuji Fujinon 2.8 50-140mm (shorter than 1/4 s)

Tested on an steve 6D system.

 

[Mt Spokane Photography]

There is no standard test, so individuals invent one. The results can vary widely depending on how its done. With a crop camera, at 200 mp, you are trying to stabilize what is effectively 320mm (1.6*200).

A 140mm lens is going to perform better.

This sounds like a case of incompetence on the part of the tester.

If he had tested a 35mm lens with IS, he would have declared it the winner.

 

[Viggo]

That's not right. I can have sharp shots with both a 70-200 and the 200 f2 at 1/15s at higher than 70% hitrate.
50

 

[Mt Spokane Photography]

That's not right. I can have sharp shots with both a 70-200 and the 200 f2 at 1/15s at higher than 70% hitrate.
50

It all depends on how you run the test. Standing still with arms braced or hard mounted to a motorcycle over washboard roads at 50 mph, there is no standard test method.

Even if there were a standard test method, perhaps mounting a camera to a vibration table which moves at various directions, speeds, and amounts, lens makers would design lenses to pass the test, and not to work as we actually use them.

 

[Deleted member 91053]

Much depends on the lens and how it handles. For example, in my hands, the 4 stop IS on my 800 F5.6 L IS is (to all intents and purposes) useless as I simply don't need it. However in the case of my 100-400 Mk2 IS is critical and this lens would be of little value without it! Even though the 100-400 has only half the potential focal length I can easily use shutter speeds of one stop or mere slower on the 800mm!

As many know I am the weirdo who hates IS - but on some lenses it does have it's place, only one so far but it is needed on the 100-400 Mk2 IMO. Never though I would say that

 

[xps]

That's not right. I can have sharp shots with both a 70-200 and the 200 f2 at 1/15s at higher than 70% hitrate.
50

It all depends on how you run the test. Standing still with arms braced or hard mounted to a motorcycle over washboard roads at 50 mph, there is no standard test method.

Even if there were a standard test method, perhaps mounting a camera to a vibration table which moves at various directions, speeds, and amounts, lens makers would design lenses to pass the test, and not to work as we actually use them.

This is the system on which it is tested on.
http://www.image-engineering.de/news/product-news/823-steve-6d-automated-ois-tests-with-six-degrees-of-freedom

 

[mb66energy]

That's not right. I can have sharp shots with both a 70-200 and the 200 f2 at 1/15s at higher than 70% hitrate.
50

It all depends on how you run the test. Standing still with arms braced or hard mounted to a motorcycle over washboard roads at 50 mph, there is no standard test method.

Even if there were a standard test method, perhaps mounting a camera to a vibration table which moves at various directions, speeds, and amounts, lens makers would design lenses to pass the test, and not to work as we actually use them.

This is the system on which it is tested on.
http://www.image-engineering.de/news/product-news/823-steve-6d-automated-ois-tests-with-six-degrees-of-freedom

The problem I see is that different people and different situations have different directions/amplitudes/spectral distributions of vibrations but the test stand is programmed with only one set of a situation or at best with a set of situations.
The only valid test is maybe ... use different lens-body combos in your photographic environments and compare. And redo that test while you are ageing

I am doing my first steps with an IS stabilized lens (EF f/4 70-200 L IS USM) and I am impressed that it gives me ca. 2.5 reliable stops (or ISOs) compared to the non-IS version of this lens.

 

[Mt Spokane Photography]

That's not right. I can have sharp shots with both a 70-200 and the 200 f2 at 1/15s at higher than 70% hitrate.
50

It all depends on how you run the test. Standing still with arms braced or hard mounted to a motorcycle over washboard roads at 50 mph, there is no standard test method.

Even if there were a standard test method, perhaps mounting a camera to a vibration table which moves at various directions, speeds, and amounts, lens makers would design lenses to pass the test, and not to work as we actually use them.

This is the system on which it is tested on.
http://www.image-engineering.de/news/product-news/823-steve-6d-automated-ois-tests-with-six-degrees-of-freedom

At least progress is being made, most lens reviewers still handhold lenses to see how slow they can set the shutter speed.

I've tested lots of equipment on vibration tables, but the cost is very high for aerospace mil-spec testing. I assume that this is less expensive.

 

[ahsanford]

There is no definitive method as our ability to manually stabilize the camera varies. So I always test this myself.

But the ballpark (really high level) method is to:

1) Turn IS off.

2) Take 10 (or 5, if you'd like) handheld pictures of a truly static target at a set shutter speed. Try to use as consistent a handheld shooting technique as you can -- I recommend against the 'I am being super grippy stable for this test' sort of approach if that's not how you shoot 90% of the time. The goal is not to hyper-concentrate and get a high score -- it's to reflect how the IS will work for how you shoot. (Han: "Fly casual.")

3) Either proofing on your computer or 100% pixel peeping on your LCD, rate how many hits vs. misses you took (i.e. 10 out of 10, 8 out of 10, etc.)

4) Double the shutter speed and repeat the process. Rinse and repeat.

The idea is to generate a hit list of shots that looks (something) like this:

WITHOUT IS
1/30 = 10 out of 10
1/15 = 10 out of 10
1/8 = 9 out of 10
1/4 = 5 out of 10
1/2 = 2 out of 10

5) Now you turn on the IS and repeat 2-4. You'll get a similar trend, but the falloff in 100% hit rate will occur with longer shutter speeds, say like this:

WITH IS
1/30 = 10 out of 10
1/15 = 10 out of 10
1/8 = 10 out of 10
1/4 = 10 out of 10
1/2 = 9 out of 10
1s = 7 out of 10
2s = 3 out of 10

You won't get a perfect comparison from this sort of approach, but from this you could either say:

My slowest 100% hit rate is 1/15 without IS and 1/4 with IS, so the IS is good for roughly two stops for that lens at that focal length (if it's a zoom).

OR:

My hit rate starts to fall off around 1/4 without IS and between 1s and 2s with IS, so it's maybe 2 1/2 stops.

It's not an exact science. Shoot, see how it does, and intelligently adjust your acceptable 'shutter speed floor' accordingly.

- A

 

[ahsanford]

Does anybody know some good tests of image stabilisation on lenses?
What is the maximal usable shutter speed on an lens an IS can stabilize?

An german magazine did one and was not really impressed by the IS of the 70-200 II 2.8
They see the 70-200 is only able to stabilize the lens at shorter than 1/200s on an 80D.
Winner is the Fuji Fujinon 2.8 50-140mm (shorter than 1/4 s)

Tested on an steve 6D system.

And despite everything I just posted -- which is the bush-league simple way to test it -- it depends. Canon lenses have a variety of different IS setups depending on what you shoot and how you shoot it. For instance, stabilizing a macro lens is different than panning while tracking a bird.

So when in doubt, ask the forum and someone with far more knowledge than I have can set you straight.

- A

 

[Crosswind]

Hi and sorry for digging out this older conversation, but my question fits exactly in here, so I didn't want to start a new topic.

For example; when I buy a Canon 35mm f/2.0 IS USM, and mount it onto my EOS M5 (via adapter), then I will basically have the f.o.v. of a classic ~50mm prime on a FF camera - but will the image stabilizer inside the lens also recognize that it is working on an APS-C body and therefore amplify/increase the effectiveness of its stabilization levels (because of the crop factor) - so that it matches the behavior of an image stabilizer inside a regular 50mm prime on FF?

My concern is, that I'd loose a bit of the stabilization-advantage because I'm not using it on FF for which this lens (and its IS) is designed. With my M5, having a ~3x higher pixel density than my old 6D, means that shaking is much more visible, so I thought the stabilizer should compensate for that fact. But I'm not sure if that's actually the case.

Thanks in advance for your reply.