Challenging Conventional Wisdom/Rules of Thumb

[RustBeltRaw]

In response to another conversation, elsewhere: I made this set of images.

Another interesting feature of that comparison is how much hand shake looks like bokeh.

[Cantab]

This discussion has inspired me to do a bit of testing with VR on/off at higher shutter speeds. We're up to about 18 hours of daylight now so finding time to do so shouldn't be a problem. If anyone reading has done any testing of there own with VR, maybe you could start another thread on that topic

I'll look forward to seeing the results of your testing.

[Dave Humphries]

This discussion has inspired me to do a bit of testing with VR on/off at higher shutter speeds. We're up to about 18 hours of daylight now so finding time to do so shouldn't be a problem. If anyone reading has done any testing of there own with VR, maybe you could start another thread on that topic

I did this once and I think (from memory) I determined that of 20 shots, 10 taken each way, 6/10 were better with VR on, there's a thread here somewhere. IF I find it, I'll add a link here, but don't hold your breath Dan.

Cheers,

[DanK]

It is argued that if your ss is faster than 1/500 (half of 1000hz sample rate) then IS/VR/OS can potentially hurt more than help.

I had never heard this, so I googled. I found this 5-year-old thread that included this response from Canon:

Canon does not quote the sampling frequency for its Image Stabilizer mechanisms. Also, Canon does not recommend users to avoid using IS at fast shutter speeds. The visual effects of IS in captured images diminish as the shutter speed increases over 1/focal length, but the use of IS for moving subjects in these conditions can be beneficial because it presents a steadier image to the camera's AF detection mechanism.

[NorthernFocus]

From the same thread:

...some photographers may prefer to shut it off at least occasionally depending on their shooting style. Bottom line, it makes no sense to declare that IS is either "all good" or "all bad" when it comes to bird photography. Use it when you need it...

For Canon (who is the claimed source in the thread) to provide and answer like this suggests to me that there are indeed conditions under which IS can be detrimental to IQ. If it were black and white that it always helps or at least doesn't hurt, why not say so?

The bottom line of all of this discussion is that in order to truly perfect the craft, one has to learn one's equipment's strengths and limitations. One can rely on rules of thumb, articles on technique, equipment reviews, etc, etc, as guidance. But at the end of the day, we all have to do what we know through our own experience works best. One unfortunate downside with the digital age is that there are more variables (or more under the photographer's control) than ever which makes solving the equation more difficult. Then again, that is one of the very things that makes photography attractive as a hobby and challenging as a profession isn't it. There's not book that one can blindly follow to go out and make a good photo

Dave, if you do run across that link by all means post it for us.

In the mean time, I'll be silent for the next few days because why? Because I'll be out shooting !

[William W]

No question IS/VR/OS works. My comment was in reference to arguments about sample rate vs shutter speed. It is argued that if your ss is faster than 1/500 (half of 1000hz sample rate) then IS/VR/OS can potentially hurt more than help. . .

AND

I had never heard this, so I googled. I found this 5-year-old thread that included this response from Canon:

The Nikon commentator, Thom Hogan, has a lot to say on the subject of 1/500s and using VR.

It seems to me that Thom is the main person who does a lot of that arguing, specifically about that shutter speed and its relationship to the sampling frequency. I’ve read most (if not all) of what Thom has written on the subject.

*

I haven’t made any extensive tests using Canon Lenses and never really have I specifically addressed the “Tv ≤ 1/500s” situation: but I have and do use a variety of Image Stabilized Lenses, hand-held at fast shutter speeds.

More recently and most regularly these lenses are, but not limited to: EF 24 to 105F/4 L IS; EF 70 to 200F/2.8L IS; EF 70 to 200F/4 IS; EF 300/2.8L IS; EF 400F/2.8L IS; EF-S 17 to 55 F/2.8 IS; EF-S 18 to 55F/3.5~5.6 IS and EF-S 55 to 250 F/4~5.6IS

In this practice, I have found no ill effects by using the Image Stabilization “ON” when shooting hand-held, at, or faster than a Shutter Speed of 1/500s.

However, I cannot conclude that having Image Stabilization “OFF” would have made for better IQ – I can only say that the images I made with IS “ON”, were, relatively for that lens, always “very sharp”.

*

Perhaps this analysis should only be limited to Nikon VR, or more correctly, perhaps should exclude Canon Image Stabilized lenses.

Or perhaps the mathematics of Thom’s predications and the resultant flow of his “logic” should be questioned? (I think this point has strong legs.)

There are quite a few (Nikon users) who disagree with Thom’s hypothesis.

WW

[William W]

(Originally Posted by William W - In response to another conversation, elsewhere: I made this set of images.)

Another interesting feature of that comparison is how much hand shake looks like bokeh.

I'd never thought of that and how it might cause confusion or misanalysis, thanks.

WW

[NorthernFocus]

This is exactly why I didn't bring up VR in the OP (and only did so because someone earier seemed to have misinterpreted one of my comments). There are a lot of opinions for/against but I've yet to seen anything that I'd call test data supporting the arguements I've read. Granted, many opinions supported by years of experience from credible sources, present company included, but no quantitative data. When I get back from my impending weekend trip, I'll try to set something up that resembles a defensible test protocol and use the FocusTune software as an impartial form of analysis of sharpness with/without VR at various shutter speeds.

To save some of you the trouble of pointing it out, I fully recognize the irrelevance of such testing. Technically/measurably ultrasharp is of absolutely no added value versus visibly sharp to the human perception. Unfortunately, my DNA seems to possess the "inquiring minds want to know" gene Hopefully, your collective conciences prevent your sleeping with the knowledge that you've stirred the aforementioned beast within and have therefore doomed me to who knows how many hours of pointless data generation, recording, and analysis. Oh, wait.... that's what I do for a living But in this case it won't pay nearly as well as it normally does

[Wayland]

Half my life ago I could confidently hold a long lens at a speed equivalent to half it's focal length.

These days I find that twice it's focal length is closer to the mark. (Disregarding anti shake systems.)

I don't think the technology has much to do with that change...

[RustBeltRaw]

All this is further proof that no amount of camera technology will ever replace photographers. You still need good techniques, you still need to fight for access, you still need to get up at 3AM, and you still need to point your camera at interesting stuff. As (I believe) the ranking young whippersnapper on this thread, I'm counting the blessing of being able to hand-hold down to about the 2/(focal length) threshold Wayland mentioned. But I am studying the ancient wisdom of great masters whose kung fu is best to try and push that. Amazing how hard it is to train your non-dominant eye. My horizon lines are going to be all over the place for a while.

I'd never thought of that and how it might cause confusion or misanalysis, thanks.

Insight from shooting indoor action at f2.0. Panning with sliver-thin DoF and servo autofocus is something of a fool's errand. Still haven't found a foolproof way of individually identifying each problem, but it helps to look for more in-focus areas in each shot. Those can be key to determining where your focal plane actually wound up. I've been shooting the same venue about once every 5 weeks for nearly 18 months, and my settings are still evolving.

[William W]

Insight from shooting indoor action at f2.0. Panning with sliver-thin DoF and servo autofocus is something of a fool's errand. . . etc

What sport?

WW

[jcuknz]

Hey DAN you are supposed to be out shooting not adding to this discussion today

The 1/f assumes you know how to hold your rig and operate the trigger.

If I say any more I will probably get lynched.

[NorthernFocus]

Hey DAN you are supposed to be out shooting not adding to this discussion today
...

Just returned. We had two whole days of uncharacteristically (mostly) sunny weather out on Prince William Sound. I did do some high ss shooting with and without VR. We'll just see what we'll see....

[Glenn NK]

If you pay attention to crop factors that's 630mm effective plus the camera has a function to shoot in 1.3 crop so effectively like shooting a 14MP full frame sensor with an 820mm lens. Using the ages old strategy of keeping a minimum shutter speed of 1/focal length, I started the shoot at 1/800ss and as light increased moved progressively to 1/1250, 1/1600, and ultimately 1/2000. Wide open aperture at 5.6 the entire time and when I got to 1/1600 I also began to tweak my ISO a bit.

Dan:

Since you are using the same camera, but changing the size of the captured image (through an electronic adjustment), this wouldn't seem to me to affect the "1/FL" shutter setting would it?

I would think the 1.4 factor for the extender should be included.

Only the actual FL and the extender would have a bearing on the optics, and hence the 1/FL shutter speed.

Glenn

[jcuknz]

My understanding of the rule is that it is talking about the 'Angle of View' and arose before we had all these modern gadgets and things to confuse people. How much displacement of the subject is acceptable during the exposure.

So if as I did some year back made a quarter crop, and some, of a surfer when hand holding my 950mm rig I should have used a 1/2000 shutter speed .... I wasn't only 1/400 but had OIS to help. Plus the fact I was tracking with a moving object rather than trying to hold the camera steady on a stationary subject. How much it was a burst sequence and I was tracking 'blind', another factor I don't know how to evaluate.

A further aspect not mentioned before is how much editing can help to make the results satisfactory
SOOC being a NO NO but after editing ... perhaps .... for web use anyway

'Going Going Gone' posted awhile back.

[revi]

It's indeed mostly about the angle of view, which is why you have to use the '35mm equivalent' focal length, and not the one engraved on the barrel.
And that guideline might have been made for moderate tele lenses held by young and fit people. Someone a bit less fit handling a long (and heavy)
lens might very well find it a bit optimistic, especially after a hike.

Otoh, panning might actually help to stabilise the camera, as you give it a definite movement, and that helps limiting trembling.

[Manfred M]

And that guideline might have been made for moderate tele lenses held by young and fit people. Someone a bit less fit handling a long (and heavy) lens might very well find it a bit optimistic, especially after a hike.

Being neither particularly young nor particularly fit, I find that that a heavier lens and camera result in a much more stable shooting platform than a smaller, lightweight camera (I'm comparing shooting my D90 with the 55-200mm against the D800 with battery grip and f/2.8 70-200mm lenses). The additional mass damps out minor shaking (thank you inertia!) means that a this style of camera gives me better results than the smaller and lighter one. With a stabilization mechanism added to the equation, I find that I can handhold at far lower shutter speeds and longer lenses than I could when I was younger. On the other hand, I do take the light weight gear on a long hike, just because it is so much easier to carry.

All that being said; this is true only if one uses good shooting techniques and bracing the camera and body properly when shooting.

[RustBeltRaw]

What sport?

Roller derby. Indoors in a wood-paneled room, frequently after dark so the the one row of windows is unlit. Massive white balance gradient across the room (5600K windows to 3200K hot lights). It's a pain, but I get stuff like this.

[William W]

Roller derby.

Ah yes! thanks.
I now remember that from another, previous conversation.

Nice ation shot.

WW

[Manfred M]

It seems to me that Thom is the main person who does a lot of that arguing, specifically about that shutter speed and its relationship to the sampling frequency. I’ve read most (if not all) of what Thom has written on the subject.

Unfortunately Thom Hogan seems to be sadly mistaken as to how VR works. His premise is that the sampling frequency is going to limit the effectiveness of how quickly the stabilization system can compensate for camera movement. His assumption is that if the sampling frequency of the system is 1000Hz (1/1000th of a second) the maximum resolution of the system can only be ½ of this value or 1/500th of a second. In other words he suggests that any shutter speed faster than that (1/500th) cannot be corrected.

http://www.bythom.com/nikon-vr.htm

This is of course utter nonsense, because he makes the assumption that we will want to apply the correction just before the shutter was released, when in fact we want to compensate for what the camera is doing during the exposure. There is absolutely no way of doing this accurately using the technique that Hogan describes in the (in)famous article.

To a large extent, normal camera shake is sinusoidal, i.e. it bobs up and down and side to side We need to correct for motion that the camera is undergoing while the shutter is traveling, not what it was doing prior to the shutter firing, because the two are unlikely to be the same.

Fortunately there is a solution. While we often refer to the measurement instruments inside the lens as an electronic gyroscope. Well, in fact these instruments are called accelerometers, i.e. they measure the rate of change of speed (a.k.a. acceleration, hence the name of the instrument), not the speed itself; although speed and position can be calculated. The rate of change of speed is far less variable than the absolute speed, and if we want to go to a fourth order calculation, we can determine if the acceleration is changing (known as “jerk”) and compensate for that. With this data, the on-board processor can move the VR mechanism (really an optical element that rotates on two axes) in in synchronization and in the opposite direction to what the shaking motion is doing, regardless of shutter speed.