Autofocus mechanism

I just read a nice article referred by Horoscope fish discussing VR , it appears that the sampling rate for the VR is around 1000Hz , since all shutter movement occurs over 1/250 th of a sec,regardless of the shutter speed setting,, that means the camera is sampling and correcting several times during the motion of shutter.. which may suggest to one to turn OFF the VR.

It seems then ,that if the VR is making several adjustments to lens elements DURING the movement of the shutter , I'm wondering if the same is the case for the Autofocus in AF-C mode .. that is to say the elements are physically adjusting during the the actual exposure period , and if its possible that some other setting or methodology might occasionally render more detail on a Bird In Flight. ( I realize it may be more difficult ).

Additionally I'm wondering if the Mechanism which is adjusting as part of VR ,, is the same mechanism or assembly which is responsible for autofocus in AF-C mode.

In order for VR to work, it would have to be making adjustments during the entire exposure. If it corrected at a slower rate, it simply could not correct camera movement at all.

What the sampling rate does imply is there would be no need to use VR above 1/1000 sec, but then again, I can't shake that fast, so I probably would not need it.

I do wonder how much VR helps and does it sometimes hurt. Could explain some soft images at times.

AF is driven by entirely different sensors, so I imagine there is no relation to the VR system.

I don't pretend to have a grip on the technical aspects but from my experience I find I get better shots, meaning sharper shots, overall when I leave VR turned off. I believe VR to be a very specific solution to a very specific problem, not a general purpose "set it and forget it" feature. Try turning VR off and see for yourself.

My recipe for sharp shots looks like this:

VR is OFF if shooting a Nikon lens that supports it unless there is some VERY GOOD reason for me to have it on. As soon as that need passes, it's turned OFF again.

Quiet Shutter Mode unless I'm convinced I'm going to need to shoot in Continuous Mode which I rarely do.

Shutter speed equates to twice the focal length. Shooting at 100mm means a minimum 1/200s shutter speed. I shoot at as high an ISO as required to stick to this guideline.

....

I agree ,that VR could explain some soft images- If there was a delay of 1-1000 of a sec between the command to release the shutter and its commencement to actually move, Id figure the camera could have two data points to predict motion and make a smooth correction during the exposure based on that prediction.
In fact ,I thought that was what predictive 'focus' was supposed to do, and that VR was similar, but since shutter operation time is always at sync speed rather than ones speed setting , then the consideration should then be that any camera motion one would see at 250th of a sec or so , should still impact an image even if the shutterspeed was 1-8000th of a sec.
1-1000th of a second is a somewhat arbitrary shutter speed selection as a breakpoint , since even at that number the camera is still having enough time to sample and adjust at least four or five times no matter what speed setting one chooses.
The book is very vague about exactly what the camera is mechanically doing.

I agree that , whatever techniques work best for you , are the ones you should use.
BUt I'm not as experienced as yourself and so I'm still open to ideas that will lead me to my own best technique, or understanding things like why prime lenses often don't have features like VR.

For Me Personally, I can hold a 300mm lens still down to about a 250th of a sec without VR reasonably reliably,( as I expect you can as well) - so I dont really adopt these sorts of rules of thumb. so Im not doubling or quadrupling my shutter speed for no reason.
I want to know what the camera is doing instead... but as you suggest , I may spend more time with the VR off.

Stoshowicz said:

I agree that , whatever techniques work best for you , are the ones you should use.
BUt I'm not as experienced as yourself and so I'm still open to ideas that will lead me to my own best technique, or understanding things like why prime lenses often don't have features like VR.

For Me Personally, I can hold a 300mm lens still down to about a 250th of a sec without VR reasonably reliably,( as I expect you can as well) - so I dont really adopt these sorts of rules of thumb. so Im not doubling or quadrupling my shutter speed for no reason.
I want to know what the camera is doing instead... but as you suggest , I may spend more time with the VR off.

Click to expand...

Well, first and foremost I encourage you to experiment and figure out what works for you. Absolutely. That being said, I'm not asking you to double or quadruple your shutter speed for no reason.

The reason this rule of thumb exists (shutter speed = 1.5x focal length) has to do with the magnification of focal length combined with the DX sensors field of view. When shooting at, say, 300mm the image we see, obviously, is magnified... And, rather critically, so is everything else; including camera shake. Just like shooting a rifle with telescopic sight attached; sure it magnifies the target but it also magnifies every tiny shake and wiggle the shooter communicates to the rifle not matter how small. Now we need to combine this magnified camera shake (brought on by the increased focal length) along WITH the 1.5x "crop factor" of the DX sensor. The situation is, dare I say, magnified again and magnified equally, meaning one and-a-half times in addition. Bit of a one-two punch going on here you see.

So... When I suggest you keep your shutter at a speed that is twice the operating focal length, it's not without some degree of logic and understanding of photographic principle backing it up. But again, please feel free to experiment and determine best what works for you.

.....

Horoscope Fish said:

I don't pretend to have a grip on the technical aspects but from my experience I find I get better shots, meaning sharper shots, overall when I leave VR turned off. I believe VR to be a very specific solution to a very specific problem, not a general purpose "set it and forget it" feature. Try turning VR off and see for yourself.

My recipe for sharp shots looks like this:

VR is OFF if shooting a Nikon lens that supports it unless there is some VERY GOOD reason for me to have it on. As soon as that need passes, it's turned OFF again.

Quiet Shutter Mode unless I'm convinced I'm going to need to shoot in Continuous Mode which I rarely do.

Shutter speed equates to twice the focal length. Shooting at 100mm means a minimum 1/200s shutter speed. I shoot at as high an ISO as required to stick to this guideline.

....

Click to expand...

This is what I have been working toward recently. I do think I get better results with it off and shooting fast enough to not have a need for it.

Horoscope Fish said:

Well, first and foremost I encourage you to experiment and figure out what works for you. Absolutely. That being said, I'm not asking you to double or quadruple your shutter speed for no reason.

The reason this rule of thumb exists (shutter speed = 1.5x focal length) has to do with the magnification of focal length combined with the DX sensors field of view. When shooting at, say, 300mm the image we see, obviously, is magnified... And, rather critically, so is everything else; including camera shake. Just like shooting a rifle with telescopic sight attached; sure it magnifies the target but it also magnifies every tiny shake and wiggle the shooter communicates to the rifle not matter how small. Now we need to combine this magnified camera shake (brought on by the increased focal length) along WITH the 1.5x "crop factor" of the DX sensor. The situation is, dare I say, magnified again and magnified equally, meaning one and-a-half times in addition. Bit of a one-two punch going on here you see.

So... When I suggest you keep your shutter at a speed that is twice the operating focal length, it's not without some degree of logic and understanding of photographic principle backing it up. But again, please feel free to experiment and determine best what works for you.


.....

Click to expand...

Yes sir I understand the reasoning, though I'm not finding it necessary to abide with its precaution. Because shutter motion duration is always of a minimum duration which is the same as syncspeed the distortions due to motion which would be seen at 1-250th of a sec are still there in axis of shutter closure. You just cant escape that diagonal distortion by- reducing the duration which individual pixels experience are being lit. The pixel in the upper left hand corner always is obscured at least 1-250th of a second before the pixel at the lowermost right corner. Unable to correct this other than by actually being very still,, one is largely just making the image darker , and so requiring higher iso or wider aperture. But yes I do understand why they use the formula they do. ( at least , this is how I'm thinking the whole thing works this way)
( eyelight was using a reference guide number of 1-1000th of a sec which is four times faster than my syncspeed)
If a person can use a tripod and avoid this precaution of 1.5-2x focal length, the expectation is based on an assumed amount of wobble that a human exhibits.

@Stoshowicz , not sure I am following your complete idea here. Are you theorizing that because the shutter travel across the sensor is at a fixed rate, that faster shutter speeds than that rate have no additional effect on reducing blur from camera shake??

Stoshowicz said:

I just read a nice article referred by Horoscope fish discussing VR , it appears that the sampling rate for the VR is around 1000Hz , since all shutter movement occurs over 1/250 th of a sec,regardless of the shutter speed setting,, that means the camera is sampling and correcting several times during the motion of shutter.. which may suggest to one to turn OFF the VR.

It seems then ,that if the VR is making several adjustments to lens elements DURING the movement of the shutter , I'm wondering if the same is the case for the Autofocus in AF-C mode .. that is to say the elements are physically adjusting during the the actual exposure period , and if its possible that some other setting or methodology might occasionally render more detail on a Bird In Flight. ( I realize it may be more difficult ).

Additionally I'm wondering if the Mechanism which is adjusting as part of VR ,, is the same mechanism or assembly which is responsible for autofocus in AF-C mode.

Click to expand...

No, not the same mechanism for VR (in the lens body) and AF-C (in the viewfinder focus system). VR does not focus (does not rack the lens in and out), but instead just moves a central lens element laterally to keep subject centered (stable position).

If the shutter speed is sufficient to freeze any camera shake, it seems at best a moot point if VR is on, and at worst, it could only hurt if it did something. Nikon lens manuals all say to turn VR off if on a tripod (not needed then... and seems not needed with a fast shutter speed.) A tripod could imply a slow shutter speed, but they say turn VR off. But if you are going to hand hold 1/10 second, I'd say turn it on.

And viewfinder AF-C is phase detect focus... instead of seeking to hunt a better focus (like Live View does), the viewfinder phase shift can already know which way to go, and how far, and it simply just goes there, a jump. That would seem an improvement if it did do anything.

But... it can't.

AF-C is different in Live View (mirror is up, shutting out viewfinder mechanism). It is called AF-F in Live View, and instead of focusing at half press, it always focuses, but says it locks when shutter is released. Viewfinder AF-C says it focuses continuously at half press, but does not mention during shutter release. But during shutter release, the mirror is up, and the viewfinder focus mechanism is shut out.

Stoshowicz said:

Yes sir I understand the reasoning, though I'm not finding it necessary to abide with its precaution. Because shutter motion duration is always of a minimum duration which is the same as syncspeed the distortions due to motion which would be seen at 1-250th of a sec are still there in axis of shutter closure. You just cant escape that diagonal distortion by- reducing the duration which individual pixels experience are being lit. The pixel in the upper left hand corner always is obscured at least 1-250th of a second before the pixel at the lowermost right corner. Unable to correct this other than by actually being very still,, one is largely just making the image darker , and so requiring higher iso or wider aperture. But yes I do understand why they use the formula they do. ( at least , this is how I'm thinking the whole thing works this way)
( eyelight was using a reference guide number of 1-1000th of a sec which is four times faster than my syncspeed)
If a person can use a tripod and avoid this precaution of 1.5-2x focal length, the expectation is based on an assumed amount of wobble that a human exhibits.

Click to expand...

Assuming I am understanding your correctly (I'm not entirely certain I am); and addressing only the bold-ed part of your post: I'm going to have to say "No".
The shutter mechanism is comprised of not one, but two curtains, which operate in concert such that all the pixels across the entire sensor receive equal exposure regardless of shutter speed.

....

Eyelight said:

@Stoshowicz , not sure I am following your complete idea here. Are you theorizing that because the shutter travel across the sensor is at a fixed rate, that faster shutter speeds than that rate have no additional effect on reducing blur from camera shake??

Click to expand...

I don't want to say yes and no , because I know it sounds crazy .. (I'm trying to get a handle on whats going on) ,, so lemme give you the scenario to explain the phenomenon as Im thinking it is , ,,, You set up an object- a tennis ball- to exactly have its image traverse the sensor in 1-25oth of a sec ,, well if that object is timed to be in the exposing zone that the shutters reveal , for the entire time that the shutters are moving then the image captured across the sensor should be a long stripe of green. If that tennis ball is moving in the opposite direction , then the resultant image of the ball will be very much compressed.
So the actual answer I would give you is that the effect will vary depending upon which direction you shake the camera. So if one wobbles randomly the expected effect of having the slot traverse the sensor will also be random as an ' improvement' .

Horoscope Fish said:

Assuming I am understanding your correctly (I'm not entirely certain I am); and addressing only the bold-ed part of your post: I'm going to have to say "No".
The shutter mechanism is comprised of not one, but two curtains, which operate in concert such that all the pixels across the entire sensor receive equal exposure regardless of shutter speed.

....

Click to expand...

Yes , I'm aware that there are two curtains, whether they traverse in the same direction and then go the other way at some point or whether they always go in opposite directions I don't know.. but I do know that curtains obscure light and at higher shutter speeds over sync speed , the entire sensor is not exposed all at the same time , there is always a differential time element as to WHEN a particular bayer array begins its duration of exposure relative to other arrays on other parts of the sensor. What I've seen of mistimed flash exposures suggests that the exposure opening is a diagonal slot.

WayneF said:

No, not the same mechanism for VR (in the lens body) and AF-C (in the viewfinder focus system). VR does not focus (does not rack the lens in and out), but instead just moves a central lens element laterally to keep subject centered (stable position).

If the shutter speed is sufficient to freeze any camera shake, it seems at best a moot point if VR is on, and at worst, it could only hurt if it did something. Nikon lens manuals all say to turn VR off if on a tripod (not needed then... and seems not needed with a fast shutter speed.) A tripod could imply a slow shutter speed, but they say turn VR off. But if you are going to hand hold 1/10 second, I'd say turn it on.

And viewfinder AF-C is phase detect focus... instead of seeking to hunt a better focus (like Live View does), the viewfinder phase shift can already know which way to go, and how far, and it simply just goes there, a jump. That would seem an improvement if it did do anything.

But... it can't.

AF-C is different in Live View (mirror is up, shutting out viewfinder mechanism). It is called AF-F in Live View, and instead of focusing at half press, it always focuses, but says it locks when shutter is released. Viewfinder AF-C says it focuses continuously at half press, but does not mention during shutter release. But during shutter release, the mirror is up, and the viewfinder focus mechanism is shut out.

Click to expand...

Im not sure I understand the conclusion I should gather ,, is the camera still trying to maintain proper focus distance (and VR ) to my target while the mirror is up ? (with either phase or contrast based focus sensors) or is it 'flying blind' for the moment? (not in live view)

Stoshowicz said:

You set up an object- a tennis ball- to exactly have its image traverse the sensor in 1-25oth of a sec ,, well if that object is timed to be in the exposing zone that the shutters reveal , for the entire time that the shutters are moving then the image captured across the sensor should be a long stripe of green. If that tennis ball is moving in the opposite direction , then the resultant image of the ball will be very much compressed.

Click to expand...

How real is this question? A 35mm DX lens at 20 feet has a field of view of 13.75 feet. If tennis ball covers the frame width in 1/250 second, it would be going 2343 MPH. I think its hard enough to deal with real problems instead of going looking for them.

At a flash sync speed of 1/250 second, the shutter is fully open, not a slit.

Stoshowicz said:

Im not sure I understand the conclusion I should gather ,, is the camera still trying to maintain proper focus distance (and VR ) to my target while the mirror is up ? (with either phase or contrast based focus sensors) or is it 'flying blind' for the moment?

Click to expand...

VR yes, works while the shutter is open, but it does not change focus.

AF-C, out of consideration while the shutter is open.

VR does not affect subject motion, only camera motion.

Stoshowicz said:

I don't want to say yes and no , because I know it sounds crazy .. (I'm trying to get a handle on whats going on) ,, so lemme give you the scenario to explain the phenomenon as Im thinking it is , ,,, You set up an object- a tennis ball- to exactly have its image traverse the sensor in 1-25oth of a sec ,, well if that object is timed to be in the exposing zone that the shutters reveal , for the entire time that the shutters are moving then the image captured across the sensor should be a long stripe of green. If that tennis ball is moving in the opposite direction , then the resultant image of the ball will be very much compressed.
So the actual answer I would give you is that the effect will vary depending upon which direction you shake the camera. So if one wobbles randomly the expected effect of having the slot traverse the sensor will also be random as an ' improvement' .

Click to expand...

Not near as crazy as some things that are heard around the web, but the handle might be a bit farther away than it appears.

Wayne beat me to the high speed of the tennis ball, but just for kicks, lets assume the tennis ball travels across the sensor in the same direction and at the same speed as the shutter curtain. The shutter curtain and tennis ball travel across the view in 1/250 second. If the shutter speed is set to 1/250 second, the 2nd curtain will start to close at the same moment the 1st curtain has reached full open. The tennis ball enters the view immediately behind the trailing edge of the 1st curtain and exits the field of view immediately after the 1st curtain reaches full open. Hold that.

Assume the tennis ball is 3" diameter, so using Wayne's 20' field of view of 13.75' the ball is 1/55 the field of view. So, the ball occupies a ball sized spot on the sensor for less than 1/55 the amount of time the rest of the image is exposed or 1/13750 second. This ball, we will never see, nor will it leave an image on the sensor.

If the ball were traveling in the opposite direction of the shutter travel then it would only be in view for half the exposure, but would still be moving too fast to see or record.

To see or record the image of the tennis ball, we need it to move slow enough to be exposed similar to the background, which means it needs to occupy the same position in the view for a significant amount of the exposure time. So, for instance, instead of traveling 13.75', it travels 3", in which case we get a 6" oval shaped green blur that has background showing through. 3" travel in 1/250 second is 62.5 FPS (42.6 MPH).

WayneF said:

How real is this question? A 35mm DX lens at 20 feet has a field of view of 13.75 feet. If tennis ball covers the frame width in 1/250 second, it would be going 2343 MPH. I think its hard enough to deal with real problems instead of going looking for them.

At a flash sync speed of 1/250 second, the shutter is fully open, not a slit.




VR yes, works while the shutter is open, but it does not change focus.

AF-C, out of consideration while the shutter is open.

VR does not affect subject motion, only camera motion.

Click to expand...

I was setting up a scenario that one could visualize that the distortion that occurs in the image, and how it affects the image,,
It does happen you know, its an artifact of the relative speeds of the thing moving across the screen and the time that the shutters take to expose the screen.
I did additional research over the weekend on this issue.
The narrowness of the effective shutter opening does suppress the effect of subject motion, and so there is some effect on blur due to subject motion , but its not a direct calculation of half the time equals half the motion. The distortion due to the time delay of the shutter getting across the sensor, will still smudge or compress the data to some extent. This effect if it affects even ten pixels across the image still can affect the image. and so there is a limit of detail one can get when the subject is moving.
That which is presented onto the sensor is always relative to the frame of the image, so one can pan with the moving subject or one can sit on the background with the subject blurring past ,but The VR is designed to remove vibrations which are of a particular frequency, so the importance- impact of the VR drops off at speeds faster than the sync speed, or at relatively slow shutter speeds where one would really require a tripod anyway.

Thanks for your input , but the comment about 'looking for problems' that don't exist isn't fair..
yes , I am looking at somewhat particular and esoteric subject matter., that's why I'm asking these questions of folks who seem experienced and well informed like yourselves,
but if all I wanted to know was vague rules of thumb and uninformative Nikon rhetoric I would have already read it in the manual.

Yes Youre correct that at sync speed the "slit would extend across the entire frame"

Yes, for focal plane shutter speeds faster than sync speed, the shutter speed (exposure time) and the shutter travel time are two very different concepts.

from Four Flash Photography Basics we must know - Maximum Shutter Sync Speed
A very interesting picture (a classic) using a focal plane shutter is this 1912 picture by Jacques Henri Lartigue. The shutter travel motion can cause distortion of a moving subject. The shutter slit was moving down the picture there (upwards in the camera, but we see it after inversion). The spectators are leaning left because the 4x5 camera was being panned right, following the race car. Their feet were exposed earlier than their heads, at slightly different times, due to the moving slit, and the panned camera. The car was going faster yet, so it leans the other way. This classic picture is responsible for us imagining speeding wheels as being slanted ovals, at least in cartoons. Shutters are much faster today, but the fastest motion can still be an issue (see the FP flash picture of the grinding disk on next page). The narrow open slit can provide a very short (fast) exposure, but it takes much longer for this slit to travel across all of the frame. The curtains cannot open or close instantly.

But yes, shutters are faster today, and "film size" is smaller (both effects are shorter travel time), so it becomes a minor effect today, hard to show except at fastest speeds. Sorry, I did think it was an unreal question.

That referenced try was at 1/8000 second (was about no motion stopping capability of HSS flash, but it shows focal plane distortion):
Four Flash Photography Basics we must know - Auto FP and HSS

WayneF said:

Yes, for focal plane shutter speeds faster than sync speed, the shutter speed (exposure time) and the shutter travel time are two very different concepts.

from Four Flash Photography Basics we must know - Maximum Shutter Sync Speed
A very interesting picture (a classic) using a focal plane shutter is this 1912 picture by Jacques Henri Lartigue. The shutter travel motion can cause distortion of a moving subject. The shutter slit was moving down the picture there (upwards in the camera, but we see it after inversion). The spectators are leaning left because the 4x5 camera was being panned right, following the race car. Their feet were exposed earlier than their heads, at slightly different times, due to the moving slit, and the panned camera. The car was going faster yet, so it leans the other way. This classic picture is responsible for us imagining speeding wheels as being slanted ovals, at least in cartoons. Shutters are much faster today, but the fastest motion can still be an issue (see the FP flash picture of the grinding disk on next page). The narrow open slit can provide a very short (fast) exposure, but it takes much longer for this slit to travel across all of the frame. The curtains cannot open or close instantly.

But yes, shutters are faster today, and "film size" is smaller (both effects are shorter travel time), so it becomes a minor effect today, hard to show except at fastest speeds. Sorry, I did think it was an unreal question.

That referenced try was at 1/8000 second (was about no motion stopping capability of HSS flash, but it shows focal plane distortion):
Four Flash Photography Basics we must know - Auto FP and HSS

Click to expand...

I didn't want to explain the questions I was addressing in real terms because I didn't want the question I was asking to be side stepped.. but Ill tell you some of the issues I am looking at,, a friend of mine shifted habit from regular shutter button operation , to back-button focusing , he feels his images are coming out clearer and really likes this other technique.. but why would this make a difference if this technique is the same as single finger pressing of the shutter release button? either the mechanisms for focus or VR are being used more effectively.. the answer may affect whether he routinely uses the VR or not..another issue is to answer, just exactly how much wing feather detail might I expect to get on birds and bugs in flight,which really isn't feasibly replicated shot to shot.. I'm occasionally really cranking up on shutter speed I can gotten up to even 6000th of a sec and I still cant see the same fine scale details that I can on similarly distanced motionless birds but that softness could be for several reasons. Another issue , is that Im looking at getting a longer lens , do I need it to have VR if its going to be used at higher shutter speeds as well? If I got a faster lens , would this be better or worse at freezing for detail than trying to use a speedlight If I'm taking my pix in decent ambient light anyway?
What I'm trying to say here is that since a modern DSLR is such a very sophisticated device , there is a great deal if information which may be either immediately salient or generally informative, and for which trial and error isn't reliably certain ( because there IS so many situation specific factors- NOT because there isn't ). Regardless of the rules of thumb that some folks may find works for them - just physically speaking, without subjective overlay , VR isnt conclusively helping or harming anything in the image at shutter speeds over sync,, so you can just leave it on if you're shooting handheld and have the advantage of a steadier view of your bird. If you do most of your photography at very slow shutter speeds,,and if VR isnt reading the motions-vibrations in those situations anyhow its not doing anything then either.
If the feature is inactive, well I just dont care what someones subjective experience tells them is going on ,, -THE THING ISNT ACTIVATED

Im impressed with your math But like I said , its to demonstrate principle. But playing along, I never said how dark the background was nor how relatively bright the ball was , nor the fov of the lens , nor the distance to the ball. So those there calculations aren't valid speculations to the scenario. ( besides, I found out that the modern shutters open vertically) And I see you must agree that the ball would be elongated ,, which was the point I was trying to illustrate. And a 42.6mph rate of motion is entirely feasible for a vertically moving birds wing.