Canon & Nikon Image Stabilisation
Paul B
stabilisation to their DSLRs? Lens based stabilisation may well be better
but is rather dependent on the lens having it to start with! So why not have
the option of either?
tn...@mucks.net
You do have an option........Sony
Charles
"Paul B" <m...@privacy.net> wrote in message
news:_46dnWwZcr4wZkHU...@pipex.net...
Not me. It makes no sense for companies and customers that have invested
heavily in lens based stabilization systems that are known to work better.
The option for folks who disagree is to check out other types of cameras.
I have both types of IS and much prefer the in-lens method.
Having both could be a problem, by the way. Both IS systems could/should
never be active at the same time (chaos!).
Paul B
I gather that lens based is better than sensor shifting (until it goes
wrong.) Lens based IS can be switched off or sensor shifting be disabled
when an IS lens is fitted. I can remember the arguments between those saying
that shutter priority was better than aperture priority which I thought was
pretty daft when it was simple enough technically to have both. Canon
obviously agreed because they brought both out on the A1.
The technical difficulties of introducing both to either brand would be
fairly simple (both makes now use sensor shift for cleaning anyway.) Of
course either make may consider it bad for lens sales.
Paul B
ray
In photography, as in many other fields, what is 'better' is rather a
specious argument. All we require is sufficiency.
Paul B
> On Thu, 09 Apr 2009 00:28:20 +0100, Paul B wrote:
>
>> Thus spake Charles:
>
> In photography, as in many other fields, what is 'better' is rather a
> specious argument. All we require is sufficiency.
Does that apply to pixel count as well? :)
C J Campbell
Stabilization really does not do much, if anything at all, for wide
angle or macro photography. In fact, it causes problems in real
close-ups. Yes, Nikon makes a VR Micro-Nikkor, but you are supposed to
turn the VR off within a certain minimum distance from the lens.
Nevertheless, I have heard that Nikon, at least, is likely to introduce
a body with stabilization. It would not be a pro body -- Nikon seems to
prefer to innovate on their consumer cameras.
It is only a matter of time before they come out with cameras that tie
your shoes, too.
--
Waddling Eagle
World Famous Flight Instructor
C J Campbell
In-lens IS works better with long lenses. In-body works better with
wide angle lenses. Reason would say that long lenses need the
stabilization more.
Paul B
I'm currently using a Canon PS710IS P&S camera - much to my initial
surprise, the IS works very well & is lens based IIRC. I'm considering going
back to an SLR system. Although I've used Canon in my film days, my
preferred option is Nikon's D300 or D700 if my budget will stretch. My
choice of lens would be their 24-70mm f2.8 to start with. Nikon don't seem
to think VR is worth it under 200mm. At least Canon's EF-S 17-55mm f2.8 has
IS but Nikon's equivalent doesn't for example. I would suggest that anything
over 35mm can & does benefit from IS. With wideangle, I see little point in
IS.
ray
Certainly. Many folks are convinced otherwise and shell out for what they
don't 'need' and will probably never make decent use of.
David J Taylor
In-lens stabilisation also gives you the advantage of a stable image in
the viewfinder. I can appreciate that for those with existing lenses
without IS/VR the in body stabilisation offers /some/ improvement, albeit
a limited one. Today, the extra cost of stabilised lenses is not great,
and even the 3rd party suppliers offer in-lens stabilisation.
David
ASAAR
> Although I've used Canon in my film days, my preferred option
> is Nikon's D300 or D700 if my budget will stretch. My
> choice of lens would be their 24-70mm f2.8 to start with.
> Nikon don't seem to think VR is worth it under 200mm.
??? The 16-85mm VR Nikkor is a very nice lens. You have a point,
though, as most of Nikon's VR lenses are for their longer lenses.
> I would suggest that anything over 35mm can & does benefit from
> IS. With wideangle, I see little point in IS.
I disagree. I took a closeup picture of some people on a train a
week ago, coincidentally shot at 35mm. Although not tack sharp, it
still managed to look pretty good. Without IS it would have been
pretty awful, as the shutter speed was a looong 1/2 second.
ransley
I agree, why be forced to buy a more expensive lens with IS, then you
could also have dual IS Lens and camera working together. I have old
good Canon lenses that could benefit from an IS body. Canon should
wake up it would be a good selling point. Knowing everything
advances , features sell, it should happen, someday.
Ron Recer
"ransley" <Mark_R...@Yahoo.com> wrote in message
news:4f0c9e73-641c-4662...@z1g2000yqn.googlegroups.com...
sensor stabilization boby after body when you can have an IS lens that works
with many bodies and most agree the lens IS is superior to the sensor IS.
Ron
Wally
<david-...@blueyonder.neither-this-bit.nor-this.co.uk.invalid>
wrote:
On the other hand, lens stabilization can reduce optical performance
because the stabilizing elements will be off center. And besides, the
lens design will need to be compromised to accomodate those
stabilizing elements to begin with.
Wally
David J Taylor
[]
> On the other hand, lens stabilization can reduce optical performance
> because the stabilizing elements will be off center. And besides, the
> lens design will need to be compromised to accomodate those
> stabilizing elements to begin with.
>
> Wally
Whilst in theory that sounds plausible, do you have actual evidence of
your supposition? Comparisons of MTF curves?
For the moving sensor on the other hand, the lens would need to be
designed to cover a bigger image circle, and I have not seen evidence that
this is done.
So both systems may have compromises....
David
Chris Malcolm
> In-lens IS works better with long lenses. In-body works better with
> wide angle lenses. Reason would say that long lenses need the
> stabilization more.
Depends what you're doing and how you're doing it. If you're shooting
from the window of a moving vehicle a long lens is out of the
question, and a wide lens benefits greatly from stabilisation. I do
use long lenses as much as wide, but I nearly always use them with a
support with stabilisation turned off, so I find stabilisation more
useful at the wide end.
--
Chris Malcolm
ransley
> "ransley" <Mark_Rans...@Yahoo.com> wrote in message
Why not have it both ways, in camera and lens at the same time. For me
id rather save the expense of in lens IS but my lenses are small.
Paul B
> I disagree. I took a closeup picture of some people on a train a
> week ago, coincidentally shot at 35mm. Although not tack sharp, it
> still managed to look pretty good. Without IS it would have been
> pretty awful, as the shutter speed was a looong 1/2 second.
You are disagreeing!? As a general rule, the longer the lens, the more
useful IS is going to be. I've taken shots down to around 0.5" myself that
were sharp but my P&S doesn't go below 35mm equiv. All I can say is that IS
is very useful unlike stuff like face recognition or in-camera sepia filters
etc
ASAAR
Of course I'm disagreeing. Not with the assertion that IS is more
useful at longer focal lengths, but that it serves no useful purpose
at 35mm and shorter focal lengths. I might argue that point only
because Rita insists that it's so, but then she won't use IS/VR even
on long telephotos. Complications may arise with wide angle lenses
on DSLRs at the slow shutter speeds that don't get along very well
with mirror slap, but for wide angle lenses on P&S cameras (that go
to 24mm and wider), that complication isn't seen.
Paul B
> Most people use the lens longer than the camera body. Why keep
> paying for sensor stabilization boby after body when you can have an
> IS lens that works with many bodies and most agree the lens IS is
> superior to the sensor IS.
> Ron
Most new Canon & Nikon bodies have ultrasonic shift so they are already half
way there so I surmise that any added expense is going to be mostly a
one-off development cost that most other manufacturers have already carried
out.
--
Basically, I hate people who preface nearly every sentence with the word
'basically'!
Toby
"Charles" <charles...@comcast.net> wrote in message
news:grj940$e5h$1...@news.motzarella.org...
Personally, I don't think that Canon/Nikon would lose anything by including
inbody IS. With lens IS being superior, it would still make sense to include
it in high-end (and other) lenses, but would still give those with IS-less
lenses a couple of extra stops of latitude. It would also be completely
possible to have the camera switch off inbody IS whenever lens IS is active.
Toby
Bob Larter
> On Apr 8, 2:32 pm, "Paul B" <m...@privacy.net> wrote:
>> Does any agree that both Canon & Nikon should fit sensor based image
>> stabilisation to their DSLRs? Lens based stabilisation may well be better
>> but is rather dependent on the lens having it to start with! So why not have
>> the option of either?
>
> I agree, why be forced to buy a more expensive lens with IS, then you
> could also have dual IS Lens and camera working together.
It wouldn't work. Having both a lens IS & a body IS active at the same
time would likely result in oscillation of both IS systems, & might even
result in physical damage to one or both of them.
--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------
Chris Malcolm
Not much bigger, and the vignetting of lenses is a gradual function of
width. I have some full frame film camera lenses which always
vignetted noticeably wide open at the corners on full frame, and now I
have a 1.5 crop sensor I can if I want to be fussy spot the slight
vignetting in the corners there too. In other words over the slight
increase required I don't think it matters.
--
Chris Malcolm
David J Taylor
Looking at the vignetting on DX lenses under certain conditions, the
drop-off is quite sharp. But it's not just vignetting, it's all the other
aberrations which are worse off-axis. As I said, both systems have
compromises, if you want to look at it that way, but at least the in-lens
IS is desgined with those effects in mind, and they can be minimised....
Cheers,
David
Paul B
>> On Apr 8, 2:32 pm, "Paul B" <m...@privacy.net> wrote:
>>> Does any agree that both Canon & Nikon should fit sensor based image
>>> stabilisation to their DSLRs? Lens based stabilisation may well be
>>> better but is rather dependent on the lens having it to start with!
>>> So why not have the option of either?
>>
>> I agree, why be forced to buy a more expensive lens with IS, then you
>> could also have dual IS Lens and camera working together.
>
> It wouldn't work. Having both a lens IS & a body IS active at the same
> time would likely result in oscillation of both IS systems, & might
> even result in physical damage to one or both of them.
I think that is a given. As I & another poster have stated if a body detects
the attachment of an IS lens, it could automatically disable sensor shift.
What I don't know is if a future camera body can switch off the lens IS so a
wide to medium zoom could switch between the 2 as the lens is zoomed. Even
if it could not disable lens IS, having both need not cause more problems
than it is supposed to solve. It's not as if IS can't be switched off as
most people do when using a tripod for instance.
Doug McDonald
>
>
> It wouldn't work. Having both a lens IS & a body IS active at the same
> time would likely result in oscillation of both IS systems, & might even
> result in physical damage to one or both of them.
These systems are open loop; they
won't oscillate.
>
>
Doug McDonald
Chris Malcolm
They may be inidividually open loop, but if the output of each affects
the input of the other...
--
Chris Malcolm
Doug McDonald
No. They are open loop optically. The weight of each is not enough to effect the
accelerometer sensors that do the measuring. Being open loop means that he
computer that drives them would decide how much to move each system.
Doug McDonald
Grimly Curmudgeon
drugs began to take hold. I remember "Paul B" <m...@privacy.net> saying
something like:
>Does any agree that both Canon & Nikon should fit sensor based image
>stabilisation to their DSLRs? Lens based stabilisation may well be better
>but is rather dependent on the lens having it to start with! So why not have
>the option of either?
It'd be nice to have the choice, I agree. I bought a K10D mainly because
of the in-body IS and the ability to turn all my manual glass into IS
lenses was a major attraction. However, if I buy a FF Canon or Nikon, I
might be tempted to buy at least one IS lens for it.
I have no particular bias either way - it's a pity that many such
discussions on the merits of such technology get bogged down into an
either/or argument by fanbois.
Bob Larter
On what basis do you claim that they're open loop? Both systems rely on
servomechanisms to move either the lens group or the image sensor, &
I've never met a servo that wasn't closed loop.
And while I might wrong about them oscillating, I guarantee that you
wouldn't get correct image-stabilisation.
Bob Larter
I was wrong about that, because I'd forgotten that they use
accelerometers or gyros to sense camera shake, rather than optics. (If
they used the image to sense movement, you _would_ get oscillation with
two separate IS systems.)
Marco Tedaldi
> Doug McDonald wrote:
>> Bob Larter wrote:
>>>
>>>
>>> It wouldn't work. Having both a lens IS & a body IS active at the
>>> same time would likely result in oscillation of both IS systems, &
>>> might even result in physical damage to one or both of them.
>>
>> These systems are open loop; they
>> won't oscillate.
>
> On what basis do you claim that they're open loop? Both systems rely on
> servomechanisms to move either the lens group or the image sensor, &
> I've never met a servo that wasn't closed loop.
> And while I might wrong about them oscillating, I guarantee that you
> wouldn't get correct image-stabilisation.
>
>
loop.
the servo is closed loop in a sense, that the motor gives feedback on
it's position.
But to start from the beginning:
We have gyros/accelerometers which give a signal about the movement of
the lens. The controler in the camera or lens gives a signal to the
actuators (servos, whatever) in which position the seonsor oder lens
element should be moved.
That's a typical open loop configuration. It is even open loop, if the
position of the lens/sensor is sensed and feed back to the controler, to
adjust the positioning.
It becomes a closed loop, as soon as soon as the output of the
correction (in this case the image) is taken into account for a further
correction, but this does not seem to be the case.
I hope this clarifies a bit..
kruemi
BTW: Is phase-shift autofocussing open loop?
--
Agfa isolette, EOS 40D
http://flickr.com/photos/kruemi
And a cool timekiller: http://www.starpirates.net/register.php?referer=9708
David J Taylor
[]
> BTW: Is phase-shift autofocussing open loop?
I think the answer is "it depends".
Which mode? Which camera?
David
Marco Tedaldi
Canon EOS 40D in single shot (not tracking).
The only question is, does it determinate how far and in which direction
it is out of focus, or just measure while focussing. To me it seems,
that it is open loop, but I can't imagine how this is done.
kruemi
David J Taylor
> David J Taylor schrieb:
>> Marco Tedaldi wrote:
>> []
>>> BTW: Is phase-shift autofocussing open loop?
>>
>> I think the answer is "it depends".
>>
>> Which mode? Which camera?
>>
> Hm.. if you ask like this:
> Canon EOS 40D in single shot (not tracking).
>
> The only question is, does it determinate how far and in which
> direction it is out of focus, or just measure while focussing. To me
> it seems, that it is open loop, but I can't imagine how this is done.
>
> kruemi
Marco, my understanding is also that single-shot is open loop. It's
simply that you measure the focus error, and then drive the lens by that
measured amount of error (all given the appropriate units and scaling, of
course). I don't believe that focus error is measured while driving the
lens, but perhaps the lens position is?
As you are suggesting, the result may not be perfect, so it may be worth a
further half-press of the shutter to allow the auto-focus to correct any
remaining error.
But I have measured any of these systems, so perhaps anyone who has can
shed light on this?
Cheers,
David
Bob Larter
Doesn't matter. If the sensor drives the focus mechanism, it's closed
loop. If the lens is set to MF, it's open loop.
Bob Larter
The fact that AF systems can 'hunt' proves that they're closed-loop.
An open-loop system won't 'hunt'.
David J Taylor
> David J Taylor wrote:
>> Marco Tedaldi wrote:
>> []
>>> BTW: Is phase-shift autofocussing open loop?
>>
>> I think the answer is "it depends".
>>
>> Which mode? Which camera?
>
> Doesn't matter. If the sensor drives the focus mechanism, it's closed
> loop. If the lens is set to MF, it's open loop.
Have to disagree there, Bob. It's only closed loop if the focus sensing
remains active and providing input to the focus mechanism. If it's a
one-off measurement and a single calculated correction, it can be open
loop.
Cheers,
David
Marco Tedaldi
> David J Taylor wrote:
>> Marco Tedaldi wrote:
>>> David J Taylor schrieb:
>>>> Marco Tedaldi wrote:
>>>> []
>>>>> BTW: Is phase-shift autofocussing open loop?
>>>>
>>>> I think the answer is "it depends".
>>>>
>>>> Which mode? Which camera?
>>>>
>>> Hm.. if you ask like this:
>>> Canon EOS 40D in single shot (not tracking).
>>>
>>> The only question is, does it determinate how far and in which
>>> direction it is out of focus, or just measure while focussing. To me
>>> it seems, that it is open loop, but I can't imagine how this is done.
>>>
>>> kruemi
>>
>> Marco, my understanding is also that single-shot is open loop. It's
>> simply that you measure the focus error, and then drive the lens by
>> that measured amount of error (all given the appropriate units and
>> scaling, of course). I don't believe that focus error is measured
>> while driving the lens, but perhaps the lens position is?
>>
>> As you are suggesting, the result may not be perfect, so it may be
>> worth a further half-press of the shutter to allow the auto-focus to
>> correct any remaining error.
>
> The fact that AF systems can 'hunt' proves that they're closed-loop.
Only partwise. It depends on what part of the system we are looking at.
But Yes, if the focus really goes hunting it's a strong indication for a
closed loop overall. The contrast AF on my old Dimage A2 went searching
a lot, but I think we agree that contrast AF is definitely closed loop.
The AF on my 40D has not done this until now (but I've only taken 6000
and some pictures until now with this camera), so I can't say if this
might happen.
But Even than it might be an open loop which just checks the result and
starts over again, if the Result is not good enough.
The "proof" for open loop might be, that there seem to be problem
sometimes with systematic focussing errors which can be eliminated by
readjusting the lens. A closed loop system would correct such errors
inside the closed loop.
> An open-loop system won't 'hunt'.
>
A closed loop system should have no remaining error...
I'm not saying that you are wrong. It might be, that it is a closed
loop, but the "hunting" argument is not enough for me. And since there
is a fact that would suggest that the system is open, I'm not fully
convinced.
Greets and thanx for your feedback
Marco
David J Taylor
> David J Taylor wrote:
>> Marco Tedaldi wrote:
>>> David J Taylor schrieb:
>>>> Marco Tedaldi wrote:
>>>> []
>>>>> BTW: Is phase-shift autofocussing open loop?
>>>>
>>>> I think the answer is "it depends".
>>>>
>>>> Which mode? Which camera?
>>>>
>>> Hm.. if you ask like this:
>>> Canon EOS 40D in single shot (not tracking).
>>>
>>> The only question is, does it determinate how far and in which
>>> direction it is out of focus, or just measure while focussing. To me
>>> it seems, that it is open loop, but I can't imagine how this is
>>> done. kruemi
>>
>> Marco, my understanding is also that single-shot is open loop. It's
>> simply that you measure the focus error, and then drive the lens by
>> that measured amount of error (all given the appropriate units and
>> scaling, of course). I don't believe that focus error is measured
>> while driving the lens, but perhaps the lens position is?
>>
>> As you are suggesting, the result may not be perfect, so it may be
>> worth a further half-press of the shutter to allow the auto-focus to
>> correct any remaining error.
>
> The fact that AF systems can 'hunt' proves that they're closed-loop.
> An open-loop system won't 'hunt'.
Yes, I had also wondered about this, but of course there are two
components - the measurement of the focus error, and the mechanical
positioning of the lens to correct this error. The mechanical lens focus
servo could well be closed-loop, i.e. measuring the focus position
continuously and attempting to precisely position the focus to the
commanded position by measuring the /mechanical/ error, and this could
cause the slight overshoot or slight oscillation seen.
I had hoped that the focus measurement /was/ closed-loop, but when I last
asked here a number of people assured me that in single-shot, at least, it
was in fact open-loop. I'd love to see a reliable reference either way.
Cheers,
David
Bob Larter
> Bob Larter wrote:
>> David J Taylor wrote:
>>> Marco Tedaldi wrote:
>>> []
>>>> BTW: Is phase-shift autofocussing open loop?
>>>
>>> I think the answer is "it depends".
>>>
>>> Which mode? Which camera?
>>
>> Doesn't matter. If the sensor drives the focus mechanism, it's closed
>> loop. If the lens is set to MF, it's open loop.
>
> Have to disagree there, Bob. It's only closed loop if the focus sensing
> remains active and providing input to the focus mechanism.
Exactly my point.
> If it's a
> one-off measurement and a single calculated correction, it can be open
> loop.
Sure, but AF systems don't do that. They 'hunt' for the sharpest reading
on the AF sensor. It's a classic-closed loop system.
Bob Larter
Try using an old, slow AF lens with it. I see it often when I use my
ancient 100-300/5.6L zoom. With USM lenses, the hunting is generally so
fast that you can't see it under typical lighting conditions.
> But Even than it might be an open loop which just checks the result and
> starts over again, if the Result is not good enough.
The "starts all over again" part is what makes it closed-loop. If it
just moves the focus *once*, then gives up, that'd be open-loop.
> The "proof" for open loop might be, that there seem to be problem
> sometimes with systematic focussing errors which can be eliminated by
> readjusting the lens. A closed loop system would correct such errors
> inside the closed loop.
Not if (for example) there's a slight difference between the distance
from the lens to the AF sensor, vs the distance to the image
sensor/film. You'd get a slight focus offset in your photo, despite the
AF system getting perfect focus.
>> An open-loop system won't 'hunt'.
>>
> A closed loop system should have no remaining error...
See above. The image sensor & the AF sensor aren't 100% guaranteed to be
'seeing' an identical image. The closed-loop AF system only guarantees
perfect focus at the AF sensor.
> I'm not saying that you are wrong. It might be, that it is a closed
> loop, but the "hunting" argument is not enough for me. And since there
> is a fact that would suggest that the system is open, I'm not fully
> convinced.
Try the experiment of seeing whether your camera hunts for focus under
very poor lighting / low contrast subjects, using the oldest, slowest
lens you own.
> Greets and thanx for your feedback
No worries.
Bob Larter
I'd bet money that it's closed-loop - at least while the shutter's
half-pressed.
It should actually be pretty easy to test by focusing on something
nearby & moving the camera back & forth with the shutter half-pressed.
<tries it out>
Okay, what actually happens on my EOS 1Dmk2 in single-shot focus is that
lens definitely hunts while it's aquiring focus, but as soon as it does
(ie; when the focus confirm light comes on), it gives up & leaves the
focus there.
ASAAR
>>> BTW: Is phase-shift autofocussing open loop?
>>
>> I think the answer is "it depends".
>>
>> Which mode? Which camera?
>
> Doesn't matter. If the sensor drives the focus mechanism, it's closed
> loop. If the lens is set to MF, it's open loop.
You weren't specific enough. The phase-shift focusing used in
DSLRs does not use the sensor. It uses small AF sensors that
usually require lenses to have apertures not much smaller than
f/5.6. Unlike when using contrast detect AF, after a reading, if
the camera determines that the focus isn't quite there, it knows
which direction to move the lens elements. Is your answer still
applicable?
Bob Larter
> On Sun, 12 Apr 2009 05:02:46 +1000, Bob Larter wrote:
>
>>>> BTW: Is phase-shift autofocussing open loop?
>>> I think the answer is "it depends".
>>>
>>> Which mode? Which camera?
>> Doesn't matter. If the sensor drives the focus mechanism, it's closed
>> loop. If the lens is set to MF, it's open loop.
>
> You weren't specific enough. The phase-shift focusing used in
> DSLRs does not use the sensor.
Sorry. By 'sensor', I meant the AF sensor.
> It uses small AF sensors that
> usually require lenses to have apertures not much smaller than
> f/5.6.
Yes.
> Unlike when using contrast detect AF, after a reading, if
> the camera determines that the focus isn't quite there, it knows
> which direction to move the lens elements. Is your answer still
> applicable?
Yes. The important question is whether the AF system hunts (ie;
self-corrects) for focus. If it does, it's a closed-loop system.
Elsewhere in this thread, I posted about actually trying it out on one
of my cameras (EOS 1Dmk2), & yes, it does hunt for focus.
nospam
<bobby...@gmail.com> wrote:
> I'd bet money that it's closed-loop - at least while the shutter's
> half-pressed.
> It should actually be pretty easy to test by focusing on something
> nearby & moving the camera back & forth with the shutter half-pressed.
> <tries it out>
> Okay, what actually happens on my EOS 1Dmk2 in single-shot focus is that
> lens definitely hunts while it's aquiring focus, but as soon as it does
> (ie; when the focus confirm light comes on), it gives up & leaves the
> focus there.
Chris Malcolm
Experimental evidence trumps opinion. My Sony A350 DSLR hunts for
focus. The AF sensor tells it which way to go off hunting, and then it
keeps moving until it finds a sharp focus. If the light is poor and
the contrast dim it can sail right past focus until it hits the end
stop, when it turns round to look again. The "keep moving until sharp
focus found" is clearly a closed loop.
There is also obviously a check on the accuracy of focus found when it
stops, because sometimes in poor light and low contrast it will judder
a bit from side to side over focus until it's satisfied. With low mass
lenses this happens so fast it's hard to notice. It's obvious with big
heavy lenses. That too is clearly a closed loop.
Both these loops also have timers or counts on them, because if they
don't succeed pretty quickly they give up and report failure to focus.
--
Chris Malcolm
Bob Larter
Interesting.
---
If the starting point is so far out of focus that the sensor can't
identify a phase difference, the camera racks the lens once forward and
once backward to find a detectable difference. If it can't find a
detectable difference during that motion, it stops.
Although the camera does not take a "second look" to see if the intended
focus has been achieved, the lens does take a "second look" to ensure it
has moved the direction and distance commanded by the camera (it is a
"closed loop" system). This second look corrects for any slippage or
backlash in the lens mechanism, and can often be detected as a small
"correction" movement at the end of the longer initial movements.
---
nospam
<bobby...@gmail.com> wrote:
> >> I'd bet money that it's closed-loop - at least while the shutter's
> >> half-pressed.
> >> It should actually be pretty easy to test by focusing on something
> >> nearby & moving the camera back & forth with the shutter half-pressed.
> >> <tries it out>
> >> Okay, what actually happens on my EOS 1Dmk2 in single-shot focus is that
> >> lens definitely hunts while it's aquiring focus, but as soon as it does
> >> (ie; when the focus confirm light comes on), it gives up & leaves the
> >> focus there.
> >
> > <http://www.fredmiranda.com/forum/topic/241524>
>
> Interesting.
>
> ---
> If the starting point is so far out of focus that the sensor can't
> identify a phase difference, the camera racks the lens once forward and
> once backward to find a detectable difference. If it can't find a
> detectable difference during that motion, it stops.
>
> Although the camera does not take a "second look" to see if the intended
> focus has been achieved, the lens does take a "second look" to ensure it
> has moved the direction and distance commanded by the camera (it is a
> "closed loop" system). This second look corrects for any slippage or
> backlash in the lens mechanism, and can often be detected as a small
> "correction" movement at the end of the longer initial movements.
and just before that,
In one "look," it calculates the distance and direction the lens must
be moved to cancel the phase differences. It then commands the lens to
move the appropriate distance and direction and stops. It does not
"hunt" for a best focus, nor does it take a second look after the lens
has moved (it is an "open loop" system).
bit of both, it seems...
Bob Larter
I suspect that the description on that page may have lost a little
clarity in translation from the technical documents.
I've definitely seen way more than the two movements during focussing
that it describes.
I notice also that it doesn't say anything about what happens when
multiple focussing points have nearly equal partial focus, & it has to
choose between them.
Bob Larter
I've seen this exact same behaviour on my Canon DSLRs with slow, old
lenses. With fast (ie; USM & light) lenses, the motions are too quick to
see.
> Both these loops also have timers or counts on them, because if they
> don't succeed pretty quickly they give up and report failure to focus.
Ditto.
David J Taylor
[]
> Experimental evidence trumps opinion. My Sony A350 DSLR hunts for
> focus. The AF sensor tells it which way to go off hunting, and then it
> keeps moving until it finds a sharp focus. If the light is poor and
> the contrast dim it can sail right past focus until it hits the end
> stop, when it turns round to look again. The "keep moving until sharp
> focus found" is clearly a closed loop.
>
> There is also obviously a check on the accuracy of focus found when it
> stops, because sometimes in poor light and low contrast it will judder
> a bit from side to side over focus until it's satisfied. With low mass
> lenses this happens so fast it's hard to notice. It's obvious with big
> heavy lenses. That too is clearly a closed loop.
>
> Both these loops also have timers or counts on them, because if they
> don't succeed pretty quickly they give up and report failure to focus.
Sony does it that way, but not all cameras. Trump annulled! <G>
David
David J Taylor
> David J Taylor wrote:
>> Bob Larter wrote:
>>> David J Taylor wrote:
>>>> Marco Tedaldi wrote:
>>>> []
>>>>> BTW: Is phase-shift autofocussing open loop?
>>>>
>>>> I think the answer is "it depends".
>>>>
>>>> Which mode? Which camera?
>>>
>>> Doesn't matter. If the sensor drives the focus mechanism, it's
>>> closed loop. If the lens is set to MF, it's open loop.
>>
>> Have to disagree there, Bob. It's only closed loop if the focus
>> sensing remains active and providing input to the focus mechanism.
>
> Exactly my point.
>
>> If it's a
>> one-off measurement and a single calculated correction, it can be
>> open loop.
>
> Sure, but AF systems don't do that. They 'hunt' for the sharpest
> reading on the AF sensor. It's a classic-closed loop system.
Not all the time, according to:
http://www.fredmiranda.com/forum/topic/241524
Cheers,
David
ASAAR
>> Sure, but AF systems don't do that. They 'hunt' for the sharpest
>> reading on the AF sensor. It's a classic-closed loop system.
>
> Not all the time, according to:
> http://www.fredmiranda.com/forum/topic/241524
And not all of the time for some of Nikon's DSLRs, such as the
D300. There are custom settings for Single Servo AF (AF-S) and
Continuous Servo (AF-C), which allow the camera to take photos only
when focus has been achieved (focus priority, slower) or when the
shutter is fully depressed but hasn't necessarily achieved best
focus (release priority, faster). Continuous Servo mode has a third
option, release+focus priority, described in the manual with :
> Photos can be taken even when the camera is not in
> focus. In continuous mode, frame rate slows for
> improved focus if the subject is dark or low contrast.
Marco Tedaldi
> David J Taylor wrote:
>> Marco Tedaldi wrote:
>> []
>>> BTW: Is phase-shift autofocussing open loop?
>>
>> I think the answer is "it depends".
> If the lens is set to MF, it's open loop.
>
Here I disagree (I think the rest has well been explained in the link in
another post) but Manual Focus is definitely "Closed Loop".
There is no control by the camera at all, but by you. So how does manual
focussing work?
I look through the viefinder and see that the image is not in focus. My
hand moves the focus Ring which changes the focus of the lens, te
resulting image is looked at by me and a decide to focus further or stop
focussing. That's a typical closed loop control.
If you leave the human out it's neither open nor closed loop, it's no
control at all. :-)
Marco
Chris Malcolm
> In article <49e133bb$1...@dnews.tpgi.com.au>, Bob Larter
> <bobby...@gmail.com> wrote:
>> >> I'd bet money that it's closed-loop - at least while the shutter's
>> >> half-pressed.
>> >> It should actually be pretty easy to test by focusing on something
>> >> nearby & moving the camera back & forth with the shutter half-pressed.
>> >> <tries it out>
>> >> Okay, what actually happens on my EOS 1Dmk2 in single-shot focus is that
>> >> lens definitely hunts while it's aquiring focus, but as soon as it does
>> >> (ie; when the focus confirm light comes on), it gives up & leaves the
>> >> focus there.
>> >
>> > <http://www.fredmiranda.com/forum/topic/241524>
>>
>> Interesting.
I found this quote particualrly interesting:
"When the camera determines how far and in what direction the lens
must move to cancel the phase difference, it does so within a
tolerance of "within the depth of focus" of lenses slower than f2.8
(down to f5.6) or "within 1/3 of the depth of focus" of lenses f2.8
and faster. The depth of focus is the range at the sensor plane within
which the image of a point will be reproduced as a blur smaller than
the manufacturer's designated "circle of confusion" (CoC). Canon's
designated circle of confusion is 0.035mm for the 24x36mm format and
0.02mm for the APS-C format. The CoC is based on maintaining the
appearance of sharpness in a 6x9 inch print at about an 10 inch
viewing distance (as revealed by the Euro-Canon web site). There is no
guarantee that images enlarged any greater than this will appear
sharp."
If that really reflects Canon's current AF policy then I'm glad my
Sony DSLR's AF is capable of significantly higher focus precision than
that! (I've just been testing and adjusting it, because it suffered
from back focus with some very critical lenses. In fact the focus
offset could be observed with all lenses at all apertures, but with
most it didn't count as back focus because it was well within the DoF.)
I suspect later Canons and/or later Canon lenses are more
sophisticated than the description in that article, because that
article also claims the camera calculates the acceptable DoF from the
maximum aperture of the lens, whereas at least some Canon lenses
specifically report to the camera what the acceptable CoC for this
lens is, and a service adjustment to the lens is capable of altering
that. Otherwise the camera might find it hard to focus a soft focus
lens :-)
> In one "look," it calculates the distance and direction the lens must
> be moved to cancel the phase differences. It then commands the lens to
> move the appropriate distance and direction and stops. It does not
> "hunt" for a best focus, nor does it take a second look after the lens
> has moved (it is an "open loop" system).
A point not mentioned in that article (which is about Canon's AF) is
that the focussing motors are stepper motors, and the system is so
designed that one step is inside the most critical DoF of that
lens. No higher focussing resolution can be got than that.
The article also remarks:
"One person on this forum has reported an e-mail from Sigma stating
that Canon cameras will tend to stop at the first part of the depth of
focus range they reach, depending on the direction the lens must move
to get there. My results have not verified that, and I don't think it
really corresponds to how Canon says the system works."
That effect will be lens specific, and will depend on whether you're
working with an aperture and lens where there are a few focus motor
steps within the DoF, and may also depend on how critical the lens
tells the AF it's supposed to be. On my Sony A350 the effect isn't as
marked as stopping at the boundary of the acceptable DoF, but the
position of the AF with respect to the DoF range does depend on which
direction it's approached from. The slight difference is irrelevant
and hard to notice unless working with extremely thin sharp DoF such
as with a 50mm lens at f1.4.
It does seem to be the case that some lenses have a reputation for
being soft in focus which is not a property of the optics but that
their AF system on at least some camera bodies isn't up to the
precision of the optics.
--
Chris Malcolm
Chris Malcolm
One exception trumps a generalisation :-)
The original poster claimed a general property of DSLR phase detection
focus. It seems from the discussion so far that at least some Sony
and Canon DSLRs are more sophisticated than that.
--
Chris Malcolm
David J Taylor
[]
> One exception trumps a generalisation :-)
>
> The original poster claimed a general property of DSLR phase detection
> focus. It seems from the discussion so far that at least some Sony
> and Canon DSLRs are more sophisticated than that.
I thought the OP asked, rather than claimed?
"BTW: Is phase-shift autofocussing open loop?"
Of course, that was when we got distracted from IS/VR and into auto-focus!
Cheers,
David
Paul B
> Bob Larter schrieb:
>> David J Taylor wrote:
>>> Marco Tedaldi wrote:
>>> []
>>>> BTW: Is phase-shift autofocussing open loop?
>>>
>>> I think the answer is "it depends".
>
>> If the lens is set to MF, it's open loop.
>>
> Here I disagree (I think the rest has well been explained in the link
> in another post) but Manual Focus is definitely "Closed Loop".
>
> There is no control by the camera at all, but by you. So how does
> manual focussing work?
> I look through the viefinder and see that the image is not in focus.
> My hand moves the focus Ring which changes the focus of the lens, te
> resulting image is looked at by me and a decide to focus further or
> stop focussing. That's a typical closed loop control.
>
> If you leave the human out it's neither open nor closed loop, it's no
> control at all. :-)
I'm not sure where the notion of an open loop crept in. The concept of an
open loop is like running up to a football, kicking it with the intention of
striking the back of a net. A servo system has some sort of feedback.
Depending on design the system maybe damped to stop overshoot by reducing
the bandwidth. If the loop cannot find a null, it effectively becomes
open-loop & a large offset applied to hunt across until the desired null is
found (& the loop closes) & if it isn't, will change sign or direction until
it does or is designed to give up rather than hunt again. Mechanical servos
are a lot more interesting than a purely electronic feedback circuits such
as a phased locked loops - the moving parts have inertia governed by their
mass.
Ken Hart
"Grimly Curmudgeon" <grimly...@REMOVEgmail.com> wrote in message
news:nov0u4d281ri2a33a...@4ax.com...
> We were somewhere around Barstow, on the edge of the desert, when the
> drugs began to take hold. I remember "Paul B" <m...@privacy.net> saying
> something like:
>
>>Does any agree that both Canon & Nikon should fit sensor based image
>>stabilisation to their DSLRs? Lens based stabilisation may well be better
>>but is rather dependent on the lens having it to start with! So why not
>>have
>>the option of either?
>
> It'd be nice to have the choice, I agree. I bought a K10D mainly because
> of the in-body IS and the ability to turn all my manual glass into IS
> lenses was a major attraction. However, if I buy a FF Canon or Nikon, I
> might be tempted to buy at least one IS lens for it.
> I have no particular bias either way - it's a pity that many such
> discussions on the merits of such technology get bogged down into an
> either/or argument by fanbois.
I like the after-market image stabilization best- you know, the devices with
three legs and a 1/4"-20 bolt that threads into the bottom of the camera. I
think they are called "tripods".
John McWilliams
It's hell for sports, though.
--
john mcwilliams
David J Taylor
[]
> I like the after-market image stabilization best- you know, the
> devices with three legs and a 1/4"-20 bolt that threads into the
> bottom of the camera. I think they are called "tripods".
.. but somewhat inconvenient when shooting from a moving boat, aircraft,
or while walking along, and they may not be allowed in certain museums
etc.
David
Ken Hart
"John McWilliams" <jp...@comcast.net> wrote in message
news:gru13i$ft7$1...@news.motzarella.org...
> Ken Hart wrote:
snip>>
>> I like the after-market image stabilization best- you know, the devices
>> with three legs and a 1/4"-20 bolt that threads into the bottom of the
>> camera. I think they are called "tripods".
>
> It's hell for sports, though.
>
Maybe, maybe not... I'd say a tripod is only "heck" for sports. When I shot
sports, I prefocused on a particular spot, and waited for the action to come
to that spot.
While I didn't use a tripod because of the bulk, I could envision using a
monopod.
John McWilliams
Mais oui. A mono for sports is just what the arm doctor orders. I
frequently use same, but the types of sports I shoot doesn't much allow
prefocussing or choosing a single lane.
--
John McWilliams