[nanok]

i am using a k10d (well, gx10, same thing, really), i am wondering if anybody is aware how the shake reduction is supposed to behave for cases where one follows action with the camera (so intentional, mostly sideways, camera movement, rather continous, but "you never know" ).

in other words, i am seeing a rather bad hit rate from my last rally shoot (like: i maybe have 3 pin-sharp, out of 3-4 hundred or so), and i must work on figuring out if 1. i am simply an idiot and do something very wrong 2. sr should not be used in such conditions (relates a lot to #1, i know) 3. the af on the k10d is really that bad (i honestly doubt that, especially with depth of field giving a hand, i would expect to get better results manual focusing, if the focus would be the issue)

all honest oppinions appreciated (note: no rants, please, if possible: i like the camera, and it's af, and for now i am planning to use it as it is, to the best of my abilities, i will not switch system, at least not for now, i just want to try and get the best i can from the k10d, and i think i am far from it at this point)

thank you all kindly

[Geekybiker]

From what I have read it auto detects panning on the k10d and adjusts accordingly. Can't offer any real experience though as I dont shoot many panning shots.

[Lowell Goudge]

I believe the shake reduction works largely on acceleration not speed or displacement of the camera, and as a result panning at a constant speed would be the same as no acceleration.

[Aegon]

Lowell, if that were true, then the next question is: can you, a human being, maintain constant acceleration with a margin for error less than the oscillating frequency of the shake reduction mechanism?

I'd think that you couldn't.

[Aegon]

The K100D can not distinguish between panning movement and regular shake. Panning the camera with the Shake Reduction on may cause the shake reduction system to overcompensate during the exposure, thus causing blur. PENTAX therefore recommends turning SR off during a panning shot.

link

*edit* There are many conflicting sources on this topic. I don't know for certain (my Pentax doesn't do SR), so I'll keep an open mind and await responses.

[Lowell Goudge]

Lowell, if that were true, then the next question is: can you, a human being, maintain constant acceleration with a margin for error less than the oscillating frequency of the shake reduction mechanism?

I'd think that you couldn't.

not acceleration speed!

acceleration is what the think keys on, constant speed is zero acceleration. if you don;t keep speed constant in the panning, SR will correct the non constant part

[flyer]

I've read somewhere (can't remember where) that the shake reduction ignores a "more or less" steady movement in one direction, so that's why there is no panning mode on the Pentax cameras. I always leave it on when I pan and it doesn't seem to hurt anything anyway, so just try and see the result you get.

[Douglas_of_Sweden]

I believe the shake reduction works largely on acceleration not speed or displacement of the camera, and as a result panning at a constant speed would be the same as no acceleration.

I believe Lowell are correct. I've used accelerometers in my profession, and today they are very accurate and can be made extremely small. I don't know how exactly they use them in the Pentax cameras, but straight forward using the acceleration would be easiest. A 3-D accelerometer show -1 g on the vertical axis and 0 g (or -9.81 and 0 m/s/s respectively) when standing still. If you mover horizontally it begins to show the strength of the acceleration on those axes. The problem would be if you manage to move with a very constant speed, which probably is your goal if you for example are following a car with the camera, and indeed many motor sport photographers are very good at this. Constant speed equals no acceleration! One of Newtons laws of motion I think. So what we do in our applications is to integrate the acceleration over time to get the speed. There are many problems with this, and how they can do this in real time inside a relatively small camera what we do with rather strong pc:s post processing our data I can't really even guess, other than I think they must do that somehow to make the shake reduction to work. Probably they have less problems with shake reductions due to relatively small distances. Similarly we can deal with the sea wave motions of a whole ship and vibrations when correcting the measurements of atmospheric turbulence, relatively "easily" if it was only this. However, when the ship is not on average still but moving (and they tend to do), an long long number of integrations of acceleration over time will add larger and larger errors. The equivalent to this would be you moving the camera with almost but not exactly constant speed to follow a car...in every time step in the processor, the camera will try to deal with the small accelerations caused by you deviating slightly from the constant speed as if yet another "shake"-acceleration component...it will not be easy. A system that successfully deals with the turbulence-like shaking from a man holding the camera still, could very well fail when you start to move the camera. It is not easy. It can probably be solved with different skill (as our corresponding problem can), but there is no simple trick around it, if I'm right in making the parallel to our problem. It would be interesting to know how well canikon and sony solve this problem.

Have you tried to turn of the shake reduction? Does it get worse or better?

PS. Why they use accelerometers? There is no simple way to measure speed. Whatever the application is, we actually measure something else, and scale it to speed. Very basic scientific problem. In the shake reduction they could have use also of knowing the inclination of the camera (tilt), and rotational speed. But that can all be calculated from an array of six accelerometers. This is interesting, I shall see if I can find some more info on this. Recently got the K20D myself so I want to understand how it works, rather than guessing as you must realizing I do above. DS

Sorry for the lecture.

[Lowell Goudge]

Douglas of Sweeden

do we really care about speed? I think it is acceleration that we want, and they apply corresponding acceleration to the sensor. If you hold put a really long lens on the camera and shoot at 1 second, you can actually hear the sensor hitting hard stop limits.

They apply an acceleration opposite to the lens motion, and scaled, I think to counter the shake

[CapitaineAbitibi]

From personal experience, I always got better results when the anti-shake was turned off when panning.

[nanok]

thank you all for your input, that's pretty much what i assumed also, but it's always good to have alternate oppinions. i do understand (roughly) how shake reduction works, however i cannot know how it is implemented on the k10d.

i think i will head to the highway before the next rally, and do some tests, see what i can figure out. my assumption right now is that the sr is okay, the af is at least okay enough (maybe it would start tobecome problematic at f2.8, but not at the apertures i can manage on my 50-200), but it's my panning that sucks (even though i am not much into motion-blur, panning is essential for such shots, unless you go well above 1/1000 shutter speed or so)

any thoughts (especially from motorsport shooters here) are still wellcome (what works for you, what doesn't, any hints). birders are wellcome too, i think it is even harder for them (tried it, found i was worthless at shooting birds in flight, even more so than i am at race cars)

ps: if i sort this out, i plan to soon have the "pro" shooters tucking their canons under their coats in the pooring rain go mad at me (delicately covering the eyepiece while the camera is dripping), i already had some "wtf is wrong with you, mate" looks. was all worth it, if only for that

[Lloydy]

Panned with the shake reduction on. Could be sharper?

K10 with 18 55 kit lens.

[Douglas_of_Sweden]

From personal experience, I always got better results when the anti-shake was turned off when panning.

Kind of what I suspected.

[Douglas_of_Sweden]

Douglas of Sweeden

do we really care about speed? I think it is acceleration that we want, and they apply corresponding acceleration to the sensor. If you hold put a really long lens on the camera and shoot at 1 second, you can actually hear the sensor hitting hard stop limits.

They apply an acceleration opposite to the lens motion, and scaled, I think to counter the shake

Well, I might be wrong, which would not be strange considering that I took a giant jump between rather different application with the accelerometers being the common feature, trying to see why an almost constant speed could be a problem.

It would probably be much easier if they could work directly with the acceleration itself, but how do you apply an acceleration? There must be some sort of little "engines" involved in this, right? Do you know any details of that? Can you explain to us? Acceleration is basically equivalent to force...

[nanok]

douglas:

thanks for the lecture

i really do not think a shake reduction system bothers to integrate ot get speed, it probably works more on frequencies, and has a lot more fun with statistics than we would imagine (trying to be "predictive" about it. on how panning myght be detected.. you might be dissapointed: i am pretty sure the key-word here is "thresholds", probably on frequencies, and in the case of canikon, i think it is just "ignore the horizontal component", at least roughly (but i think some newer systems of theirs also have an automatic way of deciding, so i guess they might use thresholds also)

another related hint, which might not be as obvious to most people: integration is actually a very "natural" process, which is "numerized" to be more universal. when you do it digitally, you have the advantage of flexibility, but it is very computationaly expensive. if you know exactly what you want to do, you might be able to design something much cheaper from a computational standpoint (maybe even analogue, in some "extreme" cases), instead of using a generic digital cpu to do it "soft".

imagine this rather rough example: you have a "simple" curve. it is "real", so you need a function to describe it (which is he first step, and headache), and only than you can integrate. or you can just "walk" it with a device, to measure distances on it. integration is in theory capabale of arbitrary precision, given enough computing power, however reality rarely is that precise itslef, and i am sure you know what i mean , otoh walking it means to have a (simple, but still dedicated) device just for it, which won't be much use for anything else, but sometimes it can work better. otoh, if you start out with "real world", you will always need a device in the real world, a way to digitze it's output (adc) a way to reverse the output from the computing task (dac) and another device to _do_ stuff in the real world, as a result of all this work. as things get complicated, you can save more significant amounts of computing power (which starts to matter in real time applications), but you start spending (exponentialy, perhaps) more on design (and software will always be cheaper, and these days "digital" hardware is hard to beat on price, with all the r&d already pumped into it). when size also matters, things get more complicated. but i am sure you know more about it than i do

(just some food for thought, i am no expert, unlike you, obviously). btw, that's some really cool job you seem to have