[reeftool]

Panning is mentioned in my K10D manual is an instance where SR should be turned off or won't be effective. Page 67. They don't call it panning but it mentions "shooting a moving subject". That's panning to me. I can't really remember an instance where I could blame a bad panning shot on the SR. Most of mine get deleted regardless of the camera setting. I usually forget and SR is on all the time. When you are panning, you are moving the camera. The SR probably isn't locking on anyhow. Shooting with a long lens takes practice. SR helps but only to a point. Panning is even tougher. If I want good shots with a long lens, I use a tripod or at least a monopod. I have had some real nice shots hand held at 300mm and up. I have had hundreds of deletes. Some people are better and more experienced hand holding and panning than others. The next time you go to a ball game, count how many pros in the press box DON'T use a tripod or monopod.

[falconeye]

When you have SR off or on it uses data from the Hall Sensors to keep the platter square. These same Hall Sensors will read when the body of the camera applies any amount of force that may lag the sensor platter. This is what makes it work while panning with SR off.
...
Remember the sensor platter is rolling on ball bearings.

The Pentax SR has two control loops:
1. an outer control loop reading the gyros which can be switched off.
2. an inner control loop forcing the sensor to shift into the exact position as required by the outer control loop. It can't be switched off as otherwise, gravity alone would make the sensor "fall down".

This second, inner control loop is using hall sensors. Also, the sensor isn't on ball bearings. By all means, the inner details of the second (inner) control loop should not be mentioned in a discussion about SR. It doesn't help to understand things. I.e., it is counterproductive in order to understand what happens during panning. The entire inner control loop is best understood by thinking about a motor which can accurately move the sensor to any position the (outer) SR control loop determines. The only exception to my rule was when we tried to determine possible causes of excess shutter-induced blur with the K-7. That's not our topic here. From now on in this thread, I will not mention the inner control loop, the hall sensors, the linear motors etc. anymore.

Because the SR gyros measure angular velocities (not accelerations!), the firmware theoretically can determine the camera to be panning. However, I am unsure if it does. AFAIK, nobody studied this yet.

In LV and when rotating the camera, I believe to see that the image moves to the sensor border and then stays there. To "crawl back" to neutral if rotatation stops. The sensor being driven to the edge is a perfect way to "support" panning as it leaves SR with only one degree if freedom. So, in video and LV, I would say that panning and SR do nicely cooperate.

In normal mode, I am not sure though. The SR motor has enough power and time to compensate for the panning rotation in a short enough exposure. One could measure this with a motorized panning head programmed to follow a known target like a race car on a track. It may then be that SR causes more blur on the car and less blur on the track.

Alternatively, it may be that the firmware detects panning. In doing a pano, I repeatedly turn the camera, stop, take a photo, turn again, etc. It is in this mode that leaving SR on can give me blurred photos. If I pull the trigger each time immediately after turning what I call "doing a power pano" (doing a pano in the most rapid possible succession to "freeze" the scene as much as possible). Whether the SR cannot cope with sudden changes of movement or only tries to detect a panning, I don't know ...

Anyway, what I tried to say above: when doing a power pano, switch SR off.

[Lowell Goudge]

But a steady angular velocity (as in a steady pan) COULD be filtered out, and then SR would work wile panning as well. Pan with live view on and you will see (I hope, as I don't have the camera with me to try it myself) that the sensor plate isn't bouncing into the wall again and again.

no disagreement. I think the problem with panning is only really that very few shots have no angular acceleration, even if panning was "smooth" i.e. no jitters as such. My experience is things are generally coming at you at about 45 degrees, and there is real acceleration as you pan. I have some shots of diving terns somewhere, that show this.image is sharp except in the direction of panning, but still sharper overall than without SR

[falconeye]

My experience is things are generally coming at you at about 45 degrees, and there is real acceleration as you pan.

An ideal system would have a kind of race track algorithm: approaching subject and angular acceleration with max. velocity at shortest distance. It could actually couple SR gyros with AF tracking, both systems helping each other out. The gyros could even help the AF points track the subject which is a problem with long FLs. Another innovation made by pentaxforums.com

[Gimbal]

Also, the sensor isn't on ball bearings.

Actually several sites states it is. I also have a vague memory of a picture showing it, I can't find it know though. But the balls are what holds the sensor plate in position between the front and back plates.

[Gimbal]

Because the SR gyros measure angular velocities (not accelerations!), the firmware theoretically can determine the camera to be panning. However, I am unsure if it does. AFAIK, nobody studied this yet.

I agree that the SR are using gyros to detect shake, but the camera (at least the latest models) also has accelerometers pointing in all tree dimensions. (Evident by the horizon and elevation angle measuring capabilities), at first I figured they had improved the SR with these new sensors as well, tracking lateral movements. But after a few simple tests it seems they are not used for this at all. Or maybe my tests where to simple. But the hardware is there, only the software needs polishing, but that will probably not happen for the current models.

[falconeye]

Actually several sites states it is. I also have a wage memory of a picture showing it, I can't find it know though. But the balls are what holds the sensor plate in position between the front and back plates.

I checked my own K-5's cut model photography and I can see something which could be ball bearing balls. So, you may be correct. I apologize.

I believe to remember that the patents describe a magnetic fixture, but of course, Pentax may have implemented this differently. However, I maintain my claim that the hall sensors don't help in understanding the fundamental workings of the SR system. E.g., if the SR shift motor would be a kind of step motor, there wouldn't be hall sensors.

[derekkite]

The outer control loop would be in two parts; the inputs from the gyros etc. which would give an indication of movement along all axis'.

Then the algorithms that translate the input into an output to move the sensor.

The focal length that is either read or entered on startup is one factor taken into consideration. I would suspect that the algorithm looks for movements that would correspond to the shaking and movements of our hands, maybe something close to an oscillation. Our heart beats, our breathing, the shaking of muscle fatigue would show up as recognizable patterns. Those waveforms could be superimposed on top of an accelerating panning motion which would be ignored by the algorthm all the while adapting to the oscillations on top. I'm speculating of course since we can't see the source code. The challenge is not responding to movement, that is a simple input-output response. The challenge is selecting which movement to respond to. I suspect the algorithms are constantly improved from model to model taking advantage of increased processing capacity.

A friend who shoots with a Nikon D7000 with a 600mm lens has a slight parkinsonian tremor in his hands. He gets stunning shots. The implementation differs; an element in the lens is adjusted, but similar input processing would be occurring that controls the output. Does anyone have a similar experience with a Pentax?

[jamesm007]

No arguments at this time from me. Your power panning is a situation where SR would not work. When I track a bird, quick, slow, up/down, around as fast as I can, but as smooth as I can with no sudden jerky movements SR will be fine. Which means panning smooth yet fast. Some people may need to turn off SR. It will depend on their results. However the Patents show how the SR system is able to work during panning. It has a high pass filter. This fully explains why some people like me do benefit from having SR on during panning and some don't. I wonder if live-view gives a good indicator of how much you can move or how hard you can move, before the pic starts getting really jumpy?

Quotes from Pentax SR literature and Patents. Pentax filed over 30 patents for its original free floating sensor Shake Reduction (SR) system. The system does in fact work during panning. Its official. But we are both right. You may need to shut it off if your moving the camera more than by hand. But other wise it works because the system is fast, and ignores panning movements.

To rap it up. Camera shake is just that shake, movement it seems is a none factor unless it becomes too aggressive. Probably why some users as me enjoy a SR benefit even panning and some users don't.

So I guess we are both right, no?

http://www.docstoc.com/docs/73689976/Shake-Reduction-Technology

Does the SR system work during pan shots of moving subjects?

Yes, it does work, but switching the SR function off is recommended In shooting conditions

where there may be camera shake more than by hand, the SR system may not appropriately

compensate for the shaking.

Additional Advantages of the PENTAX SR System

•
Since the SR system is activated only when you press the shutter release, any effect on

battery consumption is negligible.

• PENTAX SR is optional. You can leave SR on permanently, turning it off only when

using a tripod or panning (deliberately moving the camera in the direction of subject

motion to blur the background).

• Performance capabilities such as auto-focus speed, shutter lag time and continuous

shooting rate are unaffected by the SR system because its operation is instantaneous,

occurring within the normal exposure interval.

• By building the shake-reduction system into the body, the Pentax SR system provides

maximum flexibility and requires no compromises in optical quality.

• To provide optimal shake reduction, the camera must “know” the focal length of the lens

in use. Pentax F, FA, D-FA, DA*, and DA series lenses automatically relay focal length

information to the camera. With older lenses, the bodies allow users to manually input

focal length information via the Shake Reduction menu which allows focal lengths all the

way from 8mm to 800mm.

Patents

Pentax Shake Reduction Patent 20080226276

>0048]Thefirst high-pass filter circuit 27a reduces the low-frequency component of thesignal
output from the first angular velocity sensor 26a, because the low-frequencycomponent of the
signal output from the first angular velocity sensor 26a includes signalelements that are based on
null voltage and panning motion, neither of which are related to hand-shake.





[0049]Similarly,the second high-pass filter circuit 27b reduces the low-frequency component ofthe
signal output from the second angular velocity sensor 26b, because thelow-frequency component of
the signal output from the second angular velocity sensor 26b includes signalelements that are
based on null voltage and panning motion, neither of which are related tohand-shake.





>[0050]Likewise, the third high-passfilter circuit 27c reduces the low-frequency component of the signal output from the third angular velocity sensor 26c,because the low-frequency component of the signal outputfrom the third angular velocity sensor 26c includes signal elements that arebased on null voltage andpanning motion, neither of which are related to hand-shake.