Let's talk 'Shutter Shock'...

Started Nov 20, 2012 | Discussions thread
Serguei Palto Regular Member • Posts: 202
Re: Let's talk 'Shutter Shock'...

Anders W wrote:

Serguei Palto wrote:

Anders W wrote:

Serguei Palto wrote:

Anders W wrote:

Serguei Palto wrote:

Anders W wrote:

Serguei Palto wrote:

Anders W wrote:

Serguei Palto wrote:

Anders W wrote:

As to your idea about sensor shake: What about my experiment on the tiled concrete floor of my kitchen? Wouldn't you say that this rules out (or makes less credible) the idea of sensor shake being responsible for the blur? And can you think of other/better tests we might do to establish whether sensor-shake is at all involved?

And did you read the following report on the most meticulous shutter-shock investigation I am aware of, and which takes a look at vibration as well as sensor shake?


I'd be interested in any comments you might have on it.

Thanks for the references.

About your experiment.

If I understand correctly, when you have placed the camera onto the concrete floor the blur has disappeared. If so, then it demonstrates that suppressing the camera movement does reduce the "shutter-shock" effect. In this case your observations are coherent with those reported by LumoLabs in their article you advised me to look at. (http://www.falklumo.com/lumolabs/articles/k7shutter/index.html). It is an interesting article. I agree with many of their findings.

Summarizing my feelings and the observations I see the following basic reasons for the so-called "shutter shock" effect.

1. The movement of the shutter blades with a mass of about 1g for a distance of about 10mm (just approximately the height of the sensor) can do result in reactive motion of 1kg body for a distance L=(1g/1000g)*10mm=10um (this body shift just saves the position of the center of mass of the whole system if we neglect the gravitation and human hands influence). This shift is of a size of a few pixels, and such motion can definitely result in the blur if the time of the up-to-down blades movement is comparable with the exposure time (10 ms). Attaching an additional mass to the body or just fixing the body using a tripod should suppress this kind of the shock mechanism.

Because the camera is in the gravitation field, the effect, of course, should be different if one flips the camera for 180 deg (upside down reversal, as considered in the article).

This effect can be escaped if a shutter has a symmetric design with blades moving in opposite directions, so the position of the shutter center of mass is not changed. The effect can also be small in the case of non-symmetric shutter, if the time of the up-to-down blades movement is much less than the exposure time. The last is one of the possible explanations of the fact, why I do not observe the "shutter shock" effect with Panasonic G3 body. I have the feeling that in the Pany body the shutter is faster than in the Oly one.

Much of that makes sense. But I am not sure I understand and agree with your claim that a faster shutter blade movement would reduce the effect or that the speed of this movement would be higher on the G3 than on the E-M5.

First, the E-M5 can flash sync at up to 1/250 s whereas the corresponding limit on the G3 is 1/160 s. Since max flash sync speed is directly related to the speed at which the shutter can open and close, this would suggest that the E-M5 shutter-blade movement is faster, not slower, than on the G3.

Second, and more generally, I don't see why the duration of the shutter-blade movement should bear any direct relationship to the duration of the camera movement that eventually causes the blur. What I do see is that the duration of the camera shake relative to the duration of the exposure determines how bad the blur will be and explains why the shutter shock is worst about 1/100 s and less prominent at longer as well as shorter exposures. At longer exposures, the camera movement will affect a smaller and smaller part of the total exposure. At shorter exposures, only part of the camera movement will have the time to register.

2. I think the EM5 IS system has a low frequency band and rather long response time. In general, if the body motion is close to the resonance the IS feedback can even become positive (destabilizing) instead of negative one. Fortunately, in case of EM5 the IS makes neither good nor bad job for the "shock".

Yes, that's pretty much in line with my own thinking.

3. When after the phase of motion the shutter blades suddenly stop, the sound "wave package" appears and propagates across the body. Not all the frequencies can propagate. Some of them are effectively damped (depends on body design and materials used). As I already said, if the characteristic frequencies of the body, or sensor holder, or suspended sensor itself are coincide with some of the propagating wave components of the "wave package" then the vibrations of the sensor appear. From the observations I can guess that one of the characteristic circle frequencies (w) is directly related with the speeds of the shutter providing the "shock" effect. So it is in a range of 80 Hz <w <500 Hz (In terms of frequencies of oscillations f=w/(2Pi) (Pi=3.14) we have a range of about 10 Hz < f < 100 Hz). This is, in principle, a moderate frequency range for the properly designed anti-vibration systems. Thus it is not surprised that the Panasonic OS system can sometimes help as you have mentioned. Probably we have to be happy and wait for a new generation of IS system from Olympus.

The idea of a sound "wave package" is of course an interesting one. But I think it is pretty much ruled out by my test with the camera placed on a tiled concrete floor. The floor effectively prevents the camera from moving but wouldn't stop the "wave package". Nevertheless, there is no in this scenario.

As to the ability of the 14-45 OIS to reduce/eliminate the effect of the shutter shock, my tenative explanation is that the components that need to move in order to counteract the shock (camera movement), i.e., the OIS lens element or lens group, are in this case very small and light (smaller/lighter than in bigger/longer OIS lenses like the 45-200 or the 100-300 and smaller/lighter than the sensor assembly of the E-M5). This, in turn, makes them capable of moving sufficiently fast to deal with the shutter shock.

That is very simple. The body moves until the shutter blades stop (immediately as the blades stopped the body stops too; only the wave package generated by the blades can play the game after this). If the blades motion time is much less than the exposure time then the sensor is exposed basically in a stationary state (this is true only if we neglecting an influence of the wavepackage).

Well, that assumes that the camera movement is due to the acceleration rather than the decelaration of the shutter blades. If the latter is more important, as I would intuitively think, and as the investigation of the Pentax K-7 found (if I recall correctly), then things stand differently, right?

Note, in addition, that using the anti-shock delay on the E-M5 (i.e., introduce a delay between the shutter closing prior to exposure and opening again for exposure) has been found to significantly reduce the blur (though not eliminate it entirely in my testing). Again, that does not square so well with the idea that the body stops moving immediately as the blades stop, does it?

In case of G3 the shutter is asymmetric (if I am not mistaken?). Thus, when I am shooting from hands, the body must move (otherwise, it would be violation of one of the principal lows of physics). Also in the case of G3 the sensor is rigidly attached to the body, so it also must move simultaneously with the body ("simultaneously..." is an approximation supported by a rather long characteristic time we are speaking on). Thus the sensor can not be exposed in the stationary state if the exposure time is comparable with the blades movement time. The blur must be observed in this case. The fact that I do not see the blur forces me to think that the G3 shutter is fast.

In case of EM5 the situation is different. The sensor is suspended, and it is not obliged to move with the body. Similar to a mass on a spring, it has a characteristic response time. If this time is just 10 ms then if EM5 shutter is very fast, the movement of the sensor will be delayed with respect to the movement of the body - the sensors will be in motion even after the blades and body stopped. In this case the blur is inevitable. By the way, the long characteristic time of the suspended sensor does not contradict to your experiments and probably it is the main reason for the "shock" effect.

As far as wave package is concerned you are right in a first approximation (if we assume that your tiled concrete floor does not influence the wave package propagation, then your experiments tell that the influence of the wavepackage is negligible). In general, any attached mass can make an influence the waves propagation.

I agree that the OIS system can have lighter moving parts, so their characteristic times are shorter and working frequency band of OIS wider. I think you are right.

If the center of mass of the blades is stopped then the center of mass of the rest of the camera is stopped too. This is the condition for the center of mass of the whole system to be not changed despite of any events those are happening inside the system. Only forces those are external with respect to the whole camera (blades+body+lenses) can change the position of the center of mass. The blades belong to the whole camera system and they are driven by internal forces of the system, so the center of mass of the rest part of system without the blades can not move after the blades stopped. If the blades accelerate the rest of the system accelerates too but in opposite direction. If the blades decelerate then the rest decelerates too (the speed just decreases and becomes zero when the blades stopped).

Unfortunately, in the article you mentioned this process is not clearly described.

After the blades have stopped the vibrations of the parts the system consists of are still possible with respect to the common center of mass that will remain fixed. For example, in the case of a suspended sensor, if its characteristic time is long (~10 ms) the sensor can continue its motion even after the blades were stopped (again the same scenario - the center of mass of the whole system is not changed, so the motion of the center of mass of the sensor is accompanied by contrary movement of the rest part). I hope this will help you to understand why the delay of about 1/8 s can be useful - it decreases the uncontrolled motion of the suspended sensor. You can imagine a pendulum in a car which accelaretes and decelerates to undestand what I mean.

Of course, all said above is valid only if we neglect the external forces. Gravitation forces, forces from the hands change the game in different extent. For example, the reversing the top and bottom of the camera changes the result because the body acceleration either coincides or opposite to the direction of the gravitation force. If you have attached an additional mass to the camera than you should add it to the whole system.

I understand that it works the way you describe in the first paragraph if we disregard external forces, i.e., think of the camera as being at rest in gravitation-free space and then fire the shutter. But for shutter shock on earth, that's not a very interesting case, is it? If the camera is suspended in a field of gravitation, e.g., in our hands or on a tripod, then we have reason to think that it works differently in some important ways, don't we? This is particularly so since any movement by the camera in its suspended state is likely to change the balance between the gravitational and normal forces acting upon it.

What was discussed above is OK not only for the free space. If the acceleration of a camera during the blades motion is significantly higher than G (and this was just discussed in LumoLabs article) then soft meat under the skin of hands can not resist it. The gravitation can be important if the body acceleration is comparable with G. One can check the last by measuring the "shock" after shooting from hands for the two camera positions: normal and reversed (flipping the top and bottom sides of the camera).

Yes, the camera might still move. But my point is that the impact of gravity is likely to change what you say about timing as well as the final camera position. In the case without gravity, the camera movement stops when the shutter blades stop and the camera will come to rest in a slightly different position than the original one (due to the slight change of the center of mass when the shutter blades end up in a different position than they originally had). Neither of these things is likely to be true with gravity involved.

Consider what will happen as the shutter blades move with the camera in a field of gravity. Suppose the shutter blades start by moving downwards to close the shutter prior to exposure. The acceleration of the blades may cause the body to accelerate upwards. However, the upward movement may in this case, in contrast to the case without gravity, stop before the blades come to a halt, due to the force of gravity acting in the opposite direction. When the shutter blades finally come to a halt, this will cause downward acceleration resulting in downward movement of the camera after the blades have stopped moving. Due to the increased normal force this downward movement generates, the camera will finally accelerate and move upwards again and come to rest in the position it originally had.

Yes, it seems as we are almost coherent in that what provides the "shock" effect. Of course, you are right - when the blades move down, the rest of the body will move upwards, while the gravitation field provides the force acting onto the center of mass downwards. Actually, if the acceleration of the rest of the body (during the blades movement) is comparable with the gravitation acceleration (G) then the gravitation decreases the motion of the body to the upwards, so the shock effect is suppressed. But, if the body acceleration is much higher than G then the gravitation is not more important and can be neglected for the time of the blades motion.

I don't think that the impact of gravity can be ignored here. It will suppress the upward movement. This however, doesn't mean that there will be no upward movement at all. Merely less.

There is only one point that is a little bit worried me in your response. I am not quite sure that your understanding of the process is perfect. It seems that you are under an impression of the LumoLabs interpretation of the blades halting process that is actually not clearly described therein and, to my mind, is misleading in the part, where they consider the blades halting. When the blades halt, the rest of the body stops. Despite during this process we do have high deceleration of the blades there is no significant changes in the center of mass of the rest of the body, because the center mass of the blades during the halting is not changed significantly (what we get from this process is just strong internal deformations and sound wave package). The principal game influencing the "shock" effect takes place when the blades are moving on a long distance equal to a height of the sensor. That is, to my mind, a crucial point for understanding of what is happening. Of course, herein we assume that the sound wave package is less significant compared to the reactive movement in providing the "shock" effect.

What I think you miss here is the fact that even if the shock that occurs when the shutter blades come to a halt is completely inelastic (as it has to be or the blades would bounce back), the laws of physics imply that the momentum is always preserved, which means the camera will move downwards after that shock, if (as my reasoning above assumes), the upward movement has already been stopped by the impact of gravity. In addition to that, gravity itself will move the camera downward as soon as the upward movement has stopped due to the imbalance between gravitational and normal force caused by the upward movement.

Yes, " ...the laws of physics imply that the momentum is always preserved ..." Namely because of this , when the shutter blades stop the rest of the body stops too.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow