Lens with IR and K-x with internal SR

joodiespost · 12-25-2009, 03:58 PM

If both are used in combination, is one [most likely lens] turned off? Or does both work in conjunction for Super Dooper SR?

EliotK · 12-25-2009, 04:07 PM

I asked the same question in a different forum and got a very good answer.
There is a video link explaining it for a different camera, see below.

http://www.youtube.com/watch?v=KPdy52mR6Io

1. In general, the two IS systems are not compatible and make it worse when both are turned on.
2. The in-lens system is better so turn off the in-camera one. (Multiple reviews on dpreview.com confirm that as well).
3. You can always experiment and find out for yourself if you have such a system and post your findings for the rest of us to learn.

4. Manufacturers should figure out a way to design them to compliment each other, not "fight" each other for improved image stabilization.

Good luck.

joodiespost · 12-25-2009, 04:55 PM

Ty ^^ ill check that out

**interested_observer** · 12-25-2009, 07:08 PM

Originally posted by EliotK

http://www.youtube.com/watch?v=KPdy52mR6Io

4. Manufacturers should figure out a way to design them to compliment each other, not "fight" each other for improved image stabilization.

The problem when you think about trying to couple both types of stabilization together is that it is pretty well doomed to failure. The video clip demonstrates this quite well. The user holding the camera is inducing the shake, that is sensed by both the lens stabilization system AND the body stabilization system.

The lens stabilization gathers the light and as it passes through the lens, adjusts for the shake as it focuses the now stabilized light on the camera body sensor.

The camera body sensor, being stabilized also, and sensing the same shake (via motion accelerometers) as the lens stabilization system, attempts to counter the shake by moving the camera sensor in the opposite direction as the sensed shake, essentially injecting the shake back in to the stabilized image from the lens.

joodiespost · 12-25-2009, 11:02 PM

yeah. I wait for the day when SR is so advanced that camera shake is a thing of the past. ^^

**interested_observer** · 12-26-2009, 04:00 AM

Well, to go beyond what we have today, you would need a cooperative coupling of image stabilization logic. In that sensing user induced movement of the camera body by accelerometers feeding both lens and body stabilization is self defeating, and in that there are two obvious opportunities to provide stabilization, then there should also be two different methods to observe or sense the user induced "shake".

So, if we assign the accelerometers to the lens, and the lens then focuses partially stabilized light on the sensor, that means the sensor would need to 1) sense movement from the light, while collecting the light for 2) the image. There is an inherent problem in this (my view), in that the sensor would have a difficult time determining if the movement was some left over movement the lens stabilization was not able to dampen out, or if the movement was from the object being photographed. For instance a child moving their hand or the photographer panning the camera to keep one object stable (tracking a car) while intentionally "moving" another object - the static objects behind the car - the scenery, etc.

That is a pretty difficult task to design a system to do, so I really doubt if we will see that any time soon. What I think we will see, is inbody stabilization taking a new input (if it does not do it already - since when you mount a manual lens you need to specify its focal length), and that is the mounted lenses' focal length, and provide a means to adjust to a greater degree for the longer lenses. There is a limit to this, and that is the absolute physical distance the sensor can move, within the camera's circuit board harness. In my view, the lens system has an advantage in that since it is closer to the aperture, its moving of a corrective lens has a greater effect on the light, with a smaller amount of absolute physical movement. That is just pure physics. The inbody system does a great job, and I am happy for it, and very willing to give up the small difference (residual shake - or very large amounts of shake, i.e., over driven shake) for the ability to apply shake reduction across any lens mounted on the body.

There are various systems in use today that do accomplish a better job in stabilization, however the amount of computational infrastructure just would not fit into a camera body. SAR (Synthetic Aperture Radar) or Doppler Imaging (weather) systems that use radar waves to image (as opposed to light waves) are examples. They require racks of computer equipment (and large amounts of power) and even with today's technology, that aint gonna fit into a camera body any time soon (and if it could - none of us could afford it, and it would be so hot the camera body would melt in our hands - even if we could carry the battery).

joodiespost · 12-26-2009, 11:22 AM

great post. thank you for the insight ^^ Yeah, I was thinking about the limitation of the SR movement within the camera as well. But as long as we are talking about minor shakes that comes from trying to hold a camera steady, the camera should have all the movement space it needs to counter act that. So it seems like the gist is that we'll need a stronger computer. How big is the computer in a segway for example? Cause isn't that in theory also doing it's "SR" by keeping the thing vertical?

**interested_observer** · 12-26-2009, 12:02 PM

You are close with the example of a Segway, however its still apples and oranges.

Going back to photography, you have to think about moment arm, because that is a similar problem that we have in the camera. Pick up a pencil and hold it at one end, trying to control the shake at the other end - pretty easy. Then think about picking up an 8 foot 2x4 and controlling the other extreme end - extremely difficult, as every movement is amplified a great deal. That is the problem that the in-camera SR has with focal length. Using a 20mm lens, there is not much shake since you are looking at a very wide area. Then consider mounting a 300mm lens and zooming in on an object across the street. A little camera shake induces a movement of say 6" in that object (up and down and all around). So the camera has to perform in both situations, and essentially it does very well. In terms of the 300mm lens shake, the 6" movement of the object translates to some small movement of the sensor (think of a titer totter, with one side 2 feet and the other side 100 feet). A 1 foot change on the 100 foot side, translates to a very small change on the 2 foot side.

On your Segway example, all the distances are fixed. The handle does not get any longer, (a different angle as the user leans, etc.). So yes it is similar, but not quite the same.

Its not really a matter of having a more powerful computer, its understanding the problem that you are trying to solve, the inputs, what they are and how they are being measured and at what rate, plus the way in which you are going to counteract them, the rate at which you are going to apply changes, along with the overall latency of the system.

joodiespost · 12-26-2009, 12:07 PM

ahh ic, ty again for the insight

GerryL · 12-27-2009, 03:39 PM

Originally posted by joodiespost

yeah. I wait for the day when SR is so advanced that camera shake is a thing of the past. ^^

When the sensor becomes so advanced that very high ISO settings are perfected, then slow shutter speeds will be a thing of the past (this is what causes blur), not unless you set it for effect.
Even SR would be a thing of the past when this happens.
Nikon now has a DSLR that can achieve 100,000 plus ISO.
I don't know how effective it is though or noise is not an issue.
The future is near.

12-25-2009, 03:58 PM	#1
joodiespost Veteran Member Join Date: Dec 2009 Posts: 336	Lens with IR and K-x with internal SR If both are used in combination, is one [most likely lens] turned off? Or does both work in conjunction for Super Dooper SR?

12-26-2009, 04:00 AM	#6
interested_observer Loyal Site Supporter Join Date: Jun 2009 Location: Tumbleweed, Arizona Photos: Gallery \| Albums Posts: 5,707	Well, to go beyond what we have today, you would need a cooperative coupling of image stabilization logic. In that sensing user induced movement of the camera body by accelerometers feeding both lens and body stabilization is self defeating, and in that there are two obvious opportunities to provide stabilization, then there should also be two different methods to observe or sense the user induced "shake". So, if we assign the accelerometers to the lens, and the lens then focuses partially stabilized light on the sensor, that means the sensor would need to 1) sense movement from the light, while collecting the light for 2) the image. There is an inherent problem in this (my view), in that the sensor would have a difficult time determining if the movement was some left over movement the lens stabilization was not able to dampen out, or if the movement was from the object being photographed. For instance a child moving their hand or the photographer panning the camera to keep one object stable (tracking a car) while intentionally "moving" another object - the static objects behind the car - the scenery, etc. That is a pretty difficult task to design a system to do, so I really doubt if we will see that any time soon. What I think we will see, is inbody stabilization taking a new input (if it does not do it already - since when you mount a manual lens you need to specify its focal length), and that is the mounted lenses' focal length, and provide a means to adjust to a greater degree for the longer lenses. There is a limit to this, and that is the absolute physical distance the sensor can move, within the camera's circuit board harness. In my view, the lens system has an advantage in that since it is closer to the aperture, its moving of a corrective lens has a greater effect on the light, with a smaller amount of absolute physical movement. That is just pure physics. The inbody system does a great job, and I am happy for it, and very willing to give up the small difference (residual shake - or very large amounts of shake, i.e., over driven shake) for the ability to apply shake reduction across any lens mounted on the body. There are various systems in use today that do accomplish a better job in stabilization, however the amount of computational infrastructure just would not fit into a camera body. SAR (Synthetic Aperture Radar) or Doppler Imaging (weather) systems that use radar waves to image (as opposed to light waves) are examples. They require racks of computer equipment (and large amounts of power) and even with today's technology, that aint gonna fit into a camera body any time soon (and if it could - none of us could afford it, and it would be so hot the camera body would melt in our hands - even if we could carry the battery). Last edited by interested_observer; 12-26-2009 at 09:17 AM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
People Internal Struggle	D4rknezz	Post Your Photos!	8	10-14-2010 07:32 PM
What internal camera settings do you use ?	lesmore49	Pentax DSLR Discussion	12	06-14-2010 06:38 PM
News Internal Search Improved	Adam	Site Suggestions and Help	8	11-20-2009 12:34 AM
internal halo?	gemini	Pentax SLR Lens Discussion	5	06-27-2009 11:56 AM
Internal dust	dropbear	Pentax SLR Lens Discussion	16	09-17-2007 11:13 AM

12-25-2009, 04:07 PM	#2
EliotK New Member Join Date: Dec 2009 Location: Miami Posts: 10	I asked the same question in a different forum and got a very good answer. There is a video link explaining it for a different camera, see below. http://www.youtube.com/watch?v=KPdy52mR6Io 1. In general, the two IS systems are not compatible and make it worse when both are turned on. 2. The in-lens system is better so turn off the in-camera one. (Multiple reviews on dpreview.com confirm that as well). 3. You can always experiment and find out for yourself if you have such a system and post your findings for the rest of us to learn. 4. Manufacturers should figure out a way to design them to compliment each other, not "fight" each other for improved image stabilization. Good luck.

12-25-2009, 04:55 PM	#3
joodiespost Veteran Member Join Date: Dec 2009 Posts: 336 Original Poster	Ty ^^ ill check that out

12-25-2009, 07:08 PM	#4
interested_observer Loyal Site Supporter Join Date: Jun 2009 Location: Tumbleweed, Arizona Photos: Gallery \| Albums Posts: 5,707	Originally posted by EliotK http://www.youtube.com/watch?v=KPdy52mR6Io 4. Manufacturers should figure out a way to design them to compliment each other, not "fight" each other for improved image stabilization. The problem when you think about trying to couple both types of stabilization together is that it is pretty well doomed to failure. The video clip demonstrates this quite well. The user holding the camera is inducing the shake, that is sensed by both the lens stabilization system AND the body stabilization system. The lens stabilization gathers the light and as it passes through the lens, adjusts for the shake as it focuses the now stabilized light on the camera body sensor. The camera body sensor, being stabilized also, and sensing the same shake (via motion accelerometers) as the lens stabilization system, attempts to counter the shake by moving the camera sensor in the opposite direction as the sensed shake, essentially injecting the shake back in to the stabilized image from the lens.

12-25-2009, 11:02 PM	#5
joodiespost Veteran Member Join Date: Dec 2009 Posts: 336 Original Poster	yeah. I wait for the day when SR is so advanced that camera shake is a thing of the past. ^^

12-26-2009, 11:22 AM	#7
joodiespost Veteran Member Join Date: Dec 2009 Posts: 336 Original Poster	great post. thank you for the insight ^^ Yeah, I was thinking about the limitation of the SR movement within the camera as well. But as long as we are talking about minor shakes that comes from trying to hold a camera steady, the camera should have all the movement space it needs to counter act that. So it seems like the gist is that we'll need a stronger computer. How big is the computer in a segway for example? Cause isn't that in theory also doing it's "SR" by keeping the thing vertical?

12-26-2009, 12:02 PM	#8
interested_observer Loyal Site Supporter Join Date: Jun 2009 Location: Tumbleweed, Arizona Photos: Gallery \| Albums Posts: 5,707	You are close with the example of a Segway, however its still apples and oranges. Going back to photography, you have to think about moment arm, because that is a similar problem that we have in the camera. Pick up a pencil and hold it at one end, trying to control the shake at the other end - pretty easy. Then think about picking up an 8 foot 2x4 and controlling the other extreme end - extremely difficult, as every movement is amplified a great deal. That is the problem that the in-camera SR has with focal length. Using a 20mm lens, there is not much shake since you are looking at a very wide area. Then consider mounting a 300mm lens and zooming in on an object across the street. A little camera shake induces a movement of say 6" in that object (up and down and all around). So the camera has to perform in both situations, and essentially it does very well. In terms of the 300mm lens shake, the 6" movement of the object translates to some small movement of the sensor (think of a titer totter, with one side 2 feet and the other side 100 feet). A 1 foot change on the 100 foot side, translates to a very small change on the 2 foot side. On your Segway example, all the distances are fixed. The handle does not get any longer, (a different angle as the user leans, etc.). So yes it is similar, but not quite the same. Its not really a matter of having a more powerful computer, its understanding the problem that you are trying to solve, the inputs, what they are and how they are being measured and at what rate, plus the way in which you are going to counteract them, the rate at which you are going to apply changes, along with the overall latency of the system.

12-26-2009, 12:07 PM	#9
joodiespost Veteran Member Join Date: Dec 2009 Posts: 336 Original Poster	ahh ic, ty again for the insight

12-27-2009, 03:39 PM	#10
GerryL Veteran Member Join Date: Jan 2009 Location: CA Posts: 2,731	Originally posted by joodiespost yeah. I wait for the day when SR is so advanced that camera shake is a thing of the past. ^^ When the sensor becomes so advanced that very high ISO settings are perfected, then slow shutter speeds will be a thing of the past (this is what causes blur), not unless you set it for effect. Even SR would be a thing of the past when this happens. Nikon now has a DSLR that can achieve 100,000 plus ISO. I don't know how effective it is though or noise is not an issue. The future is near.