PDA

View Full Version : A technical IS question..


snaarman
4th February 2009, 08:31 AM
I use my E510 with a mixture of modern digital lenses and older manual optics. I was pondering the way IS operates recently and I have come up with this question.

I am assuming the camera has a set of accelerometers and processing electronics that drive the sensor shift mechanism. It knows what lens you have (if its a zoom it knows what focal length you've selected) and it applies the IS correction as needed during the exposure. Thus angular movements of the camera are corrected by movements of the sensor. I figure the sensor is told to move further with longer lenses that with wide angles lenses. So far so good.

If you fit a legacy lens you can dial in the focal length yourself and its up to you to get it right.

Now the question: Does the camera also know about focus distance? It strikes me that the problem of camera shake gets worse as you get really close (as per 50mm macro for example) - so the IS correction should be greater. I mention this because I use IS with my Tamron 90 macro but it seems less effective at close quarters.

Answers on a postcard please..

Pete

photo_owl
4th February 2009, 09:11 AM
Pete

I think the answer(s) are

No it doesn't and
No it doesn't

In the first case take the 50/2 which, as most is fbw, and whilst you have relative information passing back and forwards down the wires you don't have any absolute information as the physical control is open ended (you can turn and turn and turn) and not linear (faster you turn the greater the electronic gearing).

Second, I don't think so. Whilst it's certainly the case that you can observe the impact more easily the actual correction to be applied must be the same ie the right amount.

Neither of these repsonses are a comment on how it actually works though.

snaarman
4th February 2009, 09:20 AM
Pete

I think the answer(s) are

No it doesn't and
No it doesn't

In the first case take the 50/2 which, as most is fbw, and whilst you have relative information passing back and forwards down the wires you don't have any absolute information as the physical control is open ended (you can turn and turn and turn) and not linear (faster you turn the greater the electronic gearing).

Second, I don't think so. Whilst it's certainly the case that you can observe the impact more easily the actual correction to be applied must be the same ie the right amount.

Neither of these repsonses are a comment on how it actually works though.

Hmm, I wonder though.. I assume the fly by wire focus ring is a rotary encoder. It sends direction and velocity to the cpu and the cpu returns direction and step number to the focus motor. So - as you say - they "appear" to be connected but in a non linear manner.

However it can do a reset to infinity on power off, so I wonder if it can actually keep track of absolute focus motor position and therefore could infer some kind of value for focus distance... Come to think of it, and this would be a give away, does focus distance get recorded in EXIF?

Pete

photo_owl
4th February 2009, 10:43 AM
most exif viewers look for a record relating to focus distance, and the Oly cameras do fill that field - but the data is clearly rubbish when analysed.

if you are postulating that it would be possible for the camera to record then I maintain the answer is no - it won't know when the ring is overshooting either way, but still clicking/spinning away ie at min and max range the messages will still be created and sent but the motors won't be able to deliver.

snaarman
4th February 2009, 11:03 AM
Ok, so maybe the whole AF thing is optical phase detect and relative not absolute motion, within a standard feedback loop.

I guess I was wondering if you use IS with your (for example 90mm) legacy lens - maybe you get better IS results if you set it as a 200mm lens when operating really close, to offset the effect of greater magnification.??

Difficult to test I know, but just a point...

Pete

Ian
4th February 2009, 11:51 AM
I'm pretty sure that focus distance is recorded from the lens. It had nothing to do with the fly by wire nature of the manual focus contol. The position of the focusing elements will be knownand so focus distance determined, if for nothing else, for flash power output control. I don't know if the focus distance would be used in the IS calculations - a good question! I will try to get definitive answers...

Ian

Gwyver
4th February 2009, 04:35 PM
...

Now the question: Does the camera also know about focus distance?
Answers on a postcard please..

Pete

Pete,
The focus distance is known to the camera - from the AF detectors. It is recorded in the EXIF - in the Maker Notes section with the Tag name "FocusDistance" and is in units = metres. There is also another EXIF Tag called "FocusStepCount" which I assume to be the target value that is being used to drive the AF stepper motor in the lens.
The EXIF also contains fields recording (actual) FocalLength and ZoomStepCount - so the IS controller has all the necessary parameters.

Many folk state that the recorded EXIF FocusDistance value is wildly inaccurate but, and I really don't want to start any disputes, this may be because the AF didn't actually focus on the exact point that the operator had intended.