image stabilization versus AUTO

I just got an Olympus FE-190, which I like a lot. It has digital image
stabilization and, in principle, I understand how that works. But in
shooting modes on the camera, one can choose [AUTO] "for normal
shooting" or [(picture of a waving hand)] to turn on image
stabilization. Now, simple question. Why wouldn't I *always* want that
image stabilization on? What does [AUTO] buy me that the waving hand
doesn't? Or, conversely, what am I compromising by turning on image
stabilization? I gather it must be something. The manual seems to have
nothing to say on this matter.

Thanks for any knowledge you can provide.

On the contrary, IS on the FE-190 does a great job on canceling camera
motion.

So if I'm dancing, IS goes on, but ....

Agreed. My Nikon Coolpix 8800 IS can compensate for both horizontal and
vertical movement. However, the manual says to turn off IS when you use
a tripod. I sometimes forget and haven't noticed any difference.

--
- Barry as TDC Sorcerer
- Magical Manager of the Mysteriously Missing Main Street Magic Shop
- Curator: The Disney Extinct Attractions Graveyard
- <http://www.flickr.com/groups/disney_graveyard/>
- Flickr photostream: http://www.flickr.com/photos/barrywallis/
- DLR Pictures: http://members.cox.net/dl.album

Image stabilisation systems usually use so called "gyro" sensors for the
detection of angular camera movement, the ones used in digicams have good
short term stability but tend to have a drift over longer intervals. It can
well happen during long time exposures (like over 10 seconds) that a slowly
drifting sensor tries to compensate for a camera motion that is not there
because the camera is mounted on a tripod.

"DougL" <dla...@hotmail.com> schrieb im Newsbeitrag
news:1166582176.4...@f1g2000cwa.googlegroups.com...

I believe that's correct, and it begs the question. Why would I want to
turn it off?

The above posting from Cgiorgio explains exactly why and when you would want
to turn IS off.

Roy G

If the stabilization system is in the optics, it uses a lot of
electrical power, and runs the battery down faster. I suspect even the
"electronic" stabilization may use quite a bit of power. So, if you
don't need it, you might want to turn off stabilization to make battery
charge go further.

Thanks. The frame shifting required for stabilization, as well as gyro
activation may well use
substantial power. But why wouldn't they just say that in the manual?

It's clear why, on DV cameras, you'd want IS off if you were using a
tripod, in that the stabilization
would fight any slow pans you'd want to do. But that doesn't apply to
single frame cameras.

I also am led to believe from various web documents that single frame
cameras boost up the
CCD sensitivity in order to use a shorter exposure time at given
aperture stop when IS is turned on. Now, I don't know how they do that,
but I've seen those words! If there is some extra amplification stage
(whether analog or digital) that is being applied, then one might
compromise the levels and increase noise by doing so.

You would think there would be some rational explanation from the
manufacturers about this.

Some IS systems allow you to pan.

David

I believe that he means that compensation for non-existent movement
will *create* blur.

But beyond that, if I understand it correctly, sensor IS effectively
uses a smaller portion of the sensor to make the image. If camera
shake isn't an issue, turning IS off gives you a little better
resolution.

Greg Guarino

Given other responses about horizontal panning, AUTO mode is obviously
for photographing passing AUTOmobiles ! (Or cars as we rightpondians
prefer to call them..)

I'm now going to be terribly non PC and wish you all a Merry Christmas..

T.

But, if you're *panning* with the camera, the IS will not be what you
want.

--
Bill Funk
replace "g" with "a"

Gyros are used for *camera* stabilization, but I don't know of any
gyro *image* stabilization systems being currently marketed for
consumer cameras. IIRC, they use a form of acceleratometer instead.

Well, an accelerometer certainly wouldn't work. This isn't about
accelerations. But true, I've never heard of any on-chip gyro systems.
Sure this isn't just an edge-detector and follower? Two fast subframe
readouts would give you a decent shift vector.

I accept that I want to turn off the IS if I want to pan to follow a
moving object, but why doesn't the manufacturer just tell me that?

Nah, it's something more subtle.

Is there a tech report available somewhere on DIS technology for
consumer cameras?

Check this out, everyone ...

http://www.digitalcamera-hq.com/digital-cameras/cameras-with-image-stabilization_roundup.html

"2. Digital Image Stabilization. Similar in concept to "Digital Zoom,"
this isn't a true method of stabilizing a photo, it simply uses some
digital trickery to reach a somewhat comparable goal. D.I.S. merely
ramps up the ISO-level and shutter speed, which has its drawbacks.
Doing this can increase the noise-levels on the image, making it appear
grainy. Some cameras compensate for that by sharpening the image
digitally. In the end, your photo is covered in all sorts of special
effects which could severely degrade the final image."

This too ...

http://www.digicamera.com/reviews/olympus_stylus_750/index.html

"Two kinds of image stabilization
What exactly does that mean? In essence, the Digital Image
Stabilization mode of the 740 uses a high ISO sensitivity that allows a
fast shutter speed to reduce blurriness caused by camera shake or a
fast-moving subject. The 750 uses digital image stabilization also, but
adds CCD Shift as a second, optical image stabilization. CCD Shift
Image Stabilization uses internal electrical gyro sensors to detect
camera movement and then attempt to adjust the CCD image sensor. The
goal is to keep light centered on the image sensor so that the image
remains clear despite the camera not being held totally steady."

Kinda thin on details, but if you put the two together, it starts to
make sense. The compromise is laid out well in the first clip above.
Would still like to see a real TR on this, especially a chip-level gyro
element!

The optical stabilizers are mechanical, not pixel-level electronic. They
work by using accelerometers to detect camera motion, and then feeding
that into a set of actuators that moves either a lens element or the
entire sensor to (try to) null out the vibration.

Here's Panaosnic's blurb about their variant:
http://www2.panasonic.com/webapp/wcs/stores/servlet/MegaOISExplained

The other brands work in more or less the same way.

-dms

That's the case with video cameras, but with still cameras, at least the
DSLRs that have the feature, the sensor is physically moved to compensate
for motion.

> Greg Guarino

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

Acxcording to Olympus USA, Digital Image Stabilization uses a higher
ISO and a higher shutter speed.
http://www.olympusamerica.com/cpg_section/product.asp?product=1262
Go down to "Digital Image Stabilization" and click on the link.

Fuji uses a similar gimmick in at least one of their cameras. They give it
a fancy name and they attempt to confuse buyers into thinking that their
cameras have an active Image Stabilizer mechanism, when all it really is is
an arrangement to boost the speed, thus requiring a higher shutter speed.

It would be like calling an ordinary tripod an "Image Stabilizer."

It is advertising hype, and it exploits buyers' unfamiliarity with some of
the advanced camera features, in an apparent attempt to make the camera seem
more attractive to the buyer that it really is.

No. Some video cameras with "electronic image stabilization" do this,
because they need extra picture area to be able to shift each frame into
alignment with the previous frame despite camera motion.

But EIS is pretty useless for still cameras. It removes large-scale
motion between frames in a video sequence, but each frame remains
individually blurred. Still cameras want the individual frame(s) to be
sharp.

So image stabilization for still cameras is done by moving either the
image or the sensor to remove the relative motion, and that doesn't
require an extra image area, so there's no reduction in resolution.

Dave

I believe the sensor is a "gyro" in the generic sense of measuring
angular rotation around an axis. But it doesn't use a spinning
gyroscope; the measuring element is something like a vibrating ring.

Here's a datasheet for one such sensor:
http://www.analog.com/en/prod/0%2C2877%2CADXRS150%2C00.html

Dave

The other problem with leaving the IS on with a tripod is that it will blur
the image slightly, since the IS elements are moving, relative to the other
elements and the sensor. I've only seen this once on one of my own cameras,
when I left the IS on with my 28-135 IS in the studio. This, of course,
refers only to optical IS, not digital.
The reason it isn't explained in any manual, probably, is that most
consumers won't care enough to read through an explanation.

--
Skip Middleton
www.shadowcatcherimagery.com
www.pbase.com/skipm

Unless you have a Canon or Nikon, for which the IS/VR is in the lens, so the
sensor doesn't move.

Advanced IS systems can distinguish smooth panning from shakey hands.

--
John McWilliams

My DiMAGE Z5 Manual warns to avoid image stabilization when panning a
moving object or part of the image is in motion such as a waterfall or
smoke or clouds. These are obvious problem situations where image
stabilization could make matters worse.

It does use more power and shuts itself off when the batters are running
low.

--
Pardon my spam deterrent; send email to rhod...@earthlink.net
Cheers, Steve Henning in Reading, PA USA
http://home.earthlink.net/~rhodyman

What a joke! I always assumed software image stabilization took several
short sub-exposures, then aligned them and added them together. But no,
it just cranks up the shutter speed. Advertisers are such liars.

But they don't have "sensor IS".

Yes, good point. These are not spinning gyroscopes. I found these guys
as well ... http://www.invensense.com/products/index.html
It appears that they work with a tiny vibrating mass, and the Coriolis
effect registers rotation/tilt. Cool. The Invensense ones are listed as
having been developed specifically for image stabilizing applications
in cameras (including cellphone cameras), though they are now being
widely used in gaming as well (e.g. Nintendo Wii?)

So putting it all together, these chips allow compensation for rotation
of the whole camera. That is, camera jiggle. They allow compensation
for *your* motion, not the motion of something you're shooting at. For
digital image stabilization (DIS, my Olympus FE-190) pixel shifting on
the CCD is done in accordance with the gyro output , as opposed to
actual image shifting with an active optic -- optical image
stabilization (OIS). The latter is more expensive. The compensation is
for image motion caused by camera tilts, not parallactic motion at
short range caused by pure translation.

Looking at the data sheets, these chips draw about 30mW, so including
some more power for pixel shifting, A/D conversion and processor time,
I guess they do represent a (small) power drain on the camera battery
while they are turned on.

To the extent that the cameras augment the image motion compensation to
better freeze the motion by bumping up the ISO on the CCD, you'll
"slow down" your fast moving targets, but you'll get lower S/N images
as a result. So this, plus any noise in the gyro sensor would
contribute to image degradation.

(Also, as has been pointed out in the thread above, actively panning
the camera to follow a moving target is defeated by this kind of IS.
You do NOT use IS to slow down a moving target, and should not use it
while intentionally panning!)

Both of these appear to be good reasons why, if you don't need it
(especially if you're using a tripod), you should turn IS off. QED.

Inversense has a nice on-line library (under "Support") with lots of
tech and market reports on image stabilization. See this popular
account for a nice summary

http://www.businessweek.com/magazine/content/06_13/b3977094.htm

If anyone finds anything wrong with this explanation, please let me
know. It would have been so simple for camera manufacturers to sketch
this out in their user manual, though the consumer relations folks who
write the manuals may not have a clue about the technology.

Sort of; it depemds on how smooth is it, and how fast the panning is.
In any case, this is a moot point for the OP, since his camera doesn't
actually have any form of IS at all. Instead, it only increases the
ISO and shutter speed.

Nope.
On your camera, "IS" is done by increasing the ISO (making the sensor
more sensitive by amplifying the output) and increasing the shutter
speed (to freeze any motion seen by the sensor).
There is no shifting of anything going on...
http://www.olympusamerica.com/cpg_section/product.asp?product=1262
Go down to "DIGITAL IMAGE STABILIZATION", and click the "Digital Image
Stabilization" link to see the explanation.
No gyros, no mechanism to detect motion at all.

As I've pointed out, that camera doesn't actually have IS; it only
increases the ISO to allow the shutter to operate at a faster speed.
This doesn't take up any appreciable battery power, but it does add
noise. And, since it uses a small sensor, it adds a noticeable amount
of noise.

The trade off is this: "IS" on lets you freeze action (whether the
action is movement by the subject or the camera) with more noise, or
more blur in the image from the movement.
So, noise or blur; take your choice.

In other words, ersatz "IS". I'd also put the word "done" above in
quotes....

--
John Mcwilliams

That might well be true. Thanks for pointing this out. Now, Olympus
says that their IS technology on FE-190 boosts ISO, but it doesn't say
that it doesn't use active stabilization. But you could well be right,
and I'm disappointed if, in fact, it doesn't use actual tilt sensors.
In fact, elsewhere it says that the IS technique doesn't reduce image
size, which probably is shorthand for saying that it doesn't do pixel
shifting.

To the extent that no consumer cameras actually use this new
tilt-sensor technology (no, it's NOT an accelerometer it's really an
orientation sensor), I guess that's really misleading advertising.
There is no "stabilization" as such going on at all! When the image
processing community talks about IS, they aren't talking about making
shutter speeds faster!

Oh well, so I guess I figured out what DIS *could* be, if the camera
manufacturers would get their act together. Too bad if it isn't.

But it is very useful to know at least that turning on DIS increases
noise, and why I should turn it off if I don't need it.

The Canon 28-135 IS has a panning only mode. One of 2 modes for the IS
system.

From Wikipedia:

Optical Image Stabilization
An Optical Image Stabilizer, often abbreviated as OIS, is a mechanism used
in a digital still camera or video camera that stabilizes the recorded image
by varying the optical path to the sensor.[1] In Canon's implementation, it
works by using a floating lens element that is moved orthogonally to the
optical axis of the lens, using electromagnets. The vibration signal which
is compensated for by the stabilizing lens element is typically aquired
using two piezoelectric angular velocity sensors (often also called
gyroscopic sensors).[2]

>
>

This is a problem, leading some equipment to have a 'pan' mode that
disables the IS action. IS is much more effective as canceling fast
random motion, and ineffective at canceling slow, steady motion, which
is why it is designed as it is.

You are correct. I was mistaken. I was thinking of my 70-300 that has both.

Sorry, I'm suffering from thread reading overload. I read it, but it didn't
penetrate...

.. and to me, this mis-representation is quite deliberate, and reduces my
trust in a company which resorts to such tactics.

David

Well, I guess that those in-the-know ...

http://www.digitalcamera-hq.com/digital-cameras/cameras-with-image-stabilization_roundup.html

have it right. DIS is NOT active image stabilization of any kind. Just
ISO boosting. OIS involves optically moving the image on the chip in
response to a sensor, and "CCD SHIFT image stabilization" (CIS?) is
what I thought was going on, which is using pixel shifts on the chip to
move the image in response to a sensor.

I still think that DIS is a term that implies a lot more than just fast
shutter speed, and the terminology certainly smells of
misrepresentation.

Perhaps no wonder camera manufacturers aren't anxious to spell it out
carefully in their documentation.

It does sound like CIS for inexpensive consumer digital photography is
a coming thing.

Thanks. I stand corrected.

Greg Guarino

True for Olympus, but others (some Canon, some Nikon, all Panasonic, a
few Kodak) use "real" image stabilization, a moving lens element or
moving sensor.

I did not bother to check, because my six year old Olympus E-100 RS features
real optical image stabilisation, apparently using a rotating sensor.