Image Stabilization vs Noise

On 23 Dec 2006 17:48:15 -0800, Paul Rubin
wrote:

jpc writes:
So if anyone who own a IS P&S is willing to spend a couple hours of
their life to help answer this burning question, I'm willing to offer
advice and comentary. If I'm wrong, I'm wrong. But if I'm right, 2
bucks worth of chain vs 400 bucks worth of camera makes for an
excelent price/perfomance ratio

I have an Olympus E-100RS, an older Olympus P&S with IS and with RAW.
It has 1.3MP and a 1/2" sensor, giving unusually large pixels compared
with today's P&S's. On the other hand, being an older camera I think
its CCD noise performance isn't as good as newer ones. I'd be willing
to do some test shots if someone wants to analyze them or tell me how
to analyze them.

Thanks for the offer. I have a few ideas about a simple test
proceedure that should work but I want to chrck it out before I post
it. I have people coming over for Xmas and it might be a couple days
before I get back to you.

jpc

"Skip" wrote:

camera and while their wives held their tripods. There were several
situations where we had lots of time and they were able to use their
tripods, but they were constantly knocking people in the back of the
head since they had no agility.

Hadn't these people ever heard of a monopod?

If you don't spread the legs on a tripod, it basically is a monopod.
They didn't have the option to spread the legs so they were using
monopods. They still were impractical since they were on one side of
the Land Rover and the animals weren't always on that side and were
moving. A number of us were rather upset at being banged in the head by
the lenses on these dudes tripods.

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

On Sat, 23 Dec 2006 16:17:19 -0500, Stephen Henning
wrote:

jpc wrote:

Thanks for the info. Didn't expect it to work anywhere as good as a
tripod. As for IS what kind of equipment are you comparing it to?

I just returned from a Safari trip to Africa where I took 1700 420mm
(equivalent) photos hand held and most came out great. I would not have
even tried that without IS. I am using a DiMAGE Z5. I saw several
people trying to use a tripod in the back of a Land Rover and they
didn't get many shots off. They mostly used an unstabilized hand held
camera and while their wives held their tripods. There were several
situations where we had lots of time and they were able to use their
tripods, but they were constantly knocking people in the back of the
head since they had no agility.

When I memtioned tripods, I was thinking more of moonlit landscapes
and foggy city streets at midnight, not on a boucing land rover.
Under those circumstances, a tripod, monopod and whatever would be
worse than useless.

I was able to take sharp photos at 420 mm (equiv) shooting no faster
than 1/100. In fact our guides used the image on my camera to identify
distant animals rather than their binoculars. They were very impressed.

Thanks for the recomendation. A photo group L belong to is having a
major show this summer featuing animal photography--major for us
anyway. It's a juried show and if I decide I have any chance of
competing with some excellent nature photographers who have been at it
for decades, I'm thinking of investing in a more suitable camera.

That's how I got on this IS vs noise kick--trying to decide on a
camera that works well with long telephoto and very low liight shots


jpc.

David J Taylor wrote:
DougL wrote:
[]
We spend several days and a lot of words trying to establish exactly
who has what. As far as the thread participants were concerned, we
ended up concluding that point and shoot cameras did NOT have OIS. Can
you supply a pointer that says otherwise?

Would be interesting to know who really does image stabilization
(either CCD-shift or optical) in a point-and-shoot.

Yes, "moving mirror" was my shorthand. "Moving optic" would have been
more accurate.


You concluded incorrectly.

Plenty of non-SLR cameras have pure optical image stabilsation, for
example lens-based in many of the Panasonic cameras (FZ20, FZ5, FZ30,
FZ50, FZ7 and many of their compact cameras as well), and Canon S1-IS,
S2-IS, S3-IS, and Minolta have (had?) moving sensor (Z3 etc.).

Some pointers:

http://www.dpreview.com/reviews/panasonicfz20/

http://www.dpreview.com/reviews/panasonictz1/

http://www.dpreview.com/reviews/canons3is/

http://www.dpreview.com/news/0407/04070703dimagez3.asp

David

I very much appreciate the wisdom. This started out as a digital camera
newbie (me) searching for some technical info on that was going on in
point and shoot cameras with advertised IS. Many manufacturers
(Olympus, evidently) are remarkably cagey about what exactly
constitutes image stabilization in their cameras. It would be really
handy to see a general review about who-has-what.

I'm quite impressed that there is actually a moving-sensor mode for IS.
Had no idea.

Quite frustrating that IS means different things to different
manufacturers, and at least some manufacturers consider ISO boosting to
mean IS. I mean, one could consider the 1/4-20 screw on the bottom of
the camera to provide IS potential.

To recap my response to this thread, IF image stabilization on your
camera really means ISO boosting, then you are better off turning it
off if you don't need it, as there is a price to be paid for it in
image quality.

wrote:
[]
I very much appreciate the wisdom. This started out as a digital
camera newbie (me) searching for some technical info on that was
going on in point and shoot cameras with advertised IS. Many
manufacturers (Olympus, evidently) are remarkably cagey about what
exactly constitutes image stabilization in their cameras. It would be
really handy to see a general review about who-has-what.

Lies, damn lies and marketing-speak!

I'm quite impressed that there is actually a moving-sensor mode for
IS. Had no idea.

Moving lens elements too. Each has its own advantages. Seems to have
been something they were able to make relatively easily.

Quite frustrating that IS means different things to different
manufacturers, and at least some manufacturers consider ISO boosting
to mean IS. I mean, one could consider the 1/4-20 screw on the bottom
of the camera to provide IS potential.

I detest this sort of deliberate dishonesty.

To recap my response to this thread, IF image stabilization on your
camera really means ISO boosting, then you are better off turning it
off if you don't need it, as there is a price to be paid for it in
image quality.

Agreed - but sometimes the grain from higher ISOs can add character to the
image, particularly of you convert it to black and white....

Merry Christmas!

David

On Sun, 24 Dec 2006 09:55:00 -0600, jpc wrote:

I think most of the IS lenses have been pretty tightly nailed for
noise. However, a few lenses do inject interference back via the power
lines, and Canon have updated main PCBs for those lenses.

Thanks for confirming that IS can cause noise problems in
some system. My hypothisis was feeling a bit lonely (g)

Don't worry about it. Some people have a compulsion to give
others a bit of static when they have nothing better to do.

On 23 Dec 2006 17:48:15 -0800, Paul Rubin
wrote:

jpc writes:
So if anyone who own a IS P&S is willing to spend a couple hours of
their life to help answer this burning question, I'm willing to offer
advice and comentary. If I'm wrong, I'm wrong. But if I'm right, 2
bucks worth of chain vs 400 bucks worth of camera makes for an
excelent price/perfomance ratio

I have an Olympus E-100RS, an older Olympus P&S with IS and with RAW.
It has 1.3MP and a 1/2" sensor, giving unusually large pixels compared
with today's P&S's. On the other hand, being an older camera I think
its CCD noise performance isn't as good as newer ones. I'd be willing
to do some test shots if someone wants to analyze them or tell me how
to analyze them.


X-mas is over and back to image noise. My wife even bought me a new
camera-- a $10 keychain to add to my colllection of really bad
cameras. Some friends have been running around with Hoyas and other
plastic film cameras to see who can take the best BAAAAD pictures, I'm
trying to out do them all with digital.

Here is the proceedure I promised.

I- Downlaad the latest ImageJ--ver1.37-- from
http://rsb.info.nih.gov/ij/download.html and install

Depending on how old your computes is, you might want the package with
the latest Java. The documention package is a little outdated, but you
might want to download that too.

2-Go to http://www.imaging-resource.com and download the ISO400 low
light DaveBox images for the E-100RS, the EOS-5D and the S3IS since
that the camera I suspect has noisy piezo motors The low light images
are near the end of the sample images and the newer ones have the code
(camera--LL-iso#-3) where the 3 refers to a light level of 1 fc. That
is roughly what you'd find on a city street at night.

3-Crop out the two grey scales on the right side of the hinge and save
as tifs--this speeds up ImageJ . Open an image with ImageJ
grab the line icon on the tool bar and drag a straight line (the
default) down the long grey scale. Hit cnrl-K to make a stair case
graph of the intensity (0-256). The noise is superimposed on each
step.

The E100RS 400ISO graph is unusual--low noise down to middle grey then
much more noise on the lower steps. The 5D--our standard--has low
noise on all steps. The S3IS starts out noisy and keeps getting
noisier

4--Go to the Analyze/Set Measurements and check the Standard
Deviation. Mean Grey Value, and Dispay Label boxes. Drag a line in
one of the bigger boxes in the second greyscale. Hit control-K to
check that there isn't an upward or downward slope to the graph.
Close the graph and hit cntl-M. In the results box, the mean is the
intensity of light and the StdDev is the noise.

5-Create a set of grey scale patchs for your own camera.
I placed a sheet of white paper under a desk lamp, put the camera's
manual focus at infinity to blur any details and adjusted the custom
white balance, before I took a series of images at f8 and ISO400.
Starting at 1/800 sec I kept halving the shutter speed until I had
7 images . Using the proceedure I discribed I measured the signal and
noise for each patch.

By now some might wonder why go to so much trouble. So here is the
short version of why this is necessary,

The measurements I collected were

1/13 sec-- max light -- intensity 251
1/25 sec---0.5 of light intensity 203
--
--
1/400 sec--0.03% light--intensity 40
1/800 sec-0.015%light -intensity 24

Both digital and film images are gamma corrected. With film it's
because that is the way film works. With digital it's because the
camera engineers added gamma or contrast step in the firmware

At the particular setting I was using, both the signal and the noise
were applified 6.4 times over 7 stops. If I'd set my illumination more
carefull and could have gone 9 or 10 stops before the camera
saturated, I would have seen about 12X. aplification when the intesity
dropped to around 3

The means that while I was seeing something close to constant noise in
each measurement, a StdDev around 2, the real noise was 6.4 times
higher in the brighest patch.

Why? Simple. The noise is from the LIGHT not the CAMERA.

More light, more noise. With 256 times more light, 16 times more
noise. This applies to any camera of any design that's ever going to
be made. Because this is the way the universe works.

But what about CAMERA noise. That noise dominates in the dark patches
where there is very very little light to generate noise. Down in the
dark where you need all the S/N you can find, the camera design become
very important.

So if you're on safari in Africa with more light than you can use, IS
is the greatest thing since sliced bread. No matter how much noise
ithe piezo motors are creating, it won't show up in your photos. The
light noise is far too strong.

But if you're taking candid band shots in a smoky bar, and your
mechanical IS is adding 4 times more noise as may be happening in the
CanonS3, IS becomes less than useless. You're better off setting a
shutter speed that will freeze camera shake, dialing in your highest
ISO setting and correcting the exposure and noise using sortware

6--This step is obvious. Take two sets of gray scale shots with IS off
and on, and compare noise levels. If the noise levels are consistant
higher in the IS images-noisy camera. If they bounce around--we are
dealing with nosie--the Oly engineers did a good job on the design.


7--Let us know how it works out.

jpc