Another ISO question...

John Sheehy wrote:
Scott W wrote in
:

John Sheehy wrote:

If you compare a crop from the 5D sensor the size of an FZ50 sensor,
then there is no contest. The only area the 5D is going to excel in
is image read noise at ISOs 800 and 1600. The FZ50 crop will have
less read noise at ISO 100, and slightly less shot noise, image-wise
(bin/downsample or look at the fine noise with all the extra detail -
your choice).

There are a number of flaws in your argument, and you present no
actual data to prove your position. You simply state
results, but again, with no actual data to prove your position.

Let's take your small pixel to a logical end:
pixels so small the well depth is 1 photon (electron), and
with read noise of 1 electron. So every pixel has maximum
signal to noise ratio of 1, dynamic range is 1.

I don't understand this part, I assume we are talking about binning
enough pixels on the FZ50 so that the area comes out to the same on
the 5D? If so then if the read noise on the FZ50 was the same on a
pixels basis as the 5D binning 16 pixels would increase the read noise
by a factor of 4.

Yes; in electrons, or absolute intensity. Relative to the new super-
pixel's photon capture, though, it is only 1/4 as strong.
Signal:ShotNoise has increased 4x, or 2 stops.

Huh? You can not combine pixels and actually gain more
more signal/noise than one larger pixel. The 16 small pixels
binned would have signal/noise 1/4 that of a single large
pixel.

Figure it this way, you capture say 4000 electrons
over your 16 pixels, the 5D capture 4000 electrons on its one pixel.
Say both have a read noise of 5 electrons, the 5D will have a read
noise of 5 out of 4000, but the binned pixels from the FZ50 will have
a read noise of 20 out of 4000.

The fact of the matter is, however, that at ISO 100, the FZ50 original
pixels have about 3.25 electrons of read noise (slightly less at ISO
200), and the 5D has about 29 electrons of read noise, and 3.25 times 4
is still a bit less than 29. Over a stop less.

3.25 electrons sets a new low among consumer digital cameras.
Can you cite a report, or show data that proves this value?

DSLRs at low ISO are not sensor read noise limited; they are
limited by the 12-bit A/D converter noise floors. DSLRs like
the 5D have read noise of 3 to 4 electrons.

You have found a condition in the electronics that is currently
a limitation that is in the process of being improved with
higher bit A/D converters (e.g. the 1D Mark III). But you
make a non-existent ~200 megapixel small pixel sensor for your
argument. While we are nowhere near the 200 megapixel sensor,
we are getting closer to having lower noise A/Ds. To make
fair comparisons, you need to use read noise closer to
the true read noise of the sensor, and that is 3 to 4 electrons
for each sensor (and likely to get a little better, both for
large and small pixel sensors).

But the 16 pixels from the FZ50 are not likely to even capture as many
electrons as the 5D, since the fill factor is likely to be a lot less.

It isn't. The FZ50 in 16.5 pixels captures the same number of maximum
photons as the 5D; ~80,000. The FZ50 captures them faster, though, with
a higher quantum efficiency! At ISO 100, the FZ50 captures about 80K in
16.5 pixels, as compared to 52K in the 5D.

Where is the report on quantum efficiency and well capacity?
This does not make sense as there is a finite wall size needed
between pixels to prevent electrons from bleeding into adjacent
wells. This effect can be seen in plots like Figure 1 at:
http://www.clarkvision.com/imagedeta...rmance.summary
The trend toward lower pixel pitch does not project to zero
because of the walls between pixels.

So your 80,000 electrons/16.5 pixels = 4848 electrons/pixel,
which is on the high side for such small pixels.

Note even with 4848 electrons and read noise o 3.25 electrons,
the dynamic range is only 1491. That really isn't very good.
You'll clip many highlights and lose a lot in the shadows.

That is the key: pixels need to be large enough to capture a good
dynamic range. Small pixels do not do that. You need to look
at the whole picture!

The horror stories about the inefficiencies of tiny pixels are greatly
overstated. The real "horror" comes from the size of the sensors they
are historically found in; the FZ50's sensor is a good size for easy lens
design, but it does not capture a lot of photons total, even though it is
fairly efficient per unit of area. It is unreasonable to declare that
tiny-sensor images are noisier than DSLR images because of small pixels;
the real reasons are the smaller total number of photons collected, and
lack of optimized high-ISO readout on P&S cameras. Even so, the FZ50 has
a higher max-signal-to-read-noise ratios than a Nikon D2X, at all ISOs.
The FZ50 has a higher SNR than the DX2, in the shadow areas, at the pixel
level, without any binning for the FZ50.

Maximum signal-to-noise ratios in modern good digital cameras are
determined by the number of photons collected in each pixel.
It is a physical fact that photons are finite and photon density
in the focal plane of a camera is finite. Photons/square micron
in a focal plane is finite (and not a very large number).
It is simple math to see:

photons/square microns * more square microns = more photons captured

I am not advocating binning, per se. I'd rather have the original, full-
res image with all of its fine noise. I present the binned super-pixel
as a bridge concept, as it is something that one can grasp even if one
doesn't appreciate the concept that the power of noise in a displayed
image is not dependent only upon traditional noise statistics, but upon
the displayed size of the original pixel as well. They are two factors
affecting the subjective experience of noise. Another factor is the the
sharpness and contrast of the actual signal; the more easily it is
perceived, the less relevant the noise tends to be.

You and Ilya (who proposed insanely small pixel a year or so
ago) should go into business with your concept. If it works,
you'll make millions.

Most notions of binning giving better IQ come from contexts where the
result of the binning is viewed at the same pixel size as the original.
Even when they *are* displayed at the same subject size, if the original
resolution is viewed with any downsizing, the Nearest Neighbor algorithm,
or a hybrid thereof is often used which drops the influence of some
pixels, cancelling less noise. Nearest Neighbor maintains pixel-level
noise while reducing the number of pixels, a bad handicap for an image
downsized that way (NN actually increases it at the nyquist in the first
halving of resolution, with Bayer CFA cameras).

Your concept of more pixels does have merit. Certainly more
pixels in a given size sensor means less enlargement/pixel for
a given final print size. But people get too caught up in
resolution without considering all the other factors that
make images great. Many of the arguments you present have a
limit like film: film grains are very small 1-bit sensors.
Why not go back to film? The reason is there are numerous
limiting factors, including, but not limited to:
diffraction, lens aberrations, dynamic range, signal-to-noise,
ability to manufacture sharp lenses at a reasonable cost.
When all of that is considered, one finds larger pixels
actually deliver better image quality. But what drives pixels
smaller includes, but is not limited to: camera bulk (size
and weight), cost. The very high quality images with reasonable
cost lenses come in the range of 6 to 8 microns. Larger pixels
and the cost goes up (if you want to have a lot of them.
Smaller pixel and performance drops.

Take a crop of any noisy image in PS, duplicate a few copies of it, and
pixelate the copies 2x2, 3x3, 4x4, etc. The pixelated ones do not really
have less noise - the noise just gets coarser and has smoother gradients
to it. To me, that's worse - I'd rather see the fine noise and the
greater detail; my brain knows what it is, and knows to stop looking for
further details, and is happily focused. The quest for reduced pixel-
level noise has two main paths; a useful one where noise is reduced while
all else remains equal or better, and a useless one, entailing binning
and big pixels (which are actually a mechanical form of binning photons),
which is much like a dog chasing after its own tail, IMO.


We'll have to disagree.

Roger

John Sheehy wrote:
"Roger N. Clark (change username to rnclark)" wrote
in :

Not on the cameras I've tested. Comes out 51,000. The 10D was
44,000.

You might want to check the linearity of the highlights on both of those
cameras.

My 20D goea all the way up to 4095 in the RAW data at ISO 100, but if you
compare to an ISO 200 image, the RAW levels are the same throughout the
shadows and highlights, but at the high end the RAW values go
increasingly too high at ISO 100, and clips at a lower point in a
highlight gradient. My 10D is just the opposite. It rolls in extra
highlights just below RAW saturation at ISO 100. You can test these
tings by shooting an OOF gradient of a light falling off along a wall,
and make split images at ISO 100 and 200 with half the exposure time.
After you calibrate for any small differences in global RAW levels, you
might see that the images differ only in extreme highlights; the 10D
capturing more highlights before clipping at ISO 100, and the 20D
capturing less at ISO 100.

I'm using industry standard testing methods, and that is not
what I see in my camera.

Roger

John Sheehy wrote:
"Roger N. Clark wrote
Unity gain ISO on the FZ50 must be below 100.

Unity gain is a meaningless concept, in the face of analog noise. There
is no meaningful relationship between the charge of an electron and an
ADU. An ADU has a totally arbitrary value.

(I'll use DN = Data Number = your ADU; DN is used in the
sciences where digital image analyses is done.)

No it is not. While the camera manufacturer chooses the analog
gain (which in a sense is arbitrary), the measurement of
gain in electron/DN calibrates a sensor. Thus when one sees a
particular DN read in photoshop, with the gain and conversion
to 16-bit numbers, one can directly compute the number of
photons that number means. That gives you an absolute
light meter.

As to unity gain, it is quite meaningful. Regardless of
analog electronics, the underlying signal originates
as a quantum of light (a photon) that is converted to
an electron. If the noise in the analog system were zero
and the A/D converter could measure 1/10 electron (not that
that is possible) you would see the signal out of the sensor
jumping in units of 10 DN since the sensor is only
recording "whole" electrons.

Calibrated gains are used to measure photon levels all
the time in sciences using CCD, CMOS and other electronic
sensors.

Amateur astronomers are using DSLRs to record extremely
faint stars, galaxies, and nebulae. They have found that
increasing ISO above the unity gain level does not improve
the detection of faint objects, and it only reduces dynamic
range.

So, digital data does have important meaning.

Roger

ASAAR wrote:
On Thu, 12 Jul 2007 06:06:54 GMT, David J Taylor wrote:

I have heard that having an IS feature on a lens can make as much
as 3 fstops worth of difference in helping to attain a decent
shake-free image in low light/smaller aperture settings..is it as
high as 3 fstops or is that a bold claim?

On some of the Panasonic Lumix cameras, yes. On some DSLR lenses
perhaps a little less (the glass is heavier and not so easy to move
dynamically).

I don't think so. Nikon claims a 3 f/stop improvement with their
lightweight 55-200mm VR DX lens, and a 4 f/stop improvement with the
larger, heavier full frame 70-300mm VR II lens. From what I've seen
of the 55-200mm VR lens, Nikon sure isn't overstating the 3 f/stop
improvement.

Unfortunately, there is, as yet, no standardised test for IS/VR lenses, so
we don't really know what improvement is offered and at what vibration
frequencies. Moving larger masses can be done, but takes more power, so
there will be a trade-off between the amount of stabilisation (or the
frequency range over which it works) and battery life / current
consumption (which is rarely quoted).

I have seen reports suggesting that the IS is less effective on some
lenses, but without quantitative tests, who knows.

David

On Sat, 14 Jul 2007 06:38:32 GMT, David J Taylor wrote:

I have heard that having an IS feature on a lens can make as much
as 3 fstops worth of difference in helping to attain a decent
shake-free image in low light/smaller aperture settings..is it as
high as 3 fstops or is that a bold claim?

On some of the Panasonic Lumix cameras, yes. On some DSLR lenses
perhaps a little less (the glass is heavier and not so easy to move
dynamically).

I don't think so. Nikon claims a 3 f/stop improvement with their
lightweight 55-200mm VR DX lens, and a 4 f/stop improvement with the
larger, heavier full frame 70-300mm VR II lens. From what I've seen
of the 55-200mm VR lens, Nikon sure isn't overstating the 3 f/stop
improvement.

Unfortunately, there is, as yet, no standardised test for IS/VR lenses, so
we don't really know what improvement is offered and at what vibration
frequencies. Moving larger masses can be done, but takes more power, so
there will be a trade-off between the amount of stabilisation (or the
frequency range over which it works) and battery life / current
consumption (which is rarely quoted).

I have seen reports suggesting that the IS is less effective on some
lenses, but without quantitative tests, who knows.

Something made you shift gears here, and it looks like your long
standing pro-Panasonic bias is the culprit. First, the_niner_nation
wrote that he heard that lens based IS can produce up to 3 stops of
stabilization. You immediately countered with :

On some of the Panasonic Lumix cameras, yes. On some
DSLR lenses perhaps a little less (the glass is heavier and
not so easy to move dynamically).

So without either agreeing or disagreeing that lens based IS can
produce up to 3 stops of stabilization, you then stated that
Panasonic's IS can get up to that amount of improvement with some of
its cameras, but lens IS designs don't do as well. I'd like to know
which orifice that was pulled out of.

I then said that based on personal experience, Nikon's older VR
seemed to be producing at least 3 stops worth of VR improvement, and
you then countered that because there are no standardized tests, we
can't really know how much improvement there really is. You can't
have it both ways, not having standard tests, yet proclaiming that
Panasonic does IS better. I'm not going to search for it, but I'm
pretty sure that in some older message, you stated that Panasonic's
IS was good for up to 4 stops of improvement. Was that based on
some organization's testing, or your own tests, or as I suspect,
your ability to see Panasonic cameras through rose colored filters?

You then switched gears and offered yet another reason why
Panasonic's IS was better. That lens based IS necessarily needs to
move larger masses, and so it will require more battery wasting
power, shortening battery life. Pure rubbish. I hate to have to
tell you this, but Panasonic's IS also shortens battery life. And
as the Panasonic cameras are good for hundreds of shots per charge,
whether IS is used or not, when compared with lens based VR/IS,
where DSLRs can get up to thousands of shots from their batteries
per charge whether using IS or not, I don't think that you've made
even a weak case for Panasonic's IS. Maybe the sock puppet
anti-DSLR trolls will now jump in to provide you some aid and
comfort. g

ASAAR wrote:
On Sat, 14 Jul 2007 06:38:32 GMT, David J Taylor wrote:

I have heard that having an IS feature on a lens can make as much
as 3 fstops worth of difference in helping to attain a decent
shake-free image in low light/smaller aperture settings..is it as
high as 3 fstops or is that a bold claim?

On some of the Panasonic Lumix cameras, yes. On some DSLR lenses
perhaps a little less (the glass is heavier and not so easy to move
dynamically).

I don't think so. Nikon claims a 3 f/stop improvement with their
lightweight 55-200mm VR DX lens, and a 4 f/stop improvement with the
larger, heavier full frame 70-300mm VR II lens. From what I've seen
of the 55-200mm VR lens, Nikon sure isn't overstating the 3 f/stop
improvement.

Unfortunately, there is, as yet, no standardised test for IS/VR
lenses, so we don't really know what improvement is offered and at
what vibration frequencies. Moving larger masses can be done, but
takes more power, so there will be a trade-off between the amount of
stabilisation (or the frequency range over which it works) and
battery life / current consumption (which is rarely quoted).

I have seen reports suggesting that the IS is less effective on some
lenses, but without quantitative tests, who knows.

Something made you shift gears here, and it looks like your long
standing pro-Panasonic bias is the culprit. First, the_niner_nation
wrote that he heard that lens based IS can produce up to 3 stops of
stabilization. You immediately countered with :

On some of the Panasonic Lumix cameras, yes. On some
DSLR lenses perhaps a little less (the glass is heavier and
not so easy to move dynamically).

So without either agreeing or disagreeing that lens based IS can
produce up to 3 stops of stabilization, you then stated that
Panasonic's IS can get up to that amount of improvement with some of
its cameras, but lens IS designs don't do as well. I'd like to know
which orifice that was pulled out of.

I then said that based on personal experience, Nikon's older VR
seemed to be producing at least 3 stops worth of VR improvement, and
you then countered that because there are no standardized tests, we
can't really know how much improvement there really is. You can't
have it both ways, not having standard tests, yet proclaiming that
Panasonic does IS better. I'm not going to search for it, but I'm
pretty sure that in some older message, you stated that Panasonic's
IS was good for up to 4 stops of improvement. Was that based on
some organization's testing, or your own tests, or as I suspect,
your ability to see Panasonic cameras through rose colored filters?

You then switched gears and offered yet another reason why
Panasonic's IS was better. That lens based IS necessarily needs to
move larger masses, and so it will require more battery wasting
power, shortening battery life. Pure rubbish. I hate to have to
tell you this, but Panasonic's IS also shortens battery life. And
as the Panasonic cameras are good for hundreds of shots per charge,
whether IS is used or not, when compared with lens based VR/IS,
where DSLRs can get up to thousands of shots from their batteries
per charge whether using IS or not, I don't think that you've made
even a weak case for Panasonic's IS. Maybe the sock puppet
anti-DSLR trolls will now jump in to provide you some aid and
comfort. g


I completely stand by my statement that moving larger masses by the same
amount requires more power. I would be interested to know whether the
mass which needs to be moved is greater in a DSLR lens than in a compact
camera lens. Of coure, the power is taken from the battery whatever
system is used, and I have not said otherwise. If you have any figures on
battery life with and without IS, I would be interested to see them.

I make no other distinction between Canon and Panasonic compact cameras
and DSLR in-lens IS. They use the same principles, as I am sure you know.
My personal experience with the Panasonic system matches yours with the
Nikon system, in that subjectively they both produce about three stops
improvement. I have not said they don't.

There are plenty of reports on the Web which comment that different lens
arrangements produce different degrees of image stabilisation, and plenty
of subjective statements as to exactly how much is achieved. Do you not
think that a standardised, objective test would be a worthwhile
improvement?

I am not trying to "case for Panasonic's IS" - simply to report what I
know and what I have read, and I see no justifcation for the tone of your
post or your rather repeitious references to "trolls". I see roles for
both DSLR and compact cameras - and thought you did as well.

David

On Sat, 14 Jul 2007 14:10:25 GMT, David J Taylor wrote:

I completely stand by my statement that moving larger masses by the same
amount requires more power.

Typical. A truism is a truism is a truism. And you used it to
avoid having to deal with the real issue(s), which has nothing to do
with moving mass. You really are capable of understanding the point
I was trying to make, but I knew that you'd try to avoid dealing
with it.


There are plenty of reports on the Web which comment that different lens
arrangements produce different degrees of image stabilisation, and plenty
of subjective statements as to exactly how much is achieved. Do you not
think that a standardised, objective test would be a worthwhile
improvement?

Of course it would be good to have such standardized tests. But
since we don't have them, statements such as yours, claiming IS/VR
superiority for Panasonic's IS method, without (as far as I can see)
any good, supporting tests, is suspect at best. In fact, I've seen
quite the opposite, that some people report that the best IS is
found on lenses. Lenses that cost more money than I'm likely to
spend. Again, this is a red herring, tossed out to avoid dealing
with the point(s) I tried to make. So what else is new?


I am not trying to "case for Panasonic's IS" - simply to report what I
know and what I have read, and I see no justifcation for the tone of your
post or your rather repeitious references to "trolls". I see roles for
both DSLR and compact cameras - and thought you did as well.

You "thought"? No way. You're not thinking at all if you "think"
that I don't see roles for DSLR and compact cameras. I frequently
recommend P&S cameras, and recent recommendations included models
not only from Fuji and Canon (I own some of their models) but
Panasonic as well, and I've never owned any of their cameras. You
also recommend both P&S and DSLR cameras, but unlike almost all
other posters in this newsgroup, tend to frequently succumb to the
influence or power of the one, true brand.

What I don't see a role for is excessive, knee-jerk bias. I don't
think that you're a troll, but there has recently been a rash of
anti-DSLR trolls in the ng, and I was just commenting on that, and
that even if they jump in, appearing to support you, you might want
to avoid their phony support. The only one that is really guilty of
that, so far, is SMS, but then he also might as well be from their
home planet. I don't think that you're anti-DSLR at all, just
immensely, perhaps fanatically supportive of Panasonic products.

ASAAR wrote:
On Sat, 14 Jul 2007 14:10:25 GMT, David J Taylor wrote:

I completely stand by my statement that moving larger masses by the
same amount requires more power.

Typical. A truism is a truism is a truism. And you used it to
avoid having to deal with the real issue(s), which has nothing to do
with moving mass. You really are capable of understanding the point
I was trying to make, but I knew that you'd try to avoid dealing
with it.


There are plenty of reports on the Web which comment that different
lens arrangements produce different degrees of image stabilisation,
and plenty of subjective statements as to exactly how much is
achieved. Do you not think that a standardised, objective test
would be a worthwhile improvement?

Of course it would be good to have such standardized tests. But
since we don't have them, statements such as yours, claiming IS/VR
superiority for Panasonic's IS method, without (as far as I can see)
any good, supporting tests, is suspect at best. In fact, I've seen
quite the opposite, that some people report that the best IS is
found on lenses. Lenses that cost more money than I'm likely to
spend. Again, this is a red herring, tossed out to avoid dealing
with the point(s) I tried to make. So what else is new?


I am not trying to "case for Panasonic's IS" - simply to report what
I know and what I have read, and I see no justifcation for the tone
of your post or your rather repeitious references to "trolls". I
see roles for both DSLR and compact cameras - and thought you did as
well.

You "thought"? No way. You're not thinking at all if you "think"
that I don't see roles for DSLR and compact cameras. I frequently
recommend P&S cameras, and recent recommendations included models
not only from Fuji and Canon (I own some of their models) but
Panasonic as well, and I've never owned any of their cameras. You
also recommend both P&S and DSLR cameras, but unlike almost all
other posters in this newsgroup, tend to frequently succumb to the
influence or power of the one, true brand.

What I don't see a role for is excessive, knee-jerk bias. I don't
think that you're a troll, but there has recently been a rash of
anti-DSLR trolls in the ng, and I was just commenting on that, and
that even if they jump in, appearing to support you, you might want
to avoid their phony support. The only one that is really guilty of
that, so far, is SMS, but then he also might as well be from their
home planet. I don't think that you're anti-DSLR at all, just
immensely, perhaps fanatically supportive of Panasonic products.

As I have said before, I prefer to recommend products of which I have
direct or very close experience. I have no doubt that Panasonic cameras
are some of the best non-SLRs made, and many people have reported similar
experiences. I have also recommended Nikon and Ricoh cameras based on
direct experience, and Canon based on a friends experience (I even helped
him choose the camera).

I am sure you will have reports on the Internet where come DSLR lenses are
described as having rather less than 3 stops of IS gain, and I was simply
reporting this fact, together with my own experiences over a few years
with a small-sensor, small-lens camera where the three stops is a justfied
claim as far as I can tell. I hypothesised why that might be. Hardly a
"knee-jerk bias".

Note that both the Panasonic system and DSLR-lens systems are in-lens IS,
with moving elements and not otherwise as you may have implied.

David

On Sat, 14 Jul 2007 14:51:08 GMT, David J Taylor wrote:

As I have said before, I prefer to recommend products of which I have
direct or very close experience. I have no doubt that Panasonic cameras
are some of the best non-SLRs made, and many people have reported similar
experiences. I have also recommended Nikon and Ricoh cameras based on
direct experience, and Canon based on a friends experience (I even helped
him choose the camera).

You've said that before, but that sidesteps the issue. It's one
thing to recommend a product that you've used, and I can understand
that, but it's quite another to offer proclaim some performance
advantage that you probably have no real data for, just to counter
someone else's opinion about another product. And if you only want
to recommend products that you've used, it's only fair to avoid
disparaging products that you haven't used.


I am sure you will have reports on the Internet where come DSLR
lenses are described as having rather less than 3 stops of IS gain,
and I was simply reporting this fact,

Demonstrably untrue. If you said that, there would only have been
agreement. You instead rejected the statement that some lenses have
up to 3 stops of IS gain by disagreeing, saying that some Panasonic
cameras can do that well, but due to having to move more mass, lens
based IS probably doesn't do as well as the Panasonics.


Note that both the Panasonic system and DSLR-lens systems are in-lens IS,
with moving elements and not otherwise as you may have implied.

Nothing was implied. Panasonic could have used gyroscopes for all
that I care. It doesn't have replaceable lenses so as far as the
end user is concerned, it's a black box issue. So what are we to
think, that since Panasonic uses a similar lens based IS technique,
and that IS lenses for DSLRs are typically larger, that Panasonic
cameras have an inherent advantage and their IS does better than the
IS employed by DSLR lenses? Remember now, you've already mentioned
that you're not aware of any standardized IS tests, so if that's
what you think, it can only be a guess. But even if it is a guess,
it would be nice if you explicitly state what you've been implying.

ASAAR wrote:
On Sat, 14 Jul 2007 14:51:08 GMT, David J Taylor wrote:

As I have said before, I prefer to recommend products of which I have
direct or very close experience. I have no doubt that Panasonic
cameras are some of the best non-SLRs made, and many people have
reported similar experiences. I have also recommended Nikon and
Ricoh cameras based on direct experience, and Canon based on a
friends experience (I even helped him choose the camera).

You've said that before, but that sidesteps the issue. It's one
thing to recommend a product that you've used, and I can understand
that, but it's quite another to offer proclaim some performance
advantage that you probably have no real data for, just to counter
someone else's opinion about another product. And if you only want
to recommend products that you've used, it's only fair to avoid
disparaging products that you haven't used.


I am sure you will have reports on the Internet where come DSLR
lenses are described as having rather less than 3 stops of IS gain,
and I was simply reporting this fact,

Demonstrably untrue. If you said that, there would only have been
agreement. You instead rejected the statement that some lenses have
up to 3 stops of IS gain by disagreeing, saying that some Panasonic
cameras can do that well, but due to having to move more mass, lens
based IS probably doesn't do as well as the Panasonics.


Note that both the Panasonic system and DSLR-lens systems are
in-lens IS, with moving elements and not otherwise as you may have
implied.

Nothing was implied. Panasonic could have used gyroscopes for all
that I care. It doesn't have replaceable lenses so as far as the
end user is concerned, it's a black box issue. So what are we to
think, that since Panasonic uses a similar lens based IS technique,
and that IS lenses for DSLRs are typically larger, that Panasonic
cameras have an inherent advantage and their IS does better than the
IS employed by DSLR lenses? Remember now, you've already mentioned
that you're not aware of any standardized IS tests, so if that's
what you think, it can only be a guess. But even if it is a guess,
it would be nice if you explicitly state what you've been implying.

I am simply relaying reports on the Web that some IS lenses do not do as
well as three stops of gain, and that my own tests have shown that other
lenses /do/ achieve three stops. That's all.

Why not comment on why that might be, rather than nit-picking every
syllable I write - I don't think what I regard as a casual conversation
can be that important!

G

David