Another ISO question...

On Jul 16, 4:38 am, "David J Taylor" -this-
part.nor-this-bit.co.uk wrote:
Alan Browne wrote:
RichA wrote:
On Jul 15, 2:43 pm, Alan Browne

Somewhere in one of the Canon sites is an illustration of IS at
work. IIRC a double concave element moves normal to the lens axis
(x) in y,z to correct for pitch/yaw (but not roll).

Another negative for in-lens I.S. besides cost, the inclusion of
another optical element to make it work. But then maybe that element
is used in the non-I.S. lens as well?

No. And comparable Canon models without IS are a slight bit sharper
than the IS version. Google me and IS sharpness for links.

.. but are both not "sharp enough" for typical use?

David

I think typical use as a phrase should end when a lens costs over
$1000.00.
At that point, you would likely be concerned with ultimate image
quality.
But then what good are 2-3 ultimate sharpness shots when you blow
50-60 because you don't have I.S.?

On Jul 16, 5:18 pm, ASAAR wrote:
On Mon, 16 Jul 2007 12:23:11 -0700, RichA wrote:
You can't have it both ways, not having standard tests

Standard tests in the photographic World, what a novel idea. No
more childish "five stars" or "four stars" ratings.

What's responsible for this opinion? Some publication gave a
plastic bodied camera a four or five star rating?

As long as they pay for it, they can get whatever rating they like for
any camera.

On Mon, 16 Jul 2007 15:56:33 -0700, RichA wrote:

What's responsible for this opinion? Some publication gave a
plastic bodied camera a four or five star rating?

As long as they pay for it, they can get whatever rating they like for
any camera.

Oh, I didn't realize that you were also a conspiracy theorist.

BTW, when I was a little tyke, one of my favorite comic characters
was Plastic Man. Did you ever read any? Did you ever have a
traumatic experience while reading a Plastic Man comic book?
Another favorite, although he arrived and then vanished from the
comic book scene long before the internet existed, used to say when
it was time to save the day, "Flame On!" g

David J Taylor wrote:
Alan Browne wrote:
RichA wrote:
On Jul 15, 2:43 pm, Alan Browne
Somewhere in one of the Canon sites is an illustration of IS at
work. IIRC a double concave element moves normal to the lens axis
(x) in y,z to correct for pitch/yaw (but not roll).
Another negative for in-lens I.S. besides cost, the inclusion of
another optical element to make it work. But then maybe that element
is used in the non-I.S. lens as well?
No. And comparable Canon models without IS are a slight bit sharper
than the IS version. Google me and IS sharpness for links.

.. but are both not "sharp enough" for typical use?

Sure. The whole point of IS is to allow freehand shooting at lower
speeds. So for a (very small) loss of sharpness, you get 2 or 3 stops
of slow shutter speed to use... areas where the sharpest lens would fail
to provide sharp images freehand. Further, these features are mostly on
the "better" glass from Canon and Nikon, so you're "way up" the
sharpness curve to start.

Cheers,
Alan

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.

RichA wrote:
[]
I think typical use as a phrase should end when a lens costs over
$1000.00.
At that point, you would likely be concerned with ultimate image
quality.

Agreed.

But then what good are 2-3 ultimate sharpness shots when you blow
50-60 because you don't have I.S.?

Excellent point - this is the approach I take based on years of getting
sometimes blurred 35mm film telephoto shots. For most of the photos I
take, image content is rather more important than the ultimate image
quality, and IS/VR is a boon.

Cheers,
David

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

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

Perhaps industry standards make too many assumptions, then, and need to
be changed.

You can't assume linearity near sensor saturation (nor near the
blackpoint if the data has been clipped to zero at the blackpoint, or the
wrong blackpoint is assumed). You can't assume that the number of
photons measured at 50% RAW saturation is half of full well, or at 100%
RAW saturation.

All of your calculations on your website seem to start with a couple
pieces of data, and extrapolate the rest, making assumptions such as that
there are 4096 RAW levels to be used, when in fact, some DSLRs use as
little as 3000 RAW levels; RAW saturation at the lowest ISO is full well
and is proportional in photon count to other ISOs and levels, etc. That
model is a bit simpler than reality.

More thorough tests are needed.

--


John P Sheehy

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

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

Perhaps industry standards make too many assumptions, then, and need to
be changed.

You can't assume linearity near sensor saturation (nor near the
blackpoint if the data has been clipped to zero at the blackpoint, or the
wrong blackpoint is assumed). You can't assume that the number of
photons measured at 50% RAW saturation is half of full well, or at 100%
RAW saturation.

I don't assume any linearity at any level. I measure it.
In fact if you paid attention to any of my sensor analyses, you would
see plots of departure from a linear model.

All of your calculations on your website seem to start with a couple
pieces of data, and extrapolate the rest, making assumptions such as that
there are 4096 RAW levels to be used, when in fact, some DSLRs use as
little as 3000 RAW levels; RAW saturation at the lowest ISO is full well
and is proportional in photon count to other ISOs and levels, etc. That
model is a bit simpler than reality.

I make no assumption in my testing. I measure it. For example,
the 1D Mark II at ISO 50 maxes (on my 1D II) at (12-bit) DN=3071.
see: http://www.clarkvision.com/imagedetail/evaluation-1d2
If you then look at other ISOs you see I didn't scale any numbers
from the ISO 50 value. For example, read noise does not follow any
scaling trend; each was measured. The gains scaled within measured noise.

Summary: you are making invalid accusations. And to the contrary,
you have not written up nor presented ANY of your actual data,
nor presented any of your methods.
You simply state a number in these newsgroups and then object
if someone disagrees.

I would love to see your data and methods. I don't have time
to evaluate many sensors, and there are only a couple others I
know who are doing the same work (and by the way using the same
methods), so if you are really evaluating sensors, it would sure
be nice to see published methods and results.

More thorough tests are needed.

Perhaps you need to read up on noise sources and methods.

Here are some very nice articles by a top sensor manufacturer:

http://learn.hamamatsu.com/tutorials/noisegraph/

and others at: http://learn.hamamatsu.com/tutorials

Concepts in Digital Imaging Technology
CCD Noise Sources and Signal-to-Noise Ratio
http://learn.hamamatsu.com/articles/ccdsnr.html

and others at: http://learn.hamamatsu.com/articles

Roger

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

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

Perhaps industry standards make too many assumptions, then, and need to
be changed.

You can't assume linearity near sensor saturation (nor near the
blackpoint if the data has been clipped to zero at the blackpoint, or the
wrong blackpoint is assumed). You can't assume that the number of
photons measured at 50% RAW saturation is half of full well, or at 100%
RAW saturation.

I don't assume any linearity at any level. I measure it.
In fact if you paid attention to any of my sensor analyses, you would
see plots of departure from a linear model.

All of your calculations on your website seem to start with a couple
pieces of data, and extrapolate the rest, making assumptions such as that
there are 4096 RAW levels to be used, when in fact, some DSLRs use as
little as 3000 RAW levels; RAW saturation at the lowest ISO is full well
and is proportional in photon count to other ISOs and levels, etc. That
model is a bit simpler than reality.

I make no assumption in my testing. I measure it. For example,
the 1D Mark II at ISO 50 maxes (on my 1D II) at (12-bit) DN=3071.
see: http://www.clarkvision.com/imagedetail/evaluation-1d2
If you then look at other ISOs you see I didn't scale any numbers
from the ISO 50 value. For example, read noise does not follow any
scaling trend; each was measured. The gains scaled within measured noise.

Summary: you are making invalid accusations. And to the contrary,
you have not written up nor presented ANY of your actual data,
nor presented any of your methods.
You simply state a number in these newsgroups and then object
if someone disagrees.

I would love to see your data and methods. I don't have time
to evaluate many sensors, and there are only a couple others I
know who are doing the same work (and by the way using the same
methods), so if you are really evaluating sensors, it would sure
be nice to see published methods and results.

More thorough tests are needed.

Perhaps you need to read up on noise sources and methods.

Here are some very nice articles by a top sensor manufacturer:

http://learn.hamamatsu.com/tutorials/noisegraph/

and others at: http://learn.hamamatsu.com/tutorials

Concepts in Digital Imaging Technology
CCD Noise Sources and Signal-to-Noise Ratio
http://learn.hamamatsu.com/articles/ccdsnr.html

and others at: http://learn.hamamatsu.com/articles

Roger