Why not make the sensor larger?

Ilya Zakharevich wrote in news:f5d3v8$4et$1
@agate.berkeley.edu:

[A complimentary Cc of this posting was sent to
Roger N. Clark (change username to rnclark)
], who wrote in article :
converted. You don't get more photons by multiple reads
of the chip.

??? By reading N times (and recharging the sensel in between), you
can collect N*FullWell photons.

That's what you get when you take different exposures and read them. I
believe the context here was reading the same sensor well exposure multiple
times.

--


John P Sheehy

Neil Harrington wrote:
"Roger N. Clark (change username to rnclark)" wrote in
That is the main factor
that is different from film days. Like you say, the
photons / area from an extended source stays they same,
but the area of the pixel is larger in the larger sensor
camera.

I've got that all right, but the fact remains, for any given ISO and scene
brightness the f-stop and shutter speed remain the same regardless of pixel
size, correct? Otherwise what meaning would the ISO number have?

While for a given scene ISO, scene brightness, and f-stop,
shutter speed remains the same but several things happen
when you change pixel size in the digital camera:
more photons are collected by the camera with the larger
pixels, the capacity of each pixel to store the photons
converted to electrons increases, and the definition
ISO effectively changes (in terms of # photons required).

For example: double the size of the pixel, and the storage
capacity in electrons (converted photons) goes up by about
a factor of 4, and ISO is redefined to be 4 times more
photons. The larger pixel collects 4 times the light
of the smaller pixel for the same f/ratio and exposure time.

You yourself stated the photon density was a constant
for a given f/ratio (true for extended objects; not
diffraction limited objects). So assume you get
1,000 photons in a pixel that is 2-microns on a side.
What do you think happens when the pixel gets enlarged to
4 microns on a side? Answer: the pixel records 4x as many
photons: 4,000 photons.

Roger

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

You don't get more photons by multiple reads
of the chip.

No, but you would get destructively additive read noise, since it is unique
to each read. This doesn't bring the SNR anywhere near 96 dB at the pixel
level, but that's not the kind of thing you'd expect from small pixels,
anyway.

If the read noise were 3 electrons per read, then 4 reads would yield 1.5
electrons; 9 reads would yield 1 electron, etc. The shot noise would
remain as a single read, but the read noise would act like it was a stack
of separate exposures.

Regardless of the validity of the 96 dB claim, multiple reads would help
where things really matter, or where they actually have much room to
improve at practical ISOs (refillable wells would only be useful at very
low ISOs) - with read noise.

John,
I agree. But the point remains that to get 96 dB you need
to convert 65536 photons if you had read noise as low as
1 electron. 2.2-micron pixel wells support on the order
of 5,000 to 10,000 electrons, much less than the minimum
required for the claimed dynamic range.

Roger

Alfred Molon wrote:
In article , says...

Multiple reads are irrelevant.

They are, because they increase the dynamic range.

Yes, multiple reads improve dynamic range by lowering the
read noise. You must still collect enough photons
to cover that dynamic range. For example: if you
can collect a maximum of 100 photons, and you do mutliple
reads so that the read noise is effectively 0.00000001
electron, the dynamic range is still only 100
(not 100/0.00000001).

The point remains that to get 96 dB you need
to convert 65536 photons if you had read noise as low as
1 electron. 2.2-micron pixel wells support on the order
of 5,000 to 10,000 electrons, much less than the minimum
required for the claimed dynamic range.

If you can't collect the required photons, multiple reads are
irrelevant.

Roger

Ilya Zakharevich wrote:
[A complimentary Cc of this posting was sent to
Roger N. Clark (change username to rnclark)
], who wrote in article :
converted. You don't get more photons by multiple reads
of the chip.

??? By reading N times (and recharging the sensel in between), you
can collect N*FullWell photons.

If you expose the pixel again to get more photons, yes, but that
was not what was proposed. Some chips allow you to
do multiple reads to reduce the read noise, but it doesn't
increase the photon count, nor does it change the
Poisson statistics of the photons.

Roger

In article ,
says...
Ilya Zakharevich wrote in news:f5d3v8$4et$1
@agate.berkeley.edu:

[A complimentary Cc of this posting was sent to
Roger N. Clark (change username to rnclark)
], who wrote in article :
converted. You don't get more photons by multiple reads
of the chip.

??? By reading N times (and recharging the sensel in between), you
can collect N*FullWell photons.

That's what you get when you take different exposures and read them. I
believe the context here was reading the same sensor well exposure multiple
times.

With multiple read is meant that you take N exposures, one after the
other, and add them. N varies with each pixel, since pixels receive
different light levels. A new exposure for a pixel starts when this
pixel is full and is discharged.
--

Alfred Molon
------------------------------
Olympus 50X0, 7070, 8080, E3X0, E4X0 and E5X0 forum at
http://tech.groups.yahoo.com/group/MyOlympus/
http://myolympus.org/ photo sharing site

In article , says...
converted. You don't get more photons by multiple reads
of the chip.

??? By reading N times (and recharging the sensel in between), you
can collect N*FullWell photons.

If you expose the pixel again to get more photons, yes, but that
was not what was proposed.

Yes, you would expose the pixel again to collect more light.
--

Alfred Molon
------------------------------
Olympus 50X0, 7070, 8080, E3X0, E4X0 and E5X0 forum at
http://tech.groups.yahoo.com/group/MyOlympus/
http://myolympus.org/ photo sharing site

Alfred Molon wrote:
In article , says...
converted. You don't get more photons by multiple reads
of the chip.
??? By reading N times (and recharging the sensel in between), you
can collect N*FullWell photons.
If you expose the pixel again to get more photons, yes, but that
was not what was proposed.

Yes, you would expose the pixel again to collect more light.

Then you have subject movement causing blur.

Roger

Alfred Molon wrote:
In article ,
says...
Ilya Zakharevich wrote in news:f5d3v8$4et$1
@agate.berkeley.edu:

[A complimentary Cc of this posting was sent to
Roger N. Clark (change username to rnclark)
], who wrote in article :
converted. You don't get more photons by multiple reads
of the chip.

??? By reading N times (and recharging the sensel in between), you
can collect N*FullWell photons.
That's what you get when you take different exposures and read them. I
believe the context here was reading the same sensor well exposure multiple
times.

With multiple read is meant that you take N exposures, one after the
other, and add them. N varies with each pixel, since pixels receive
different light levels. A new exposure for a pixel starts when this
pixel is full and is discharged.

Then you are not improving read noise.
Read noise goes up with added frames,
signal go up linearly, thus you gain by root N.
But you risk subject movement. This becomes an effective
solution for static subjects, and is used in
astrophotography. It will not work in
sports/wildlife action photography, nor in
low light snapshots of people indoors due to
subject movement.

Roger

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

For example: double the size of the pixel, and the storage
capacity in electrons (converted photons) goes up by about
a factor of 4, and ISO is redefined to be 4 times more
photons. The larger pixel collects 4 times the light
of the smaller pixel for the same f/ratio and exposure time.

ISO and flux have nothing to do with photons per pixel, or even photons
captured per unit of area; it has to do with photons available, on the
focal plane, per unit of area.

--


John P Sheehy