"TURN THOUSANDS OF DOLLARS INTO HUNDREDS!!!"

Perhaps at some point what the sensor is able to capture will
exceed the lenses ability to hide its own flaws. This already happens
with high end digital backs.


Any ideas of where in megapixels that theoretically occurs? I realize
it depends on lens quality. I recently stepped up from a 4 mp to a
10.1 mp and cropping rather severely is much sharper and holds
togeather very well IMO.

Luke

Luke wrote:
Perhaps at some point what the sensor is able to capture will
exceed the lenses ability to hide its own flaws. This already happens
with high end digital backs.

Any ideas of where in megapixels that theoretically occurs? I realize
it depends on lens quality. I recently stepped up from a 4 mp to a
10.1 mp and cropping rather severely is much sharper and holds
togeather very well IMO.

The other points about megapixels often missed is that if lens quality
etc. is up to it, they affect picture quality not by quantity
increase, but with respect to the proportional increase, and you need
to square the change in resolution to get the necessary increase in
megapixels.

For example to double the resolution of an image you need four times
as many pixels. So your change from 4mp to 10mp gave you an increase
in theoretical resolution of about 60%. To double resolution you would
have had to go to 16mp. And to make the same image change in
resolution again as you made in shifting from 4mp to 10mp you'd have
to go next to go to 25mp. If the lenses etc. were up to it.

How much smaller a change in image resolution is a shift from 3mp to
6mp compared to shifting from 6mp to 9mp? The first is an improvement
of 41%, the second an improvement of 15%, i.e. the first is very
obvious, and the second would be difficult to see without careful
comparative tests.

--
Chris Malcolm DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]

On Jan 22, 2:25*am, Chris Malcolm wrote:
Luke wrote:
Perhaps at some point what the sensor is able to capture will
exceed the lenses ability to hide its own flaws. This already happens
with high end digital backs.
Any ideas of where in megapixels that theoretically occurs? I realize
it depends on lens quality. I recently stepped up from a 4 mp to a
10.1 mp and cropping rather severely is much sharper and holds
togeather very well IMO.

The other points about megapixels often missed is that if lens quality
etc. is up to it, they affect picture quality not by quantity
increase, but with respect to the proportional increase, and you need
to square the change in resolution to get the necessary increase in
megapixels. *

For example to double the resolution of an image you need four times
as many pixels. So your change from 4mp to 10mp gave you an increase
in theoretical resolution of about 60%. To double resolution you would
have had to go to 16mp. And to make the same image change in
resolution again as you made in shifting from 4mp to 10mp you'd have
to go next to go to 25mp. If the lenses etc. were up to it.

How much smaller a change in image resolution is a shift from 3mp to
6mp compared to shifting from 6mp to 9mp? The first is an improvement
of 41%, the second an improvement of 15%, i.e. the first is very
obvious, and the second would be difficult to see without careful
comparative tests.

--
Chris Malcolm * * * * * * * * * *DoD #205
IPAB, *Informatics, *JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]

Interesting. So it would require a megapixel of 40 to double my 10 mp
camera.

I'm not well educated in this area, and haven't seen it stated,
however wonder if digital cameras pack these megapixel counts into the
exact size of a 35 mm image?

I now use an Olympus e510 and am very happy with the results of the
two lens I've gotten. Since I take a lot of bird pictures I am lusting
for the Zuiko 300mm f2.8 but at aprice of about $ 6000.00 it is just a
dream. You may know. Are there possible great gains to be made in
future lens quality AND megapixel resolution coming?

Luke

Luke wrote:

So it would require a megapixel of 40 to double my 10 mp camera.
^
the resolution of
Obviously!

I'm not well educated in this area, and haven't seen it stated,
however wonder if digital cameras pack these megapixel counts into the
exact size of a 35 mm image?

No, most do not.

I now use an Olympus e510

That one uses a sensor size a quarter of the 35mm format.

Are there possible great gains to be made in future lens quality

Only incrementlly, like better coating, less flare/ghosting,
better image stabilisation.

AND megapixel resolution coming?

Nope. We are near the top of the physically possible
already. There's only so many things one can do not to loose
and to convert photons.
http://www.clarkvision.com/imagedeta...hotons.and.qe/
shows that even the old 10D and the 50mm f/4 (17 elements) +
1.4x TC (another 5 elements) captures 9.5% of all photons in the
middle of the green band.

Even if we assume a 100% perfect lens (no light loss at all) and a
100% perfect sensor, less than 4 stops of gain will be possible.
That means you cannot even double the resolution[1] of a camera
with the same sensor size even with perfect and only theoretically
possible lenses and sensors without recording less photons than
we do today, i.e. induce more noise[2].

What you could try is increasing the sensor wells (the area holding
the electrons the captured photons are turned into) by a huge
amount, thus reducing the base sensitivity of the sensor. So where
you now have ISO 100 or 200 as lowest settings, you would be able
to use ISO 25 or 12. Larger sensor wells mean larger sample sizes
(statistically) --- by turning an ISO 100 into ISO 12 you'd reduce
the noise by a factor of 2.8 (for a cost of needing 8(!) times
the light). I don't see people in general wanting to go back to
very slow sensors, though, just to have some more resolution.


The other way to deal with the problem is to use larger
sensor sizes. They, of course, need larger optics --- and
tend to have a small DOF, so you want a smaller aperture,
which means less light, which ... you see?

-Wolfgang

[1] Any pixel would have to be turned into 2 x 2 pixels, each
only a quarter of the original pixel size (actually less,
the borders between pixels also take space, but are blind).

[2] Photons are emitted randomly, only conforming to a statistical
average. Our sensors are already photon noise limited ---
it's *the* dominant noise in all but the darkest areas of
an image. And there's no way of getting around that one.

Luke wrote:
On Jan 22, 2:25?am, Chris Malcolm wrote:

The other points about megapixels often missed is that if lens quality
etc. is up to it, they affect picture quality not by quantity
increase, but with respect to the proportional increase, and you need
to square the change in resolution to get the necessary increase in
megapixels. ?

For example to double the resolution of an image you need four times
as many pixels. So your change from 4mp to 10mp gave you an increase
in theoretical resolution of about 60%. To double resolution you would
have had to go to 16mp. And to make the same image change in
resolution again as you made in shifting from 4mp to 10mp you'd have
to go next to go to 25mp. If the lenses etc. were up to it.

How much smaller a change in image resolution is a shift from 3mp to
6mp compared to shifting from 6mp to 9mp? The first is an improvement
of 41%, the second an improvement of 15%, i.e. the first is very
obvious, and the second would be difficult to see without careful
comparative tests.

Interesting. So it would require a megapixel of 40 to double my 10 mp
camera.

I'm not well educated in this area, and haven't seen it stated,
however wonder if digital cameras pack these megapixel counts into the
exact size of a 35 mm image?

Most digital cameras pack them into a smaller frame, the degree of
reduction in size being known as the "crop factor".

I now use an Olympus e510 and am very happy with the results of the
two lens I've gotten. Since I take a lot of bird pictures I am lusting
for the Zuiko 300mm f2.8 but at aprice of about $ 6000.00 it is just a
dream. You may know. Are there possible great gains to be made in
future lens quality AND megapixel resolution coming?

It's soon going to be possible to pack enough pixels into a 35mm film
sized sensor to achieve the resolution of a medium format film camera
such as a Hasselblad. Most of today's SLR and DSLR lenses aren't good
enough to fully exploit that kind of resolution, but we do now have
the technology to be able make good enough lenses. That will create a
technological opportunity to produce digital cameras with medium
format image quality and 35mm film SLR camera size.

--
Chris Malcolm DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]

On Jan 23, 2:47*am, Wolfgang Weisselberg
wrote:
Luke wrote:
So it would require a megapixel of 40 to double my 10 mp camera.

* * * * * * * * * * * * * * * * * * * * * * * * *^
* * * * * * * * * * * * * * * * * * *the resolution of
Obviously!

I'm not well educated in this area, and haven't seen it stated,
however wonder if digital cameras pack these megapixel counts into the
exact size of a 35 mm image?

No, most do not.

I now use an Olympus e510

That one uses a sensor size a quarter of the 35mm format.

Are there possible great gains to be made in future lens quality

Only incrementlly, like better coating, less flare/ghosting,
better image stabilisation.

AND megapixel resolution coming?

Nope. *We are near the top of the physically possible
already. *There's only so many things one can do not to loose
and to convert photons. *
* *http://www.clarkvision.com/imagedeta...hotons.and.qe/
shows that even the old 10D and the 50mm f/4 (17 elements) +
1.4x TC (another 5 elements) captures 9.5% of all photons in the
middle of the green band.

Even if we assume a 100% perfect lens (no light loss at all) and a
100% perfect sensor, less than 4 stops of gain will be possible.
That means you cannot even double the resolution[1] of a camera
with the same sensor size even with perfect and only theoretically
possible lenses and sensors without recording less photons than
we do today, i.e. induce more noise[2].

What you could try is increasing the sensor wells (the area holding
the electrons the captured photons are turned into) by a huge
amount, thus reducing the base sensitivity of the sensor. *So where
you now have ISO 100 or 200 as lowest settings, you would be able
to use ISO 25 or 12. *Larger sensor wells mean larger sample sizes
(statistically) --- by turning an ISO 100 into ISO 12 you'd reduce
the noise by a factor of 2.8 (for a cost of needing 8(!) times
the light). *I don't see people in general wanting to go back to
very slow sensors, though, just to have some more resolution.

The other way to deal with the problem is to use larger
sensor sizes. *They, of course, need larger optics --- and
tend to have a small DOF, so you want a smaller aperture,
which means less light, which ... you see?

-Wolfgang

[1] Any pixel would have to be turned into 2 x 2 pixels, each
* * only a quarter of the original pixel size (actually less,
* * the borders between pixels also take space, but are blind).

[2] Photons are emitted randomly, only conforming to a statistical
* * average. *Our sensors are already photon noise limited ---
* * it's *the* dominant noise in all but the darkest areas of
* * an image. *And there's no way of getting around that one.

I understand perhaps 50% of what you have given me here but thank you
for giving me that much. If I understand you correctly we will be
getting improvements on into the future but in much more limited
increments than in the past. I look forward to responses from others
who may have better understanding than I.

I will now be more careful in my lens pruchases so as to hopefully be
buying lens that will have some useful life with the improvements in
resolution to come.

Luke

Wolfgang Weisselberg wrote:

What you could try is increasing the sensor wells (the area holding
the electrons the captured photons are turned into) by a huge
amount, thus reducing the base sensitivity of the sensor. So where
you now have ISO 100 or 200 as lowest settings, you would be able
to use ISO 25 or 12. Larger sensor wells mean larger sample sizes
(statistically) --- by turning an ISO 100 into ISO 12 you'd reduce
the noise by a factor of 2.8 (for a cost of needing 8(!) times
the light). I don't see people in general wanting to go back to
very slow sensors, though, just to have some more resolution.

Could you explain this? Thicker silicon?

The other way to deal with the problem is to use larger
sensor sizes.

Paul Furman wrote:
Wolfgang Weisselberg wrote:

What you could try is increasing the sensor wells (the area holding
the electrons the captured photons are turned into) by a huge
amount, thus reducing the base sensitivity of the sensor. So where
you now have ISO 100 or 200 as lowest settings, you would be able
to use ISO 25 or 12. Larger sensor wells mean larger sample sizes
(statistically) --- by turning an ISO 100 into ISO 12 you'd reduce
the noise by a factor of 2.8 (for a cost of needing 8(!) times
the light). I don't see people in general wanting to go back to
very slow sensors, though, just to have some more resolution.

Could you explain this? Thicker silicon?

The only way to combat photon noise is larger wells. *How*
they are going to build larger wells is a technical problem ---
though people have been working on building good 3d items in
silicon for years and years.

-Wolfgang