Full Frame censor: Low light, but NOT high ISO

I've read a lot of comments down the board and elsewhere about FF sensors
having less noise at high ISO in low light, but has anyone got any
experience of the differences between FF and smaller sensors when shooting
long exposures in low light, but still at ISO 100? At times I've had a few
issues with noise with the 400D (colour noise, banding etc), although this
principally only becomes apparent with under-exposures, which are quite easy
to do when guessing exposure lengths! I'm considering at some point in time
upgrading from the 400D (Xti) to a 5D - (maybe not until after the 5DMk2 has
been released), and wonder how the 5D performs in such conditions. I only
ever see mention of its high ISO performance though.

"Woollyzone" wrote in message
...
I've read a lot of comments down the board and elsewhere about FF sensors
having less noise at high ISO in low light, but has anyone got any
experience of the differences between FF and smaller sensors when shooting
long exposures in low light, but still at ISO 100? At times I've had a few
issues with noise with the 400D (colour noise, banding etc), although this
principally only becomes apparent with under-exposures, which are quite
easy to do when guessing exposure lengths! I'm considering at some point
in time upgrading from the 400D (Xti) to a 5D - (maybe not until after the
5DMk2 has been released), and wonder how the 5D performs in such
conditions. I only ever see mention of its high ISO performance though.

Nobody? Perhaps I should have added that Canon/Nikon/Olympus/Pentax/Sony*
(*delete where applicable) cameras and lenses suck/are ****e* (*delete where
applicable depending on which side of the Atlantic you live), to elicit a
response ;-)

"Woollyzone" wrote in
:

I've read a lot of comments down the board and elsewhere about FF
sensors having less noise at high ISO in low light, but has anyone got
any experience of the differences between FF and smaller sensors when
shooting long exposures in low light, but still at ISO 100? At times
I've had a few issues with noise with the 400D (colour noise, banding
etc), although this principally only becomes apparent with
under-exposures, which are quite easy to do when guessing exposure
lengths!

It's also especially easy to do with the 400D (XTi), as the camera's
metering exposes for 120% of the stated ISO, but has headroom enough
(relative to the 5D and most other Canons) for 85% of the stated ISO.

IOW, the 5D's "ISO 100" is really ISO 120 or 125, both in terms of
metering, and in terms of having 3.5 stops of headroom in the most
sensitive RAW color channel (green). The 400D meters for about ISO 120,
but has enough RAW headroom to be metered at ISO 85. This is by DSLR
standards; by the standards of slide film and many P&S digitals and
medium format digital backs, the 5D's ISO 100 has headroom for ISO 60,
and the XTi's, ISO 43.

With my XTi, I set the contrast to minimum, and then the camera's
histogram and flashing indication of clipping are fairly accurate at
showing the clipping of the green channel of the RAW data, if the
highlights are white or green. I expose just short of the clipping or
with just a few tiny areas flashing. The JPEGs embedded in the RAW (or
the JPEGs themselves if you shoot JPEG-only) will be a bit pale often, if
you expose just short of clipping, but they straighten up with a little
bit of simple gamma adjustment (the middle point in a "Levels" tool).

I'm considering at some point in time upgrading from the 400D
(Xti) to a 5D - (maybe not until after the 5DMk2 has been released),
and wonder how the 5D performs in such conditions. I only ever see
mention of its high ISO performance though.

The existing 5D performs well by brute force, mainly because it has 12
MP. It is not very efficient at capturing photons per unit of area (2/3
stop less efficient than the 1Dmk3, 1Dsmk3, 20D, 30D, 40D, Nikon D3, etc)
but it has FF area, and 12MP of resolution. A crop from a 5D the
physical size of the 400D's sensor would be vastly inferior to the full
image from the 400D.

Comparing the full frame from both, at ISO 100, the 5D will have less
noise from the highlights down through the brighter shadows, but the 400D
will actually be slightly less noisy in the very deepest shadows, because
it has less read noise relative to RAW saturation (slightly higher
classic DR). Any 5Dmk2 is likely to have far better DR at ISO 100,
unless the quantum efficiency (ability to capture photons) is improved to
the point where ISO 100 is no longer possible with full highlight
headroom, as happens in the 20D, 30D, and 40D. Canon has traditionally
been more interested in having base ISOs with simple numbers like 50 or
100 than they are with having a base ISO optimized for maximum DR; a very
bad decision, IMO. I think that the base ISO's RAW sensitivity (or
"gain", as it is often called) should be whatever it needs to be (ISO
113; ISO 87, ISO 139, etc) to have RAW saturation just short of photowell
saturation. They could just round it off to the nearest common ISO, as
far as what they call it is concerned, but get the ratios right under the
hood.


--


John P Sheehy

John Sheehy wrote:
"Woollyzone" wrote in
:

I've read a lot of comments down the board and elsewhere about FF
sensors having less noise at high ISO in low light, but has anyone got
any experience of the differences between FF and smaller sensors when
shooting long exposures in low light, but still at ISO 100? At times
I've had a few issues with noise with the 400D (colour noise, banding
etc), although this principally only becomes apparent with
under-exposures, which are quite easy to do when guessing exposure
lengths!

It's also especially easy to do with the 400D (XTi), as the camera's
metering exposes for 120% of the stated ISO, but has headroom enough
(relative to the 5D and most other Canons) for 85% of the stated ISO.

IOW, the 5D's "ISO 100" is really ISO 120 or 125, both in terms of
metering, and in terms of having 3.5 stops of headroom in the most
sensitive RAW color channel (green). The 400D meters for about ISO 120,
but has enough RAW headroom to be metered at ISO 85. This is by DSLR
standards; by the standards of slide film and many P&S digitals and
medium format digital backs, the 5D's ISO 100 has headroom for ISO 60,
and the XTi's, ISO 43.

With my XTi, I set the contrast to minimum, and then the camera's
histogram and flashing indication of clipping are fairly accurate at
showing the clipping of the green channel of the RAW data, if the
highlights are white or green. I expose just short of the clipping or
with just a few tiny areas flashing. The JPEGs embedded in the RAW (or
the JPEGs themselves if you shoot JPEG-only) will be a bit pale often, if
you expose just short of clipping, but they straighten up with a little
bit of simple gamma adjustment (the middle point in a "Levels" tool).

I'm considering at some point in time upgrading from the 400D
(Xti) to a 5D - (maybe not until after the 5DMk2 has been released),
and wonder how the 5D performs in such conditions. I only ever see
mention of its high ISO performance though.

The existing 5D performs well by brute force, mainly because it has
12 MP.

Mainly because it has 160% of the area overall, so there's room for some
inefficiency perhaps and still perform better? Hmm 8MP to 12MP is 150%
so pretty similar size pixels.

It is not very efficient at capturing photons per unit of area (2/3
stop less efficient than the 1Dmk3, 1Dsmk3, 20D, 30D, 40D, Nikon D3, etc)
but it has FF area, and 12MP of resolution. A crop from a 5D the
physical size of the 400D's sensor would be vastly inferior to the full
image from the 400D.

Comparing the full frame from both, at ISO 100, the 5D will have less
noise from the highlights down through the brighter shadows, but the 400D
will actually be slightly less noisy in the very deepest shadows, because
it has less read noise relative to RAW saturation (slightly higher
classic DR). Any 5Dmk2 is likely to have far better DR at ISO 100,
unless the quantum efficiency (ability to capture photons) is improved to
the point where ISO 100 is no longer possible with full highlight
headroom, as happens in the 20D, 30D, and 40D. Canon has traditionally
been more interested in having base ISOs with simple numbers like 50 or
100 than they are with having a base ISO optimized for maximum DR; a very
bad decision, IMO. I think that the base ISO's RAW sensitivity (or
"gain", as it is often called) should be whatever it needs to be (ISO
113; ISO 87, ISO 139, etc) to have RAW saturation just short of photowell
saturation. They could just round it off to the nearest common ISO, as
far as what they call it is concerned, but get the ratios right under the
hood.

Paul Furman wrote in
:

Mainly because it has 160% of the area overall, so there's room for
some inefficiency perhaps and still perform better? Hmm 8MP to 12MP is
150% so pretty similar size pixels.

That's the brute force I spoke of. That's why I said the 5D had less noise
in all ranges but the deepest of shadows.

--


John P Sheehy

"John Sheehy" wrote in message
...
"Woollyzone" wrote in
:

Thanks John, I think I need to keep that post to re-read and refer back to
in order to fully understand all of that!