alundeb wrote:
All of this makes sense, except there is no activation threshold. We cannot count individual photons, only measure the average effect on the current, and that effect does not disappear below any threshold.

jorkata wrote:
And what is this thing that they call 'noise floor' then .

alundeb wrote:
The noise level with zero signal.
.


Say the noise floor is equivalent to 10 e- RMS per pixel.

If you then have a photon capture of average 0.1 e- per pixel during one exposure, you measure signal plus noise as 10.1 e- average over the sensor. (This is true even if the quantization is 1 e-)

The signal to noise ratio is 1:100

If the sensor has 36 megapixels, we can resample to say 36 pixels.

The signal will then be 0.1 e- * 1 000 000 = 100 000 e-

The noise will be 10*sqrt(1 000 000) = 10 000 e

The signal to noise ratio has improved to 10:1

jorkata wrote:

The 1DX and 5DIII are made with the exact same technology?.



How do you know that?

But you are ignoring read noise.

Ideally, read noise will scale with pixel size.
And indeed, smaller pixels typically have lower read noise than larger pixels.
The scaling is not perfect, though.

alundeb wrote:
jorkata wrote:

The 1DX and 5DIII are made with the exact same technology?.



How do you know that?

Aah. Very safe, educated assumption, of course.
Canon surely cannot afford two different processes for two so similar sensors of the same generation.
No pixie dust on the 1DX sensor, if that's what you are thinking.
Just larger pixels and slower pixel readout.
The 16 readout channels on the 1DX give you faster overall readout but slower pixel readout - for better draining of the photodiode and hence better SNR.

jorkata wrote:
But you are ignoring read noise.

Ideally, read noise will scale with pixel size.
And indeed, smaller pixels typically have lower read noise than larger pixels.
The scaling is not perfect, though.

I wasn't ignoring it. In the FWC case the effect was negligible. And in the "noise floor" case, there was "only" read noise.

As has been said earlier in this thread, for ultra high ISO dynamic range, larger pixels are indeed better even at normalized resolution. Nobody questions that.

jorkata wrote:

The 1DX and 5DIII are made with the exact same technology?.

alundeb wrote:
How do you know that?

jorkata wrote:
Aah. Very safe, educated assumption, of course.
Canon surely cannot afford two different processes for two so similar sensors of the same generation.
No pixie dust on the 1DX sensor, if that's what you are thinking.
Just larger pixels and slower pixel readout.
The 16 readout channels on the 1DX give you faster overall readout but slower pixel readout - for better draining of the photodiode and hence better SNR.


You don't necessarily need a different process to make different sensor technology:

From CPN Europe on the 1D X:

"The new photodiode construction has resulted in an improved photoelectric conversion rate that gives increased light sensitivity. Combined with improved transistors inside the pixels, the signal-to-noise ratio has been made even better, allowing access to the higher ISO speeds without increasing image noise. "

alundeb wrote: ...If you are talking about a smaller sensor, it requires LOWER ISO to produce an image from the same perspective / subject distance, with the same field of view, the same depth of field and the same shutter speed, because you will use a wider F-stop to get the same physical aperture diameter( and hence DOF) with a shorter lens.

I wasn't talking about a smaller sensor; that would introduce yet another set of variables into the equation as you have shown. I was talking only about sensel size and SNR.

I'm fine with 16. At the sizes I print, it's plenty of resolution. Larger resolution files are nice if you print large all the time, but it's rare I print larger than 24" wide, and even at 36" wide, 16MP files yield very nice prints.

All the larger sizes would do for my photography is take up extra hard drive space.