OFFICIAL: Canon 1D X announced

Correct me if I'm wrong guys, but I understood that noise was related to the voltage over the sensor which alters with ISO...higher ISO=more voltage=more noise. This would indicate that a different architecture of the sensor...shallower and gapless microlenses etc...that required a lesser voltage might result in lower noise at any given ISO. Or was I wrongly informed?

Ricardo222 wrote in post #14044896
Correct me if I'm wrong guys, but I understood that noise was related to the voltage over the sensor which alters with ISO...higher ISO=more voltage=more noise. This would indicate that a different architecture of the sensor...shallower and gapless microlenses etc...that required a lesser voltage might result in lower noise at any given ISO. Or was I wrongly informed?

To my knowledge one can't make it quite that simple. In CMOS sensors you have different kinds of noise (shot noice, temporal noice, reset noice, ...)
Try Google on it and you'll find a lot of articles on the subject

Edit - thinking a little bit more...
Actually you could say "higher ISO=more voltage=more noise" in terms of the signal being amplified.
Example: The sensor has a max ISO value of let's say ISO6400. To shot at ISO 12800 you amplify the signal by 2, ISO 25600 amplify by 4, and so on.
This means that when doing this amplification you not only amplify the signal from the photodiods but also the noise.
What gives you good/bad result is the ratio between signal and noice after the amplification

Candersson wrote in post #14045037
To my knowledge one can't make it quite that simple. In CMOS sensors you have different kinds of noise (shot noice, temporal noice, reset noice, ...)
Try Google on it and you'll find a lot of articles on the subject

Edit - thinking a little bit more...
Actually you could say "higher ISO=more voltage=more noise" in terms of the signal being amplified.
Example: The sensor has a max ISO value of let's say ISO6400. To shot at ISO 12800 you amplify the signal by 2, ISO 25600 amplify by 4, and so on.
This means that when doing this amplification you not only amplify the signal from the photodiods but also the noise.
What gives you good/bad result is the ratio between signal and noice after the amplification

Thanks Thomas...that makes it a bit clearer. One things for sure...I think we can assume improvements in IQ from the latest sensors...the question will be open to every purchaser as to what line they feel the need to follow...speed of AF and frame rate or pixel density for high resolution, or the best mix of the two?

Ricardo222 wrote in post #14045477
Thanks Thomas...that makes it a bit clearer. One things for sure...I think we can assume improvements in IQ from the latest sensors...the question will be open to every purchaser as to what line they feel the need to follow...speed of AF and frame rate or pixel density for high resolution, or the best mix of the two?

Very true.
Obviously Canon has chosen, for now with the 1DX, to go the route to develop their current sensor technology with medium number of pixels, high speed and improved ISO.
This requires high processing power (= new generation -> Digic 5+) and of course not to forget the new AF.

There has been a lot of talk on different forums on Canons ability to continue this route and their current manufacturing capabilities.
It's said that with the current setup they are using some old CMOS technology, way behind Sony and the others.
That may be the case with the sensors we see in the current new models 5DIII/1DX, but personally I'm not really sure that these Internet analyzers know exactly what Canon has up their sleeve
Hopefully we will see some new Sensor technology from Canon in the near future
In the meantime I will stay in line for the 1DX, keeping the rest of my setup and do some shooting

Candersson wrote in post #14045632
Very true.
Obviously Canon has chosen, for now with the 1DX, to go the route to develop their current sensor technology with medium number of pixels, high speed and improved ISO.
This requires high processing power (= new generation -> Digic 5+) and of course not to forget the new AF.

There has been a lot of talk on different forums on Canons ability to continue this route and their current manufacturing capabilities.
It's said that with the current setup they are using some old CMOS technology, way behind Sony and the others.
That may be the case with the sensors we see in the current new models 5DIII/1DX, but personally I'm not really sure that these Internet analyzers know exactly what Canon has up their sleeve
Hopefully we will see some new Sensor technology from Canon in the near future
In the meantime I will stay in line for the 1DX and keeping the rest of my setup and do some shooting

Ahhh...that's what I like to hear! Someone going out to do some shooting! I wonder if some of the techie nuts who spend so much time on these gear threads ever have time to go out and actually make real pictures. I will look forward to your impressions of the 1Dx, while I have fun with a 5D3! It would be interesting to compare notes sometime.

I'm really looking forward to the 1DX, but unfortunately we will probably not see it in the stores here in Sweden until late May, early June...

You bet we will be flooded with impressions/tests/comp​arisons as soon as people start receiving the new bodies. It will be great fun!

It's sad to read all this moaning on the new models. Its lack of features, bad ISO, high MP, low MP, slow, big, and what not...
People just don't seem to understand that there will never be a camera for every need, that's just the way it is. Accept it and move on!

I don't buy that Canon's CMOS is behind Sony's and the non-Sony Nikons.

Everybody talks about comparing raws from 5d2, 1d4 and 1Dx. How about we compare Nikon and Canon with no NR first?

uOpt wrote in post #14046492
I don't buy that Canon's CMOS is behind Sony's and the non-Sony Nikons.

Everybody talks about comparing raws from 5d2, 1d4 and 1Dx. How about we compare Nikon and Canon with no NR first?

I have a suspicion this new generation of FF sensors from Nikon/Sony and Canon are going to be very close. Just a hunch but I think the ultimate winner will be this body but not by much.

K6AZ wrote in post #14046611
I have a suspicion this new generation of FF sensors from Nikon/Sony and Canon are going to be very close. Just a hunch but I think the ultimate winner will be this body but not by much.

As they say, am "Betting the farm on it."

Ricardo222 wrote in post #14044896
Correct me if I'm wrong guys, but I understood that noise was related to the voltage over the sensor which alters with ISO...higher ISO=more voltage=more noise. This would indicate that a different architecture of the sensor...shallower and gapless microlenses etc...that required a lesser voltage might result in lower noise at any given ISO. Or was I wrongly informed?

As has already been mentioned, what's happening is that when you shoot at higher ISOs, the signal is being amplified, but so is the noise.

However, it should be made clear what "noise" really is: it is variation in the signal. Random noise is variation that is random, and pattern noise is variation that is not random.

When you look at an object, you're seeing the light coming from the object. But that light is quantized. What you see as a light intensity in a photograph is actually the accumulation of individual photons that were reflected from the object. Now, the object itself is covered by more than one pixel in the final image, and thus by more than one pixel in the sensor, and that means that the lower the total light collected, the greater the amount of variation there is relative to the total. As a result, noise naturally shows up at higher ISO shots independent of the quality of the sensor, simply because the light is quantized.

Now, you'd think that the above would simply result in luminance noise (though there is variation in the number of photons collected by a given pixel, there is no variation in the wavelength of the photons), but the construction of the sensor causes random variations in the perceived colors as well. This is because the sensor is actually a bayer matrix of red, green, and blue filtered photosites, and it is those photosites that collect the light. So you have random variation of how many photons are collected by the reds versus the greens versus the blues, and the end result is both color noise and luminance noise.

This also has implications as far as which color channels see the most noise. Bayer arrays have twice as many green photosites as they do red or blue photosites, which means that the green channel will have less noise than the red or blue channels will. Blue photons are more energetic than red photons, so you tend to get more charge from them than from red photons, as a blue photon is more likely to displace an electron than a red one. This, I suspect, is why the noise in the red channel tends to be worse in Canon cameras than in the green or blue channels.

^^ Phew KC...that is a terrific description of what goes on...thank you for going to that trouble to explain it. I lack the technical knowhow to understand the finer points of it, but I get the gist! I think!

Interesting dissertation...

As it relates to video (and digital images are essentially video that doesn't move), the Green channel is considered the "Luminance" channel in an RGB television signal.

And just to add some more,

The sensor technique/design used by Sony/Nikon is much denser which means that the Analogue/Digital circuits can be placed on the sensor.
The design used by Canon force this to be done outside of the sensor and by that you get more read noise that needs to be handled.
(Another added factor is the method used to read the photodiodes in terms of columns and rows).

For those interested there are some interesting papers available that also bring up the development of the photodiodes.
Sometimes we see the QE factor mentioned and used to compare different sensors (Quantum Efficiency - the amount of photons being received by the photodiode and converted to an electrical signal).

Is there no way to reverse the phase of the noise to cancel it out?

theviper wrote in post #14048296
Is there no way to reverse the phase of the noise to cancel it out?

As you don't have only one type of noise that would most likely be very complicated.

The thing about discussing the sensor construction etc. might be OT but I think it adds a little good knowledge to us all.

One thing it adds is to grasp how complicated it really is.
In retrospect to all the people moaning about this and that