[Vihmameister]

With regard to noise most current generation cameras today baring the new sony FF 12mp sensor produce the same level of noise when you take photographs with the same DOF and shutter speed. Its only when one format gives you a shallow DOF will you see an advantage in noise performance. For instance the k3 with a F/1.8 lens and a FF with a F/1.8 lens but when you can use the same DOF between formats K3 with a F/2.8 and FF F/4.2 then that low light performance is evenly matched and both cameras would capture the same amount of noise. Now with the canon sensors you lose DR but as far as SNR they are even with just about everyone.

omg...

first 12MP?

K3=20 MP, APS-c
D750 24 MP on 35,x x 24MM FF
6D 20 MP on 36x24 FF
Some sensors are 36 and some even 51 MP
What matters is sensor density, Misunderstanding is that more dense sensor is always superior- not correct. It depends how much data (light) you want to put there -what depends on frame size and lens used.

Now to "each sensor catches same amount" noise sentence. let's make it wood-simple and lets try paint it to red too.
3 people are looking sunset, they have normal sight, determined by doctors- no eye illnesses, no any defects in cornea, iris or whatever in analog receipting organ, it means eye. But eye is not picture processor. Brain is.
Now all three must draw what they see.
Format to draw in 2 cases are A4 paper, but one is given B5 size paper.
Now one of them is artist, uses fine pen and draws quickly. Another is just advanced student and can draw quite well. third is child, not offered nothing but 12 color pencils.
Would say draw equal pictures? Answer is simple no.
The PICTURE processing, ability to filter what is seen and technical skill to represent it differ and a lot.

All three could see same thing, but make different level of "noise" while they draw it back, separate different level of details as they filter them by brain and one must fit what is seen to smaller size - even more difficult to keep details. Got it???
So are the picture handling engines - called picture processors. Same light and same theoretical noise in same conditions, but very different result.

[Flugelbinder]

There's a reason why Nikon's flagship "only" has 16MP, and before that had 12 for years and was used by so many great wedding photographers...

[Ian Stuart Forsyth]

omg...

first 12MP?

K3=20 MP, APS-c
D750 24 MP on 35,x x 24MM FF
6D 20 MP on 36x24 FF
Some sensors are 36 and some even 51 MP
What matters is sensor density, Misunderstanding is that more dense sensor is always superior- not correct. It depends how much data (light) you want to put there -what depends on frame size and lens used.

Now to "each sensor catches same amount" noise sentence. let's make it wood-simple and lets try paint it to red too.
3 people are looking sunset, they have normal sight, determined by doctors- no eye illnesses, no any defects in cornea, iris or whatever in analog receipting organ, it means eye. But eye is not picture processor. Brain is.
Now all three must draw what they see.
Format to draw in 2 cases are A4 paper, but one is given B5 size paper.
Now one of them is artist, uses fine pen and draws quickly. Another is just advanced student and can draw quite well. third is child, not offered nothing but 12 color pencils.
Would say draw equal pictures? Answer is simple no.
The PICTURE processing, ability to filter what is seen and technical skill to represent it differ and a lot.

All three could see same thing, but make different level of "noise" while they draw it back, separate different level of details as they filter them by brain and one must fit what is seen to smaller size - even more difficult to keep details. Got it???
So are the picture handling engines - called picture processors. Same light and same theoretical noise in same conditions, but very different result.

I brought up the new sony 12mp sensor as it has technology that is of the next generation sensor design that is not found in any other format from apsc to MF so we should exclude it from the discussion.
Its total light is the key to noise and to be clear if you are using MF FF or apsc if you have the same DOF and shutter speed between formats then all 3 sensors will have the same total light projected onto them. The only time that a format has better low light performance is when 1 format has a shallower DOF that another format does not have access to.
Studio shot comparison: Digital Photography Review

Now if you look at the comparison above you will see that the K3 when shot at the same DoF as the FF counter parts the FF cameras have the same amount of noise in the final image and you lose the low light shooting ability.
And thus to have this ability in ff you need the selection of lenses that allow you to shoot a shallower DOF
In your test photos above you are shooting at a DOF that any format from apsc to MF has access to. The only time that FF would give you an advantage in low light is for example you were using a apsc 50mm F/1.4 then using a FF 80mm F/1.4 would give you an advantage with regard to noise performance.
Before you bring up in my llink that the K3 was shot at a lower iso than the FF you need to shoot the FF cameras at a iso 2.25 times higher to achieve the same DOF as the apsc format
Very helpful sites you should visit
What is equivalence and why should I care?: Digital Photography Review
Equivalence

[Vihmameister]

It maybe now goes out of topic a bit. Equivalence article is good reading. Did also seek a bit info on that new 12 MP A7S camera and found article, when A7S, A7R and Canon EOS 5D mark iii are compared. (a bit unfair, but still interesting).
OK, one thing is, if we try to measure real sensor performance, calculate it or "predict" one on based equivalence. In this means, those calculations are valuable.
In real life lets take this one:

Its total light is the key to noise and to be clear if you are using MF FF or APS-c if you have the same DOF and shutter speed between formats then all 3 sensors will have the same total light projected onto them. The only time that a format has better low light performance is when 1 format has a shallower DOF that another format does not have access to.

On thing is to compare cameras, another to use one or another equipment with same conditions. Lets take thunderstorm, when sky is already darkened.
You want wide scene here, don't you? Then you want probably have low or no noise on end result and you want to have capture fast enough to capture lightening, if you can get one. Overall scene must not mixed to total gray chocolate mass? And now you have three cameras to use.

Key factor in this case is - you must choose appropriate settings for scene, on short end of your lens, wide aperture and highest noise free ISO, you know it will not make too much grain on it. What does this mean- , to get picture - you go to limits on other technical factors- what predict your usable time-frame, not vice versa. Here, as scene is not still at all, equality does not matter, but shutter speed you can use, still having meaningful result. And FF beats crop sensor usually here hands down, because time needed to have same amount of light captured, noise-free is shorter. In practice, lets look one example of my own: https://www.flickr.com/photos/kriimurohelisedsilmad/14683361464/
Picture is taken as

• ƒ/4.0
• 28.1 mm
• 0.3 sek
• 800 ISO

Camera used K-30. It was best of hundreds of shots -on "long playing" thunder. Do you see limitations? In shorter times -iso noise was too high to get capture for people for who noise does matter.
So ISO 800 was highest i actually could use on my lens and camera. (kit lens) Now imagine i had FF with F4 or even F2.8 lens with usable ISO near 2250, (as both Nikon D750 and Canon 6d do)
Is that possible that i would get much better picture out there? First normheads style answer - we will never know, because it did not never happen. (And thats the real problem, this moment is gone, as long we do not manipulate TIMEflow itself by ourselves or camera).
Second bet -yes result would be different with better FF camera. Maybe not that good, as we see here: Öine äike, Midnight Lightning
but better than my previous sample.
So my main point is -time is money again. You pay money to get equipment witch one allows to take right shot in right time-frame. And that means -duration of the shot is important.
We can expand duration, but for anything still or fast, we need shorter time - and high ISO performance, not only DOF. Lets take now again river between woods? Or waves on dark day? Maybe you can see why "creamy" rivers and sea are so popular? There is not only "artistic", but technical side behind it. People do shots they CAN DO with camera they have.

And, back to topic - articles authors own comment on that:

However, if you did, for whatever reason, choose to shoot two cameras at equivalent focal lengths, apertures and ISOs (and therefore, the same shutter speed), you' d get the same brightness and motion blur and the same depth-of-field. Not necessarily the same noise (as this can depend on sensors, as state in the article

And another one:

It's important to bear in mind that the ISO setting doesn't make that much difference to the noise. It's usually the shutter speed and aperture you use.

Here i agree, that sensor "only" is too limited in terms to compare, so ISO only is also too limited.

But, lets take now one more test comparison into account. On that not so dens 12 MP Sony. (i will not try to make Sony looking down as "inferior" here). High ISO Compared: Sony A7S vs. A7R vs. Canon EOS 5D III: Digital Photography Review
The main articles topic is on - how good is new 12MP Sony in dynamic low light with new sensor -4 EV vs canons -3 and etc, and also one may found, how important is not to oversaturate sensor with light. But look at test charts on sample picture.
Article itself explains methods to get them "equalized" on other conditions. Default test opens on 2 Sony-s but you can switch to Canon vs Sony too.
For me is not interesting what happens on 25000 ISO and up, but vice versa, how do details and overall light, picture as such on different regions look as we go down, till to ISO 100.
The lower you get, the better looks "old and not so fancy" canon's result from that side.They all are quite equal at 3200, but canons clarity wins hands down below that on any ISO. And yes, canon is noisier at 25000.
That clarity on end result, right time-frame, ergonomics, lack of major limitations are just some technical sides to consider if i am seeking as buyer/user. I am not looking for GPS, WIFI, whatever speed cinema frame-rate and other bells we may have.
No optical prism viewfinder - not camera for me. Too heavy (Nikon) may also be show-stopper for me, if similar is available in more ergonomic way (i.e Canon). To small (small and very compact is not again -more easier to handle) again no.

So, knowing all i do know and what i don't, i still will prefer Canon or Nikon current high end over Sony A7S as it is. Maybe for 99% people out there those Sony's are ideal or sufficient on all conditions they use it. As it is said:" it's not hard to protect any opinion. It's hard to have one, specially argument-ed way".

[Flugelbinder]

It's a matter of logic and for what (little) I understand, also physics.
The best analogy I've read about sensor light gathering ability, described a sensor as a lawn filled with buckets and the light as a sprinkler; as the water is scattered around the lawn, if you have big, fat buckets/fotosites (think a ff sensor with 'only 12 or 16 MP), at the end there will be a lot of water inside each bucket. Now if you use smaller buckets, even if you have more buckets/fotosites (a 24MP APS-C, or a 36MP FF), there will be less water inside each bucket.
He was a lot better with words than me, but I believe I got the analogy right...
Again, there's a reason why manufacturers are staying with 16MP (Nikon) or even 12MP (Sony). They already have the MP race going with other bodies; now it's time to (finally) the high-ISO/low-light capability race to get some speed...

[Stagnant]

Scale down the K-3 images to the same resolution as Sony A7S and I suspect / speculate, that until you reach extreme iso values the differences will be negligible. I shot quite some concerts with K-3 and Sigma 18-35 F1,8 and in all, but most extreme cases the low light performance is quite satisfying for my purposes. I normally scale down the resulting images to the height of my monitor (1440 pixels). I did some prints (15 * 20 cm) of images taken at ISO 6400 in dimly lit pubs and they worked fine. Of course you have to process your images accordingly. That's where DxO's PRIME shines.



More from that set : https://flic.kr/s/aHsk9e7BmF .K-3 + Sigma 18-35 F1.8.

[Vihmameister]

i must append... to my previous comment this quote from dp site.

Why is it people keep talking about the "poor" dynamic range of Canon's "outdated" sensors and are unable to use their eyes enough to see the DR of the 5D Mk3, particularly in terms of the highlights, is smoking the Sony? Particularly on those big tripod like things in the distance.

LIKE2
REPLY

Flag as inappropriate



By Rishi Sanyal (10 months ago)
No. Those lights were turned off by the city of Seattle by the time I got to the 5D Mark III. The 5D Mark III has demonstrably lower low ISO DR than either Sony camera, though differences even out at higher ISOs. The A7S has higher ISO performance and higher DR than either camera at higher ISOs.
Perhaps an article is in order that correlates DxO data with visual, real-world results to dispel this lingering belief that somehow entirely valid measurements can be disproved simply by 'using your own eyes'.


LIKE6







By GaryJP (10 months ago)
Then it's not much of a comparison image is it? It happens to be the first part of the image that got magnified on my computer, I also note that other highlight areas (lighted windows) do not show "demonstrably" better DR in the highlight areas either. At least not on this sample. They are actually much of a muchness. Even the Seattle Tower light areas have a little more detail on the 5D. You guys shot the images. I didn't.

Or i fell onto exact same assumption as Gary out there, or that "comparison" on dp is not accurate enough. For "eye" Canon wins both. And i totally agree about lighting in windows, and some arcs on the picture. If all shots are over saturated, then in Sony-s pictures light look still worse for me, despite Sony win on DR.

Stagnant - good shots.

[john5100]

Stagnant - very nice work in such a dim spot

[Stagnant]

Thanks Vihmameister and john5100 !
I do realise, that if I had to print those photos big (bigger than 15 * 20 cm), I would mostly likely end up with unsatisfactory results, and would need a better high iso performance. Yet again most people are unlikely to print at large sizes. All in all, in my experience, K-3 + appropriate lens and enough post processing skills can result in usable images under most shooting conditions.

[Ian Stuart Forsyth]

It maybe now goes out of topic a bit. Equivalence article is good reading. Did also seek a bit info on that new 12 MP A7S camera and found article, when A7S, A7R and Canon EOS 5D mark iii are compared. (a bit unfair, but still interesting).
OK, one thing is, if we try to measure real sensor performance, calculate it or "predict" one on based equivalence. In this means, those calculations are valuable.
In real life lets take this one:

On thing is to compare cameras, another to use one or another equipment with same conditions. Lets take thunderstorm, when sky is already darkened.
You want wide scene here, don't you? Then you want probably have low or no noise on end result and you want to have capture fast enough to capture lightening, if you can get one. Overall scene must not mixed to total gray chocolate mass? And now you have three cameras to use.

Key factor in this case is - you must choose appropriate settings for scene, on short end of your lens, wide aperture and highest noise free ISO, you know it will not make too much grain on it. What does this mean- , to get picture - you go to limits on other technical factors- what predict your usable time-frame, not vice versa. Here, as scene is not still at all, equality does not matter, but shutter speed you can use, still having meaningful result. And FF beats crop sensor usually here hands down, because time needed to have same amount of light captured, noise-free is shorter. In practice, lets look one example of my own: https://www.flickr.com/photos/kriimurohelisedsilmad/14683361464/
Picture is taken as

• ƒ/4.0
• 28.1 mm
• 0.3 sek
• 800 ISO

And this brings me to my point that the format that can shoot with the shallowest DOF is the one that will give you the best low light performance , if both formats can be shot at the same DOF and shutter speed then there is no benefit with regard to lower noise in the image. In your case if you are shooting with FF and let’s say you primary work is done with a 50mm F6.7 1/30 sec iso 102400 Apsc could match the same low light performance as that FF with a 35mm F4.5 1/30 sec iso 46000

Camera used K-30. It was best of hundreds of shots -on "long playing" thunder. Do you see limitations? In shorter times -iso noise was too high to get capture for people for who noise does matter.
So ISO 800 was highest i actually could use on my lens and camera. (kit lens) Now imagine i had FF with F4 or even F2.8 lens with usable ISO near 2250, (as both Nikon D750 and Canon 6d do)

And again I will state that the format that can give you access to lenses with shallower DOF is the one that will give the better low light ability not sensor size. That FF with a F4 lens can be match with a Apsc lens like F2.8 so you would not gain any low light performance going with FF and the same with a FF and the FF F/2.8 lens any F/1.8 on apsc would match that level of noise

Is that possible that i would get much better picture out there? First normheads style answer - we will never know, because it did not never happen. (And thats the real problem, this moment is gone, as long we do not manipulate TIMEflow itself by ourselves or camera).
Second bet -yes result would be different with better FF camera. Maybe not that good, as we see here: Öine äike, Midnight Lightning
but better than my previous sample.
So my main point is -time is money again. You pay money to get equipment witch one allows to take right shot in right time-frame. And that means -duration of the shot is important.
We can expand duration, but for anything still or fast, we need shorter time - and high ISO performance, not only DOF. Lets take now again river between woods? Or waves on dark day? Maybe you can see why "creamy" rivers and sea are so popular? There is not only "artistic", but technical side behind it. People do shots they CAN DO with camera they have.

First you have to understand that iso is not what causes the noise we see in the image per say, Its the total light. any format (using the same level of technology) that can shoot the same DOF and shutter ( this gives us the same total light) as any other format will show the same level of noise


Here I have 2 images one taken with FF and the other using a sensor 4 times smaller than the FF
Both are shot at the same FOV DOF and light intensity ( flash) and viewed at the same output level
If you look closely you will see that both photos contain the same level of noise even thou they are shot at different ISO and it’s the total light that determines the level of noise. Different formats ,Same DOF, same shutter speed, same noise independent of iso. So if you can project the same total light onto any format then the appearance of noise will be the same

And, back to topic - articles authors own comment on that:
And another one: Here i agree, that sensor "only" is too limited in terms to compare, so ISO only is also too limited.

But, lets take now one more test comparison into account. On that not so dens 12 MP Sony. (i will not try to make Sony looking down as "inferior" here). High ISO Compared: Sony A7S vs. A7R vs. Canon EOS 5D III: Digital Photography Review
The main articles topic is on - how good is new 12MP Sony in dynamic low light with new sensor -4 EV vs canons -3 and etc, and also one may found, how important is not to oversaturate sensor with light. But look at test charts on sample picture.
Article itself explains methods to get them "equalized" on other conditions. Default test opens on 2 Sony-s but you can switch to Canon vs Sony too.
For me is not interesting what happens on 25000 ISO and up, but vice versa, how do details and overall light, picture as such on different regions look as we go down, till to ISO 100.
The lower you get, the better looks "old and not so fancy" canon's result from that side.They all are quite equal at 3200, but canons clarity wins hands down below that on any ISO. And yes, canon is noisier at 25000.
That clarity on end result, right time-frame, ergonomics, lack of major limitations are just some technical sides to consider if i am seeking as buyer/user. I am not looking for GPS, WIFI, whatever speed cinema frame-rate and other bells we may have.
No optical prism viewfinder - not camera for me. Too heavy (Nikon) may also be show-stopper for me, if similar is available in more ergonomic way (i.e Canon). To small (small and very compact is not again -more easier to handle) again no.

So, knowing all i do know and what i don't, i still will prefer Canon or Nikon current high end over Sony A7S as it is. Maybe for 99% people out there those Sony's are ideal or sufficient on all conditions they use it. As it is said:" it's not hard to protect any opinion. It's hard to have one, specially argument-ed way".

https://www.pentaxforums.com/forums/169-pentax-full-frame/257389-what-could-d...ml#post3244754
See my post here why the new sony 7s is different than other sensor out there. If Apsc sensor used the same technology as the a7s then we would see the same performance when shot at the same DOF and shutter speed

[Vihmameister]

Here I have 2 images one taken with FF and the other using a sensor 4 times smaller than the FF
Both are shot at the same FOV DOF and light intensity ( flash) and viewed at the same output level
If you look closely you will see that both photos contain the same level of noise even thou they are shot at different ISO and it’s the total light that determines the level of noise. Different formats ,Same DOF, same shutter speed, same noise independent of iso. So if you can project the same total light onto any format then the appearance of noise will be the same

Looks like you are now overlooking equivalence yourself. Also you do not provide EXIF of images, just show some very similar shots- not too much to calculate.
About format and noise is simply not correct. Even total light captured per sensor area sameness will not mean same kind of signal, nor same kind or amount of noise. There are much more factors.

I think you know theory quite well, but for some reason like to focus only on some aspects of it.
For reminding look this simple documents: Digital Camera Image Noise: Concept and Types
Image Noise: Examples and Characteristics
and this Understanding Digital Camera Sensors

IF and only IF different sensors have same pixel size, same density per area, same sensitivity, same micro-lens structure, same image processing algorithms, same kind of demosaicing, same signal amplifiers, same signal to noise ratio, same amount of static noise, same signal filters and etc. There are so much more things that makes them differ.
Only IF this part is absolutely equal, we can discuss equivalence and DOF as factor to ISO speed or noise. But in practice, it is not. Even if manufacturers use same sensor model, it does not mean same performance on different cameras, nor same result on picture.
and even then : Tutorials – The RAW File Format
So stop focusing on DOF and equivalence alone and look whole picture, as much you can, before you make any assumptions how different cameras with same sensor size "can perform".

Different formats ,Same DOF, same shutter speed, same noise independent of iso. So if you can project the same total light onto any format then the appearance of noise will be the same

Incorrect,. If you will consider all process how digital image is actually created, then you may understand why it is incorrect.

See my post here why the new sony 7s is different than other sensor out there. If Apsc sensor used the same technology as the a7s then we would see the same performance when shot at the same DOF and shutter speed

I did, also i did see reaction to post you did:

So, your conclusion is that this technology, and only this, is the factor which affects ISO noise? Nothing else? If you think so, it's OK for me. Even if I don't agree. No more comments from me from now on. I'm not in a mood to debate anymore. Forget what I have said.

My first reaction was quite same, as Jimmy-s, but as i have time enough currently and i (and maybe others too) can every time learn from situations like this one, decided to point out some failures in your understanding. Currently looks like, from one side you do understand that 12MP may not equal to another 12 MP, but you do stop on half way, and tend to limit "equivalence" to only limited set of factors out and in USE in different Cameras and different brands. IF sensor matters alone or it's technology, why would do Sony work with Nikon then? Interesting, isn't it?
I hope, that at the very end - we do not need now dig deeper into every Sony's or other existing sensors technology, to just find out, that small differences here or there, impact end result, besides the fact that camera or format DOF and can be equalized.

[JimmyDranox]

What is a sensor? What is a pixel, from the point of view of his electronical functionality? How it works? What is thermal noise? Why high ISO is noisier?

That are some basic questions. And if someone can answer to this, he cannot say that ISO noise is not dependent of pixel size. At least until a totally new technology of transforming light in electric signal will be discovered.

[FantasticMrFox]

And if someone can answer to this, he cannot say that ISO noise is not dependent of pixel size.

The simple answer is: it isn't, at least technically. The amount of noise present is not dependent on pixel size, but its visibility is. The noise is always the same (for the same sensor size), what changes is the signal-to-noise ratio which is higher for a single pixel the more light it collects due to its larger size. It's the signal-to-noise ratio that determines how visible the noise is, but the total amount of noise is the same.

Or at least that's what I've extracted from several sources. Correct me if I'm wrong.

[Rondec]

What is a sensor? What is a pixel, from the point of view of his electronical functionality? How it works? What is thermal noise? Why high ISO is noisier?

That are some basic questions. And if someone can answer to this, he cannot say that ISO noise is not dependent of pixel size. At least until a totally new technology of transforming light in electric signal will be discovered.

SNR and noise are the same for nearly all sensors of the same size and same generation tech, but dynamic range differs. All of the focus on noise obscures the fact that the biggest problem with shooting at high iso is that dynamic range sucks and even if you have a relatively noiseless image, it isn't going to be a good image -- certainly not one where you can push the shadows at all.

That's where the tech in the A7s is most helpful.

[FantasticMrFox]

... the biggest problem with shooting at high iso is that dynamic range sucks and even if you have a relatively noiseless image, it isn't going to be a good image -- certainly not one where you can push the shadows at all.

Considering that recent Nikon and Sony sensors are virtually ISO-less there is not much of a difference between shooting at high ISOs and pushing shadows - both work on the same principle. The latter would just mean selectively increasing the ISO in the shadows instead of across the whole frame, naturally that won't give great results. There's just a limit to how much a camera can artificially increase a signal from an area that didn't yield a lot of information/light in the first place.