[Nicolas06]

Surely any one sensel does not know or care how many other sensels there are and the number of photons it will gather depends purely on its exposed area, all other things being equal ? The maximum measurable 'signal' will be whatever count the sensel saturates at, which is nothing to do with how many other sensels there are.

Because of how isos are defined, at the same iso setting, because the FF sensor is 2.3 time bigger its capacity has to be at least 2.3 time bigger than the minimal capacity required for the APSC sensor to expose correctly that sensitivity.

There no ways arround that otherwise the FF sensor would clip highlights and would not be able to be considered as being able to handle that iso setting. It would be limited to higher iso sensitivity.

If an FF sensor was only able to store the same light in absolute term as an APSC sensor rated for iso 100, then the minimum base iso of that FF sensor would be iso 230.

Anothing interresting thing is that as you said any photosite doesn't care of other. It's individual capacity can be deduced from its iso rating and size or if your prefer its iso rating can be deduced from its max capacity and size. But you can also have many small photosites or few big ones. This doesn't matter that much as the noise is mostly random having more sample allow to average and discard noise, simulating larger but fewer photosites. This is fundamental and core to the notion of capacity of a photosite. If noise was not random and could not be averaged, a larger capacity photosite would not allow to decrease noise at all. This is all linked together.

---

Last edited by Nicolas06; 07-09-2016 at 04:05 AM.

[Rondec]

Surely any one sensel does not know or care how many other sensels there are and the number of photons it will gather depends purely on its exposed area, all other things being equal ? The maximum measurable 'signal' will be whatever count the sensel saturates at, which is nothing to do with how many other sensels there are.

You have to have a level playing field. A size print you want to make or a certain size megapixel image to which you will normalize all images to compare. DXO Mark has chosen 8 megapixels, but you could just a 30 inch print or a certain viewing size. But this is the only way to compare a 36 megapixel full frame image to 24 megapixel APS-C image to a 16 megapixel APS-C image.

At a pixel level, the K-1 and K5 are exactly the same with regard to dynamic range and SNR. At an image level, the K-1 is better because you aren't going to see the same level of noise until you print a whole size bigger and if you print the same size, the noise will be quite a bit less with the K-1. Dynamic range is more because you don't see noise in the shadows nearly as quickly with the K-1 image at your given size as you will with the K5 (the K5 pixels are going to be magnified a lot more).

If you choose a small enough print/view size then everything is going to look fine up to a pretty high iso level on all formats.

[Nicolas06]

At a pixel level, the K-1 and K5 are exactly the same with regard to dynamic range and SNR.

But because K1 is more advanced, I thing that even a K1 APSC crop get significantly better high iso performance than 5 year old K5.

The areas where K1 is outstanding are resolution, build quality and high iso.

[clackers]

They will not have the same noise. The noise will be significantly lower on the FF because while the light density will be the same, the total amount of light received will be different.

You're saying a lot here, Nicolas, but don't understand the subject, I'm afraid.


"Total light" has nothing to do with it. That is a cult of belief, not facts.


It's because you are looking at the wrong values.


Here are the measurements that matter - K-5 IIs vs D810, because they have the same pixel size.


The SNR 18% for this APS-C camera and that FF one is the *same*.


You can do the following:




Go to Pentax K-5 II vs Nikon D800



Click on Measurements



Click on SNR 18%



Click on Screen



Repeat for Dynamic Range, Tonal Range, Color Sensitivity



For fun, click on ISO Sensitivity

[Nicolas06]

You're saying a lot here, Nicolas, but don't understand the subject, I'm afraid.
"Total light" has nothing to do with it. That is a cult of belief, not facts.
It's because you are looking at the wrong values.
Here are the measurements that matter - K-5 IIs vs D810, because they have the same pixel size.
The SNR 18% for this APS-C camera and that FF one is the *same*.
You can do the following:
Go to Pentax K-5 II vs Nikon D800
Click on Measurements
Click on SNR 18%
Click on Screen
Repeat for Dynamic Range, Tonal Range, Color Sensitivity
For fun, click on ISO Sensitivity

To consider screen or print results to be the most relevant depend of what you typically do with your FF camera:
A) if you always crop your FF to APSC size, but would not crop at all if you used an APSC body instead, then the "screen" measurement is what matter.
B) if you try to get a given framing that rougly cover your sensor area then print score does matter.

Now ask yourself what is more likely to happen for people that took care to buy the lenses that suit their need and they body format and you'll understand why it is funy 5 minutes to speak of screen score and discover than APSC and FF have excatly the same performance it is irrelevant in practice for most.

Myself I didn't buy an APSC body so I would crop to 1" sensor sze, and I think most people didn't buy an FF so they could crop to all their pictures to APSC or why not 1/2.3" size.

[Rondec]

You're saying a lot here, Nicolas, but don't understand the subject, I'm afraid.


"Total light" has nothing to do with it. That is a cult of belief, not facts.


It's because you are looking at the wrong values.


Here are the measurements that matter - K-5 IIs vs D810, because they have the same pixel size.


The SNR 18% for this APS-C camera and that FF one is the *same*.


You can do the following:




Go to Pentax K-5 II vs Nikon D800



Click on Measurements



Click on SNR 18%



Click on Screen



Repeat for Dynamic Range, Tonal Range, Color Sensitivity



For fun, click on ISO Sensitivity

In real world shooting, what do you think you see if you shoot a K-1 at iso 6400 and a K5 at iso 6400? The images should look exactly the same based on these graphs, not? And they would, if I cropped my K-1 image down to K5 size. But if I print an image that is 16 by 20, the K-1 image will look better, guaranteed. More of the same pixels equals better results when printed or viewed at the same size. I don't need DXO Mark graphs to tell me that.

I shoot with a K3 and K5 II and at low iso it they are great, but at iso 1600 and above the K-1 is a lot better. I guess you can figure out the reasons why that might be.

[kh1234567890]

But if I print an image that is 16 by 20, the K-1 image will look better, guaranteed. More of the same pixels equals better results when printed or viewed at the same size.

Surely it will depend on printer resolution - at 20x16 and the usual 300 dpi you'll be upscaling the K-5 image but downscaling the K-1 one with a corresponding decrease in visible noise and an increase in apparent sharpness ?

[Rondec]

Surely it will depend on printer resolution - at 20x16 and the usual 300 dpi you'll be upscaling the K-5 image but downscaling the K-1 one with a corresponding decrease in visible noise and an increase in apparent sharpness ?

Sure. What else would happen?

I would just add that if you never print above 8 by 10 and never shoot above ISO 800 there is probably not a real world difference.

[Nicolas06]

Surely it will depend on printer resolution - at 20x16 and the usual 300 dpi you'll be upscaling the K-5 image but downscaling the K-1 one with a corresponding decrease in visible noise and an increase in apparent sharpness ?

What ever the printing size, the noise will be more magnified on the K5-IIs than on D800 so if you view it from the same distance, the noise will be more visible on the K5-II print.

I fully agree that it will only matter for large print size or high iso setting and would otherwise make no visible difference because we are limited by ou eyes. the printing system could be better, but there no point to make thing our eyes can't see anyway.

If you go past high iso, say 3200, 6400 or 12800, the K1 picture will look noticably better, even at moderate printing size. Like it or not.

Me I don't care because I tend to favor shots taken at low iso. Still I understand that with K1 I can get away with 3200 iso instead of iso 800 and still get good enough quality. That would be nice to have that.

[clackers]

A) if you always crop your FF to APSC size, but would not crop at all if you used an APSC body instead, then the "screen" measurement is what matter.

This proves you don't understand DxO.

I hope you realise that the 'Print' stat DxO show you is not a real measurement like the 'Screen' one is.

It isn't fake as in 'faked moon landings', but it's close.

They did not actually check the SNR, dynamic and tonal ranges of a downsized JPG. How many people have looked at DXOmark over the years, even been influenced to buy something because of its ratings, and not realised that? Or you, in your arguments?

From a purely mathematical point of view they added 3dB advantage for every doubling in the number of pixels, to their 'Screen' results, assuming that everyone shoots a picture of a uniform monocolour background without a subject.

Nothing to do with FF or APS-C. The same amount is added if the additional pixels are on a phone or a medium format camera.

DXOmark's downsampling assumption will have the K-50 being better at this than the Sony A7S, and the K-3 downsampling equal to the Nikon D750 and better than the Canon 5D MkIII. It should not be treated seriously!

Stick to the 'Screen' measurements, though, and you won't go too far wrong.

[clackers]

I shoot with a K3 and K5 II and at low iso it they are great, but at iso 1600 and above the K-1 is a lot better. I guess you can figure out the reasons why that might be.

Yep, it's just the number of pixels in the case of the K5II, I'm happy to report, that helps when you do your noise reduction and downsampling. It's also the size of the pixels in the case of the K-3.

So it's not sensor size - APS-C cameras can have more pixels than FF.

The actual noise at ISO 1600 will be about 26 dB for both the K-5II and K-1. That's on a scale where 0dB is a pixel of noise for every pixel of the subject. Every 3dB gives you a 100% improvement.

[Nicolas06]

This proves you don't understand DxO.

I hope you realise that the 'Print' stat DxO show you is not a real measurement like the 'Screen' one is.

It isn't fake as in 'faked moon landings', but it's close.

They did not actually check the SNR, dynamic and tonal ranges of a downsized JPG. How many people have looked at DXOmark over the years, even been influenced to buy something because of its ratings, and not realised that? Or you, in your arguments?

From a purely mathematical point of view they added 3dB advantage for every doubling in the number of pixels, to their 'Screen' results, assuming that everyone shoots a picture of a uniform monocolour background without a subject.

Nothing to do with FF or APS-C. The same amount is added if the additional pixels are on a phone or a medium format camera.

DXOmark's downsampling assumption will have the K-50 being better at this than the Sony A7S, and the K-3 downsampling equal to the Nikon D750 and better than the Canon 5D MkIII. It should not be treated seriously!

Stick to the 'Screen' measurements, though, and you won't go too far wrong.

Let me see if I understand. You have no proof, no source for what you say. still you go as far as say the measurement are wrong, the theory is wrong and you ignore the practical results that actually match the theory and measurements.

That a solid faith in your own judgement.

When it come to blind faith and denying of actual reality, I don't think it is any productive to discuss more.

[clackers]

Let me see if I understand. You have no proof, no source for what you say. still you go as far as say the measurement are wrong, the theory is wrong and you ignore the practical results that actually match the theory and measurements.

You've a lot of learning to do before talking so frequently and copiously about these matters, Nicolas.


Their only measurements are the screen tab. They are the contents of your RAW file on your SD card, it is the Real World.


You do not understand the methodology of the print tab - it's a postprocessing 'what if', a hypothetical, and they only take into account numbers of pixels - never sensor format size.


My source is DxoMark themselves - what is yours?


Detailed computation of DxOMark Sensor normalization - DxOMark

[noelpolar]

You've a lot of learning to do before talking so frequently and copiously about these matters.

Man, is that an understatement that could apply to so many (mostly while I'm so silent on these matters as well!)

[clackers]

Man, is that an understatement that could apply to so many (mostly while I'm so silent on these matters as well!)

Well, I've learnt so much good stuff these last few years from yourself and all the other members that it's a shame to see misinformation like that Total Light/DxO print graph BS pop up again in a quality forum like this.


And Nicolas, it's only about what you've written, nothing personal ... in fact I'm excited you're getting the FA31 (I don't have one) and am looking forward to your pics! :-)