Advantage Full Frame???

[sam]

Sounds lot of techie crap...

Disagree!

I am no techie but I am very thankful that members like amitk26 spend so much time & effort to make it easy for me to understand a bit about such complex things related to photography today.

I will probably never read a book or do much research about this subject and therefore I appreciate this 'spoonfeeding'!

Thanks to amitk26 and all the other contributors!!

Regards

-Sam

[Arun B]

Thanks Amit for the education - it solves lot of questions, and, initiates to read more. 

[aniltulsi]

Amit, thanks for the elaborate explanations, and I appreciate the time spent by you to explain the technological developments. So as explained by you, the better image quality if FF is due to:

1.   Newer fabrication technology (32nm, 22nm)
2.   Arrangement and packing of pixels on the sensor
3.   Interpolation of physical pixels to get photographic pixel
4.   Noise reduction algorithm

Unfortunately, it only explains the technological developments, and does not explain the following:

1.   D700 and D300 were launched close to each other, so the sensor technology was at same level of sophistication. Still D700’s image quality is far superior to that of D300.
2.   D7000 was launched 2 years after D700, with all the developments in the sensor technology. Still the D7000’s image can not match D700’s, or even old 5D’s quality, launched is 2005, with a gap of 5 years.

So size does matter 

I do have issue when I observe the pixel level clarity of D800 as compared to 5DM3 or D4. As these are at same level of technological level, I conclude that individual pixel size does impact the image quality. I still doubt the quality enhancement of D800 over D700. I’ll wait for Anirban to confirm. DxO uses down-sampling, so does not give 'real' pixel level comparison.

Then I go back to my original issue. If the size does matter, why is the difference between D7000 and D800 quality, if the pixel size is same? The gap of one year has brought even better processes? Can the technology development in 3 years bring in such huge enhancement for D800 to exceed D700’s quality with 3 times the pixel density? Which is that specific technology (none of the above tech are good enough to justify this)?

I am sorry if I am not able to understand some basics you might be telling, but I am not convinced.

[amitk26]

the better image quality if FF is due to:

1.   Newer fabrication technology (32nm, 22nm)

Well what this means is a FF sensor can pack more pixels by reducing the empty space between the pixels by using smaller fabrication technology ( 1/3 compared to 65nm and 90nm). This also means less power consumption and less heat and thus less thermal noise.

2.   Arrangement and packing of pixels on the sensor

There is not much change in arrangement though in terms of Bayer pattern , but the gap between pixels is reduced definatly to take care of the not catching photon issue you had.

3.   Interpolation of physical pixels to get photographic pixel
4.   Noise reduction algorithm


Unfortunately, it only explains the technological developments, and does not explain the following:

1.   D700 and D300 were launched close to each other, so the sensor technology was at same level of sophistication. Still D700’s image quality is far superior to that of D300.

Even if sensor is same there is no compulsion on a manufacturer to pack all the features and IP blocks in the silicon and kill product
differentiation.  As an example Canon 7D has similar 18MP sensor to cheaper cams but it does not mean they will provide put same IP in the hardware , Signal processing can be kept different to differentiate a high end product.

2.   D7000 was launched 2 years after D700, with all the developments in the sensor technology. Still the D7000’s image can not match D700’s, or even old 5D’s quality, launched is 2005, with a gap of 5 years.

So size does matter 

Now goal post is shifted you have turned focus to overall sensor size , but if you compare the similar grade FF sensor of 2005 to 2012
definitely the factors mentioned above have changed. Take case of 5D vs 5DMKIII .
In case of APS-C the overall surface area is less and you are not checking the images with same level of blowup.
Crop a portion of image equivalent to D700 from D7000 blow up to 100% and check.

Which is that specific technology (none of the above tech are good enough to justify this)?

Is that so ?

All the technologies described above may be actually collective work of thousands of engineers spread across multiple companies in the food chain of vendors.

It is just like saying a high end audio system just gives low noise , flat response curve from 20Hz to 20Kz and introduce no distortion in Amplification and playback but these 4 things are not sufficient to explain why it sounds good.

[aniltulsi]

Thanks Amit for the explanation. I do agree these technologies are important, but if they can improve the quality to this extent is difficult, looking at the technology level we are.

However, I just checked this Comparometer:

http://www.imaging-resource.com/IMCOMP/COMPS01.HTM

Compare the image quality of D800, D700, D7000 and 5DM3 at ISO-3200 at 100%. My observations are:

1. D800 do not match the image quality of D700 and 5DM3
2. D800's image quality is similar to D7000's, but image size is bigger.

So my initial assumption that D800's image quality was superior to others was wrong, on which assumption this thread was based on. And this means my original belief that larget pixel is superior prevails.

[neocore]

So my initial assumption that D800's image quality was superior to others was wrong, on which assumption this thread was based on. And this means my original belief that larget pixel is superior prevails.

It still is superior today, unless you dont shoot high iso's. DR is much higher, excellent for Landscape photographers(But nothing to warrant a upgrade from d700 as a sensible buyer) .

How often do we shoot more than iso 1600?(This question haunts my mind, 90% of time we judge IQ by noise why is it so).

[aniltulsi]

Yes neocore, I agree that the dynamic range is better. The pixel quality is due to two reasons:

1. Electric Charge retention capacity (genetared by Photons), whic gives the dynamic range.
2. Number of photons collected (depending on area), giving the high ISO performance.

Amit has explaned is details about the technology improvements, which does improve the charge retention of each pixel. However, smaller pixels have smaller area, reducing high ISO performance, which I do see in D800.

But you are right. We don't use high ISO most of the times, but it may also be due to the fact that the quality at high ISO is bad. In past, we did not go beyong ISO-400, not we do go till ISO-1600, right? If we start getting better results, we may also start using ISO 6400 or 12800 too.

[neocore]

Yes neocore, I agree that the dynamic range is better. The pixel quality is due to two reasons:

1. Electric Charge retention capacity (genetared by Photons), whic gives the dynamic range.
2. Number of photons collected (depending on area), giving the high ISO performance.

Amit has explaned is details about the technology improvements, which does improve the charge retention of each pixel. However, smaller pixels have smaller area, reducing high ISO performance, which I do see in D800.

But you are right. We don't use high ISO most of the times, but it may also be due to the fact that the quality at high ISO is bad. In past, we did not go beyong ISO-400, not we do go till ISO-1600, right? If we start getting better results, we may also start using ISO 6400 or 12800 too.

But whenever people talk about high iso's, i see their comparisons strictly based on noise. Even those super duper high iso noise free cameras like D4,D700 lose their tonality and DR in high iso's.

For your question of, we go till iso 1600, i will try not to shoot at iso1600 and up, i still shoot at iso 3200+ sometimes for casual work or web use.

To me the ability of high iso is over rated(This will not be the reason for me to upgrade from aps-c to full frame).

P.S: you might know that the more expensive MF leaf80 dont go more than iso 800.

[amitk26]

Thanks Amit for the explanation. I do agree these technologies are important, but if they can improve the quality to this extent is difficult, looking at the technology level we are.

However, I just checked this Comparometer:

http://www.imaging-resource.com/IMCOMP/COMPS01.HTM

Compare the image quality of D800, D700, D7000 and 5DM3 at ISO-3200 at 100%. My observations are:

1. D800 do not match the image quality of D700 and 5DM3
2. D800's image quality is similar to D7000's, but image size is bigger.

So my initial assumption that D800's image quality was superior to others was wrong, on which assumption this thread was based on. And this means my original belief that larget pixel is superior prevails.

D700 image looks better but in what ways ?  In what ways do you see dynamic range in these images ?
Perhaps if there was a picture of some color and grey-scale chart it would have been better comparison.



At low ISO that is 100 the left color white color is almost washed out in D700 but grey with pattern visible in D800 ( Not a sign of higher DR in D700 if lens and settings were same) hair details of the model image are clearer in D800 , I could see better definition of lips.
It appears to me that lens reached it's limit in terms of lp/mm resolution so details near to shirt collier are not coming out so well.
but if you zoom out a bit to go to view of D700 it will look better ( just thought can't be proved)

Both of these are in camera Jpeg so may be color saturation /hue is different.  But notice that from ISO 3200 to ISO 25600 there is very little degradation in image on D800. Where as the CA and Noise on D700 worsens.

If the theory of larger pixel has to prevail at high ISO D700 image should have been better ( As you wrote bigger pixel gives better high ISO performance).
So I am not able to figure what exactly makes image on D700 better.

To me it looks this bigger pixel means better was a meme spread by Nikon folks which is backfiring on them once they have a high MP FF cam.

[aniltulsi]

Exactly neocore, the big pixel providing better ISO does not neccsarily have better dynamic range. Let me explain this. As I said before, the pixel is the combination of micro-lens, filter and sensor. The light gathering capability is provided by the micro-lens, which covers the full pixel area, but the underneath sensor may be small (due to various reasons already discussed), and may not be able to retain all the charge, and which may overflow, reducing the dynamic range. The technology improvement mensioned by Amit is able to increase the area and depth of the underneath sensor, providing better charge retention and thus, better dynamic range. The dynamic range is the difference between the lowest and the highest electric charge the pixel can retain.

At high ISO, the problem is different. The sensor did not have sufficient photons in first place due to low light, and we amplify what ever signal we got to match the normal ISO level (100). So if in good light, you get electric charge from 1 to 100 (symbolic value), in bad light, you may get from 1 to 50 only. Even if you amplify it, both lower and upper value is amplified, reducing the real dynamic range. Add noice to it, this is even worse. So even D4 will not give good dynamic range at high ISO.

[amitk26]

Exactly neocore, the big pixel providing better ISO does not neccsarily have better dynamic range. L

But worse high ISO performance does not give high Dynamic range either , Please look at the left collar of the model in D700 and D800 picture and check why it looks washed out for D700 but greyish with cloth pattern visible in D800 in ISO 100 images ?  The D700 in-camera Jpeg looks more saturated but indication of dynamic range is how well it handles highlights and low-lights.

So how so called big pixels of D700 are translating in to poor DR ?

[aniltulsi]

Amit, the D700's images are less grainy than the D800's. I do see the fall is quality of D700 is large at ISO-25600, but still better than D800's, in terms of graininess. However, the dynamic range of D800 looks better than D700's (look at the difference between black and white of the images at the end, look at the color palet)

If we go back to the pot theory, the large mouth pot will collect more water in given time, but a pot with greater depth can collect more water. So a sensor may have better ISO due to larger size, but may not have better change retention capacity due to underneath sensor not being deep enough. Newer fabrication techneques explained by you increases the size and depth of the CMOS sensor, improving the charge retention, and dynamic range.

[Madan]

Noise at high ISO, less DR,

I did a test shot, at ISO 12800, just for larks.
Pardon the PP, it's a slam-blam affair, but where is the noise ? I don't see it.
It's a D4. 16 MP

How does this image show up wrt details ?

[amitk26]

Amit, the D700's images are less grainy than the D800's. I do see the fall is quality of D700 is large at ISO-25600, but still better than D800's, in terms of graininess. However, the dynamic range of D800 looks better than D700's (look at the difference between black and white of the images at the end, look at the color palet)

But is it not contradicting your initial assertion that so called bigger pixels should have given more dynamic range.
About high ISO in D700 how exactly it better then D800 with so much on CA and blue spots ?

explained by you increases the size and depth of the CMOS sensor, improving the charge retention, and dynamic range.

Sir the charge retention thingy is applicable to CCD and not CMOS.

beauty is in eye of beholder and may be that is why D700 images from site are looking slightly better  to you


Madan : Beautiful Image I do not see any problem with DR or Noise , Practically speaking all these Cams are fantastic.

My aim was just to find truth of this theory of increase in pixels is bad , I feel may be it is useless to have more megapixels and ISO
above certain number but at the same time there is no evidence that it is necessarily bad for normal usage.  I just referred to sample images from website provided by Anil and felt that at ISO 100 D800 has better DR despite alleged smaller pixels and at ISO25600 it has less noise / CA then D700. So I feel that the meme of larger pixel size can be put to rest that's all. This collaborates with the theory I understand. I earnestly believe that moors law has not failed and engineers at Nikon do work for improvements and not worsening .

Physically there is no reason to believe that surface area of individual pixel is = sensor size / number of pixel becuase it also depends on fab technology.

At the same time this development in technology may just have diminishing returns and marginal value addition and so I do not see great need in spending hard earned money but that does not mean one should go and trash new technology without any hard evidence.

I do realize that I have burnt disproportionately high bandwidth on this mundane pov so last post on this topic from my side.
I do not own any of these FF cams and do not aspire to own any either but my stake in this discussion was just from point of view of understanding.

[sam]

Clap, clap, clap!!

Take a bow amitk26!

& thanks aniltulsi for this thread!

So much in so little!

Regards,

-Sam

[amitk26]

Take a bow amitk26!
So much in so little!

Buddy it is some sort of idiosyncrasy to burn bandwidth here on such mundane things,
but when consumers like you pay close to 40K for a latest Smartphone  (or 2 lakh for latest Nikon /Canon cam in this case) and millions of such pieces are sold , A small fraction is used to pay salaries of people like us whose sole purpose of life is to clean up few waveforms here and there and reduce few milliseconds from execution time, squeeze in a new feature here and there.

So it is hard to stand an assertion that it is worse then what was there 5 years back even if the discussion is about another company's product using similar technologies because that is denial of efforts of while ecosystem and community involved.

How useful is another topic and does not bother me ..
A more personal example Is a quad core phone announced this month a of any-use to anyone ? Does a movie making software on a phone of any use ? May or may not be ,  does not bother me .. but can it be worse then a product 2 year back .. No way until the logic provided for such assertion can be validated.

[aniltulsi]

Amit, well, not exactly contradicting, as I had talked about quality and been more concerned about the high iso, and noise than dynamic range, due to my type of photography. Dynamic range is required more by landscape and portrait photographers and high iso by sports and indoor non-flash users, like me.

And the charge retention was more related to the charge built up before it is transfered to the memory after digitization, not the functioning of CMOS or CCD. I should have used the term charge built up, difference between high and low of which gives the dynamic range.

I respect your views, but have my firm belief of relation between qualty (noise at high iso) and pixel size.

Excellent picture Madan. This makes my understanding even stronger that larger pixel is better pixel

[helion]

Is our theory of bigger is better is wrong?

YES

Since a Photon has dual nature both wave and particle the closest thing which comes to size is the wavelength which is 340nm to 740nm for visible light. This is in order of 1000 times less then the pixel size.

So clearly the size of individual pixel can no way a limiting factor to determine the light sensitivity of the sensor.

However there are several other factors such as the CMOS process , Amplification technique and design used , S/N ratio , DSP in the Camera to gather and process the image, Image processing algorithms to generate the RAW image etc.

Since the pixel size would be measured in a dimension perpendicular to the direction of impact of light, I wonder if the longitudinal measurement ie wavelength should factor in while considering this limit.

[Beyond Legacy]

This was me while reading the thread


BTW, a query that came in my mind while reading was; what is the shape of pixels that are present on sensor... Circular, square or any other???

[ISO]

Full frame sensor has the advantage over APS-C sensor, of being of bigger in size, so that each pixel is able to catch more light, which in turn provides better ISO and better dynamic range.

Comparing D700 with D300, both with 12 Mp, the pixel size for each is:
D700: sensor size = 36x24=864 sq mm.
           Pixel size = 864/12000000 = 72 micro sq mm
D300: sensor size = 24x16=384 sq mm.
           Pixel size = 384/12000000 = 32 micro sq mm

So each pixel of D700 is 2.25 times the D300's, to the quality difference makes sense.

Let's do this for D800 (36 Mp) and D7000 (16 Mp):

D800: sensor size = 36x24=864 sq mm.
           Pixel size = 864/36000000 = 24 micro sq mm
D7000: sensor size = 24x16=384 sq mm.
           Pixel size = 384/16000000 = 24 micro sq mm

So each pixel sixe of D800 is exactly same as that of D7000, so where does the quality difference come from?

How come the D800's dynamic range and ISO performance is better than D700's, who's pixel size is three times that of D800?

Is our theory of bigger is better is wrong?

Will we soon see APS-C sensors performing better tha D700?

I agree that technology evolves, and we do see improvements in each generation of sensor, but the pixel size difference between D800 and D700 is huge, and I don't know if any technology that has evolved to bring such enhancement. Does anyone know?

I am totally lost now... 

Hello Anil

If APS-C and FF both have same pixel size then for same FOV FF will have advantage as there will be more information.

Bigger pixels definately have advantage.

www.clarkvision.com/articles/does.pixel.size.matter/