D3s Sensor Data - relative to D3

Wouldn't you know it: The first thing Marianne does with her new D3s, is to take a bunch of photos of a blank wall!

I'm glad that I did, though, because it's yielded some interesting data to share with everyone who's curious about how Nikon endowed the D3s with its high-ISO performance. I hope you will find this report to be worth my staying up until 1am to finish the set through ISO 12800.

Base ISO

This was the first check to make. If the new sensor has much improved quantum efficiency, this could result in the base ISO rising, so I wanted to make sure that ISO 200 is still a "real" ISO. The check came out good; just as with the D3, ISO values below 200 are artificial, but from 200 and up, the RAW data scales linearly with ISO setting, given a constant exposure.

Full Well Capacity and Quantum Efficiency

I re-tested both of my D3 bodies, plus the new D3s, for this - just to make sure I produced a valid comparison. For some obscure reason - sunspots or moon phase or other strangeness - photons are behaving better today, and I achieved higher FWC results for my D3's than I have before. Because of this discrepancy, I am only going to report relative performance between the D3s and D3, instead of giving absolute measurements.

To make a long story short, I found a 37% increase in quantum efficiency for the D3s, i.e., given the same exposure, the D3s collects that many more photons in each sensel. That is almost 1/2 stop improvement. Along with that increase, the full well capacity has gone up by just about the same factor, which is the reason why the D3s keeps the same base ISO.

The Q.E. improvement will give the D3s that 1/2 stop advantage in shot noise across the entire ISO range, and across the tonal range as well. It isn't a huge change, but I would say it is still important, as it does affect apparent noise levels in the entire image.

Color Filter Array

So far, the data I have suggests that the color filters haven't changed. At least, I know the R:G:B ratios seen by the D3s for my light source are essentially identical to what the D3 sees. Additional testing with various subject colors will be needed for confirmation that the CFA is indeed the same.

Low ISO Performance

At low ISO, the camera's DR is limited by the A/D converter S/N ratio. It appears that the same converters are being used, as I find no difference in DR from ISO 200-400.

Higher ISO Settings

As we proceed up the ISO scale from 800, the D3s starts to acquire a substantial advantage over the D3, due to improvements in read noise. The behavior of the D3 with regard to read noise, is that it scales with ISO, from 800 and up; this is equivalent to a constant read noise measured in electrons.

The D3s design is a significant departure in its behavior as gain is increased for the higher ISO settings. As gain rises through the ISO 800-6400 range, read noise referred to electrons gradually decreases, and at 6400, it is almost down to half of the D3 noise level. This improvement, combined with the 1/2 stop gain in shot noise performance, results in a DR advantage of up to 1.5 stops for the D3s, compared to the D3.

Here is how the D3s DR compares to that of the D3 over this ISO range:

• ISO 800 - D3s +0.3 stops

• ISO 1600 - D3s +0.7 stops

• ISO 3200 - D3s +1.2 stops

• ISO 6400 and 12800 - D3s +1.5 stops

Above ISO 6400, D3s read noise is holding constant in electron units, which means it will show a steady DR decrease of one stop, for each one-stop increase in ISO (which is how the D3 behaves from ISO 800 and up).

In summary, the D3s shows modest improvement at low ISO settings due to the 1/2 stop reduction in shot noise (which however does not improve the noise level in blacks below ISO 800). There is moderate improvement in the ISO 800-1600 range, and substantial improvement - more than a full stop - from ISO 3200 and up.

I suppose I can live with that!

Just one Q:
how can more captured photons give less shot noise?

Marianne Oelund wrote:

Here is how the D3s DR compares to that of the D3 over this ISO range:

• ISO 800 - D3s +0.3 stops

• ISO 1600 - D3s +0.7 stops

• ISO 3200 - D3s +1.2 stops

• ISO 6400 and 12800 - D3s +1.5 stops

Above ISO 6400, D3s read noise is holding constant in electron units, which means it will show a steady DR decrease of one stop, for each one-stop increase in ISO (which is how the D3 behaves from ISO 800 and up).

In summary, the D3s shows modest improvement at low ISO settings due to the 1/2 stop reduction in shot noise (which however does not improve the noise level in blacks below ISO 800). There is moderate improvement in the ISO 800-1600 range, and substantial improvement - more than a full stop - from ISO 3200 and up.

Looks to be a very nice and worthwhile improvement on an excellent camera to begin with, especially for an "s" upgrade. I can see where these benefits will be very important for photographers who are constantly shooting at ISO 1600 and above. The new improvements allow the D3s to compliment the D3x even more so than the D3 did. Thanks for sharing your results.

Bill

Marianne Oelund wrote:

Wouldn't you know it: The first thing Marianne does with her new D3s, is to take a bunch of photos of a blank wall!

I'm glad that I did, though, because it's yielded some interesting data to share with everyone who's curious about how Nikon endowed the D3s with its high-ISO performance. I hope you will find this report to be worth my staying up until 1am to finish the set through ISO 12800.

Base ISO

This was the first check to make. If the new sensor has much improved quantum efficiency, this could result in the base ISO rising, so I wanted to make sure that ISO 200 is still a "real" ISO. The check came out good; just as with the D3, ISO values below 200 are artificial, but from 200 and up, the RAW data scales linearly with ISO setting, given a constant exposure.

Full Well Capacity and Quantum Efficiency

I re-tested both of my D3 bodies, plus the new D3s, for this - just to make sure I produced a valid comparison. For some obscure reason - sunspots or moon phase or other strangeness - photons are behaving better today, and I achieved higher FWC results for my D3's than I have before. Because of this discrepancy, I am only going to report relative performance between the D3s and D3, instead of giving absolute measurements.

To make a long story short, I found a 37% increase in quantum efficiency for the D3s, i.e., given the same exposure, the D3s collects that many more photons in each sensel. That is almost 1/2 stop improvement. Along with that increase, the full well capacity has gone up by just about the same factor, which is the reason why the D3s keeps the same base ISO.

The Q.E. improvement will give the D3s that 1/2 stop advantage in shot noise across the entire ISO range, and across the tonal range as well. It isn't a huge change, but I would say it is still important, as it does affect apparent noise levels in the entire image.

Color Filter Array

So far, the data I have suggests that the color filters haven't changed. At least, I know the R:G:B ratios seen by the D3s for my light source are essentially identical to what the D3 sees. Additional testing with various subject colors will be needed for confirmation that the CFA is indeed the same.

Low ISO Performance

At low ISO, the camera's DR is limited by the A/D converter S/N ratio. It appears that the same converters are being used, as I find no difference in DR from ISO 200-400.

Higher ISO Settings

As we proceed up the ISO scale from 800, the D3s starts to acquire a substantial advantage over the D3, due to improvements in read noise. The behavior of the D3 with regard to read noise, is that it scales with ISO, from 800 and up; this is equivalent to a constant read noise measured in electrons.

The D3s design is a significant departure in its behavior as gain is increased for the higher ISO settings. As gain rises through the ISO 800-6400 range, read noise referred to electrons gradually decreases, and at 6400, it is almost down to half of the D3 noise level. This improvement, combined with the 1/2 stop gain in shot noise performance, results in a DR advantage of up to 1.5 stops for the D3s, compared to the D3.

Here is how the D3s DR compares to that of the D3 over this ISO range:

• ISO 800 - D3s +0.3 stops

• ISO 1600 - D3s +0.7 stops

• ISO 3200 - D3s +1.2 stops

• ISO 6400 and 12800 - D3s +1.5 stops

Above ISO 6400, D3s read noise is holding constant in electron units, which means it will show a steady DR decrease of one stop, for each one-stop increase in ISO (which is how the D3 behaves from ISO 800 and up).

In summary, the D3s shows modest improvement at low ISO settings due to the 1/2 stop reduction in shot noise (which however does not improve the noise level in blacks below ISO 800). There is moderate improvement in the ISO 800-1600 range, and substantial improvement - more than a full stop - from ISO 3200 and up.

I suppose I can live with that!

While I didn't understand everything in your short essay, I would like to thank you for your efforts in providing some good information to the community.

Hats off.

R

Marianne Oelund wrote:

Wouldn't you know it: The first thing Marianne does with her new D3s, is to take a bunch of photos of a blank wall!

I'm glad that I did, though, because it's yielded some interesting data to share with everyone who's curious about how Nikon endowed the D3s with its high-ISO performance. I hope you will find this report to be worth my staying up until 1am to finish the set through ISO 12800.

Base ISO

This was the first check to make. If the new sensor has much improved quantum efficiency, this could result in the base ISO rising, so I wanted to make sure that ISO 200 is still a "real" ISO. The check came out good; just as with the D3, ISO values below 200 are artificial, but from 200 and up, the RAW data scales linearly with ISO setting, given a constant exposure.

Full Well Capacity and Quantum Efficiency

I re-tested both of my D3 bodies, plus the new D3s, for this - just to make sure I produced a valid comparison. For some obscure reason - sunspots or moon phase or other strangeness - photons are behaving better today, and I achieved higher FWC results for my D3's than I have before. Because of this discrepancy, I am only going to report relative performance between the D3s and D3, instead of giving absolute measurements.

To make a long story short, I found a 37% increase in quantum efficiency for the D3s, i.e., given the same exposure, the D3s collects that many more photons in each sensel. That is almost 1/2 stop improvement. Along with that increase, the full well capacity has gone up by just about the same factor, which is the reason why the D3s keeps the same base ISO.

The Q.E. improvement will give the D3s that 1/2 stop advantage in shot noise across the entire ISO range, and across the tonal range as well. It isn't a huge change, but I would say it is still important, as it does affect apparent noise levels in the entire image.

Color Filter Array

So far, the data I have suggests that the color filters haven't changed. At least, I know the R:G:B ratios seen by the D3s for my light source are essentially identical to what the D3 sees. Additional testing with various subject colors will be needed for confirmation that the CFA is indeed the same.

Low ISO Performance

At low ISO, the camera's DR is limited by the A/D converter S/N ratio. It appears that the same converters are being used, as I find no difference in DR from ISO 200-400.

Higher ISO Settings

As we proceed up the ISO scale from 800, the D3s starts to acquire a substantial advantage over the D3, due to improvements in read noise. The behavior of the D3 with regard to read noise, is that it scales with ISO, from 800 and up; this is equivalent to a constant read noise measured in electrons.

The D3s design is a significant departure in its behavior as gain is increased for the higher ISO settings. As gain rises through the ISO 800-6400 range, read noise referred to electrons gradually decreases, and at 6400, it is almost down to half of the D3 noise level. This improvement, combined with the 1/2 stop gain in shot noise performance, results in a DR advantage of up to 1.5 stops for the D3s, compared to the D3.

Here is how the D3s DR compares to that of the D3 over this ISO range:

• ISO 800 - D3s +0.3 stops

• ISO 1600 - D3s +0.7 stops

• ISO 3200 - D3s +1.2 stops

• ISO 6400 and 12800 - D3s +1.5 stops

Above ISO 6400, D3s read noise is holding constant in electron units, which means it will show a steady DR decrease of one stop, for each one-stop increase in ISO (which is how the D3 behaves from ISO 800 and up).

In summary, the D3s shows modest improvement at low ISO settings due to the 1/2 stop reduction in shot noise (which however does not improve the noise level in blacks below ISO 800). There is moderate improvement in the ISO 800-1600 range, and substantial improvement - more than a full stop - from ISO 3200 and up.

I suppose I can live with that!

Ralf Ronander wrote:

Just one Q:
how can more captured photons give less shot noise?

Photon noise decreases with more photons. Read up on Poisson distribution.

The read noise is addressed later down in Marianne's post.

Thank you, Marianne. A useful piece of work on your part, and an incredible achievement on Nikon's. A reduction of read noise by a factor of two puts it up with Canon's latest, but the really amazing part is the 37% increase in QE, over a sensor which was already the best in the business for QE.
You haven't, by any chance, got the figures in electrons?

Marianne Oelund wrote:

Wouldn't you know it: The first thing Marianne does with her new D3s, is to take a bunch of photos of a blank wall!

I'm glad that I did, though, because it's yielded some interesting data to share with everyone who's curious about how Nikon endowed the D3s with its high-ISO performance. I hope you will find this report to be worth my staying up until 1am to finish the set through ISO 12800.

Base ISO

This was the first check to make. If the new sensor has much improved quantum efficiency, this could result in the base ISO rising, so I wanted to make sure that ISO 200 is still a "real" ISO. The check came out good; just as with the D3, ISO values below 200 are artificial, but from 200 and up, the RAW data scales linearly with ISO setting, given a constant exposure.

Full Well Capacity and Quantum Efficiency

I re-tested both of my D3 bodies, plus the new D3s, for this - just to make sure I produced a valid comparison. For some obscure reason - sunspots or moon phase or other strangeness - photons are behaving better today, and I achieved higher FWC results for my D3's than I have before. Because of this discrepancy, I am only going to report relative performance between the D3s and D3, instead of giving absolute measurements.

To make a long story short, I found a 37% increase in quantum efficiency for the D3s, i.e., given the same exposure, the D3s collects that many more photons in each sensel. That is almost 1/2 stop improvement. Along with that increase, the full well capacity has gone up by just about the same factor, which is the reason why the D3s keeps the same base ISO.

The Q.E. improvement will give the D3s that 1/2 stop advantage in shot noise across the entire ISO range, and across the tonal range as well. It isn't a huge change, but I would say it is still important, as it does affect apparent noise levels in the entire image.

Color Filter Array

So far, the data I have suggests that the color filters haven't changed. At least, I know the R:G:B ratios seen by the D3s for my light source are essentially identical to what the D3 sees. Additional testing with various subject colors will be needed for confirmation that the CFA is indeed the same.

Low ISO Performance

At low ISO, the camera's DR is limited by the A/D converter S/N ratio. It appears that the same converters are being used, as I find no difference in DR from ISO 200-400.

Higher ISO Settings

As we proceed up the ISO scale from 800, the D3s starts to acquire a substantial advantage over the D3, due to improvements in read noise. The behavior of the D3 with regard to read noise, is that it scales with ISO, from 800 and up; this is equivalent to a constant read noise measured in electrons.

The D3s design is a significant departure in its behavior as gain is increased for the higher ISO settings. As gain rises through the ISO 800-6400 range, read noise referred to electrons gradually decreases, and at 6400, it is almost down to half of the D3 noise level. This improvement, combined with the 1/2 stop gain in shot noise performance, results in a DR advantage of up to 1.5 stops for the D3s, compared to the D3.

Here is how the D3s DR compares to that of the D3 over this ISO range:

• ISO 800 - D3s +0.3 stops

• ISO 1600 - D3s +0.7 stops

• ISO 3200 - D3s +1.2 stops

• ISO 6400 and 12800 - D3s +1.5 stops

Above ISO 6400, D3s read noise is holding constant in electron units, which means it will show a steady DR decrease of one stop, for each one-stop increase in ISO (which is how the D3 behaves from ISO 800 and up).

In summary, the D3s shows modest improvement at low ISO settings due to the 1/2 stop reduction in shot noise (which however does not improve the noise level in blacks below ISO 800). There is moderate improvement in the ISO 800-1600 range, and substantial improvement - more than a full stop - from ISO 3200 and up.

I suppose I can live with that!

Kaj E wrote:

Ralf Ronander wrote:

Just one Q:
how can more captured photons give less shot noise?

Photon noise decreases with more photons. Read up on Poisson distribution.

He's right, the noise increases as the number of photons, but only as the square root of the number, so the signal to noise ratio increases as the square root of the number of photons.

Paradoxically, the magnitude of shot noise is highest in the brightest areas, but you can't see it because its masked by the signal. That is the basis of Nikon's lossy raw compression, not so much encoding precision is needed in bright areas because the dithering of the noise means that the number of useful levels is smaller.

...and congratulations with your technical background, it's both a pleasure and highly interesting to read you.

Kindest regards,
Stany
http://www.fotografie.fr/
http://www.fotografie.fr/fotoforum/index.php

I prefer one really good picture in a day over 10 bad ones in a second...

Hi Marianne

I'm not going to pretend I understand this at your level but I think I can grasp what you are saying.

Thanks for taking the time and effort to share your findings, I consider myself a photographer and not a scientist but from your expertise comes a further understanding that's for sure!

Best regards
Tony

http://www.wildoat.zenfolio.com

It is very understandable information,

Even to me
Thank you for sharing.
Michel

Marianne Oelund wrote:

Wouldn't you know it: The first thing Marianne does with her new D3s, is to take a bunch of photos of a blank wall!

I'm glad that I did, though, because it's yielded some interesting data to share with everyone who's curious about how Nikon endowed the D3s with its high-ISO performance. I hope you will find this report to be worth my staying up until 1am to finish the set through ISO 12800.

Base ISO

This was the first check to make. If the new sensor has much improved quantum efficiency, this could result in the base ISO rising, so I wanted to make sure that ISO 200 is still a "real" ISO. The check came out good; just as with the D3, ISO values below 200 are artificial, but from 200 and up, the RAW data scales linearly with ISO setting, given a constant exposure.

Full Well Capacity and Quantum Efficiency

I re-tested both of my D3 bodies, plus the new D3s, for this - just to make sure I produced a valid comparison. For some obscure reason - sunspots or moon phase or other strangeness - photons are behaving better today, and I achieved higher FWC results for my D3's than I have before. Because of this discrepancy, I am only going to report relative performance between the D3s and D3, instead of giving absolute measurements.

To make a long story short, I found a 37% increase in quantum efficiency for the D3s, i.e., given the same exposure, the D3s collects that many more photons in each sensel. That is almost 1/2 stop improvement. Along with that increase, the full well capacity has gone up by just about the same factor, which is the reason why the D3s keeps the same base ISO.

The Q.E. improvement will give the D3s that 1/2 stop advantage in shot noise across the entire ISO range, and across the tonal range as well. It isn't a huge change, but I would say it is still important, as it does affect apparent noise levels in the entire image.

Color Filter Array

So far, the data I have suggests that the color filters haven't changed. At least, I know the R:G:B ratios seen by the D3s for my light source are essentially identical to what the D3 sees. Additional testing with various subject colors will be needed for confirmation that the CFA is indeed the same.

Low ISO Performance

At low ISO, the camera's DR is limited by the A/D converter S/N ratio. It appears that the same converters are being used, as I find no difference in DR from ISO 200-400.

Higher ISO Settings

As we proceed up the ISO scale from 800, the D3s starts to acquire a substantial advantage over the D3, due to improvements in read noise. The behavior of the D3 with regard to read noise, is that it scales with ISO, from 800 and up; this is equivalent to a constant read noise measured in electrons.

The D3s design is a significant departure in its behavior as gain is increased for the higher ISO settings. As gain rises through the ISO 800-6400 range, read noise referred to electrons gradually decreases, and at 6400, it is almost down to half of the D3 noise level. This improvement, combined with the 1/2 stop gain in shot noise performance, results in a DR advantage of up to 1.5 stops for the D3s, compared to the D3.

Here is how the D3s DR compares to that of the D3 over this ISO range:

• ISO 800 - D3s +0.3 stops

• ISO 1600 - D3s +0.7 stops

• ISO 3200 - D3s +1.2 stops

• ISO 6400 and 12800 - D3s +1.5 stops

Above ISO 6400, D3s read noise is holding constant in electron units, which means it will show a steady DR decrease of one stop, for each one-stop increase in ISO (which is how the D3 behaves from ISO 800 and up).

In summary, the D3s shows modest improvement at low ISO settings due to the 1/2 stop reduction in shot noise (which however does not improve the noise level in blacks below ISO 800). There is moderate improvement in the ISO 800-1600 range, and substantial improvement - more than a full stop - from ISO 3200 and up.

I suppose I can live with that!

Great effort !!!

Thank you for this succinct review of the D3S v D3. It is educational, informative and most helpful.