This topic is in the 2011 archive. You can read it and like its posts here. It is read only.
Archive 2011Nikon D800 pictures..almost here??
p.4 #1Nikon D800 pictures..almost here??
Steen DK wrote:
Oh well, so what’s left ?
If there will actually be a specific D800 version / model with the anti-aliasing filter removed, then I may still buy it, depending of the price level.
I wouldn't be terribly surprised if the base model simply didn't come with an AA filter and there's some rumor confusion on that point. As you move that high up in resolution, issues with spatial frequency and moire become less of a problem.
Someone smarter than me once explained as you approach 40mp in the 35mm FX format the issue simply starts to take care of itself and AA filters become unnecessary.
Oh well, so what’s left ?
If there will actually be a specific D800 version / model with the anti-aliasing filter removed, then I may still buy it, depending of the price level.
I wouldn't be terribly surprised if the base model simply didn't come with an AA filter and there's some rumor confusion on that point. As you move that high up in resolution, issues with spatial frequency and moire become less of a problem.
Someone smarter than me once explained as you approach 40mp in the 35mm FX format the issue simply starts to take care of itself and AA filters become unnecessary.
Nov 20, 2011 at 03:34 PM
p.4 #2Nikon D800 pictures..almost here??
If Nikon has implemented PIXEL BINNING this could be an amazing all around camera and not so much a niche landscape & studio camera.
If they haven't got any sort of binning option... well... that sucks.
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.
If they haven't got any sort of binning option... well... that sucks.
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.Nov 20, 2011 at 03:36 PM
p.4 #3Nikon D800 pictures..almost here??
Jammy Straub wrote:
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.
Hi Jammy,
I would imagine that the camera will allow the user several file sizes to select from. Just like the current Nikon line up does where you can select large, medium, and small file sizes. If that's the case full resolution when you think you may need it and the ability to select a much smaller file size for everything else.
Jack
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.
Hi Jammy,
I would imagine that the camera will allow the user several file sizes to select from. Just like the current Nikon line up does where you can select large, medium, and small file sizes. If that's the case full resolution when you think you may need it and the ability to select a much smaller file size for everything else.
Jack
Nov 20, 2011 at 03:47 PM
p.4 #4Nikon D800 pictures..almost here??
allstarimaging wrote:
Jammy Straub wrote:
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.
Hi Jammy,
I would imagine that the camera will allow the user several file sizes to select from. Just like the current Nikon line up does where you can select large, medium, and small file sizes. If that's the case full resolution when you think you may need it and the ability to select a much smaller file size for everything else.
Jack
The trick is Nikon currently only allows you to select different sizes in JPG mode, the real trick is making it happen in RAW mode You can do it with Canon, Nikon hasn't done it yet.
Jammy Straub wrote:
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.
Hi Jammy,
I would imagine that the camera will allow the user several file sizes to select from. Just like the current Nikon line up does where you can select large, medium, and small file sizes. If that's the case full resolution when you think you may need it and the ability to select a much smaller file size for everything else.
Jack
The trick is Nikon currently only allows you to select different sizes in JPG mode, the real trick is making it happen in RAW mode
You can do it with Canon, Nikon hasn't done it yet. Nov 20, 2011 at 03:50 PM
p.4 #5Nikon D800 pictures..almost here??
Jammy Straub wrote:
If Nikon has implemented PIXEL BINNING this could be an amazing all around camera and not so much a niche landscape & studio camera.
If they haven't got any sort of binning option... well... that sucks. I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.
I'm with you Jammy - but some people are saying we are dinosaurs for not wanting 36MP for whatever we do. I'd rather have the same process and electronic technology let loose on a 16MP or 18MP FX chip for even better color sensitivity and tonal range. I really like to stretch my shots regarding exposure in the shadows or in extremely low light situations and many times the color info is just not there - so a sensor sharing the same technology as a 36MP one, but with half the pixels ought to have significantly better characteristics in these areas. Unless binning can do this as well.
Hard drives are cheap (unless you use various flavors of advanced RAID configurations to store your images, then adding storage is NOT that cheap), but needing the absolute fastest processor available to efficiently process 36MP images is not cheap at all - in fact they are exponentially more expensive than lesser performing chips.
John
If Nikon has implemented PIXEL BINNING this could be an amazing all around camera and not so much a niche landscape & studio camera.
If they haven't got any sort of binning option... well... that sucks.
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.I'm with you Jammy - but some people are saying we are dinosaurs for not wanting 36MP for whatever we do. I'd rather have the same process and electronic technology let loose on a 16MP or 18MP FX chip for even better color sensitivity and tonal range. I really like to stretch my shots regarding exposure in the shadows or in extremely low light situations and many times the color info is just not there - so a sensor sharing the same technology as a 36MP one, but with half the pixels ought to have significantly better characteristics in these areas. Unless binning can do this as well.
Hard drives are cheap (unless you use various flavors of advanced RAID configurations to store your images, then adding storage is NOT that cheap), but needing the absolute fastest processor available to efficiently process 36MP images is not cheap at all - in fact they are exponentially more expensive than lesser performing chips.
John
Nov 20, 2011 at 04:13 PM
p.4 #6Nikon D800 pictures..almost here??
jhinkey wrote:
I'm with you Jammy - but some people are saying we are dinosaurs for not wanting 36MP for whatever we do. I'd rather have the same process and electronic technology let loose on a 16MP or 18MP FX chip for even better color sensitivity and tonal range. I really like to stretch my shots regarding exposure in the shadows or in extremely low light situations and many times the color info is just not there - so a sensor sharing the same technology as a 36MP one, but with half the pixels ought to have significantly better characteristics in these...Show more →
Both the D7000 and 5D Mark II demonstrated that these attributes aren't compromised by going to smaller pixels. In fact the D7000 proved the opposite.
I'm with you Jammy - but some people are saying we are dinosaurs for not wanting 36MP for whatever we do. I'd rather have the same process and electronic technology let loose on a 16MP or 18MP FX chip for even better color sensitivity and tonal range. I really like to stretch my shots regarding exposure in the shadows or in extremely low light situations and many times the color info is just not there - so a sensor sharing the same technology as a 36MP one, but with half the pixels ought to have significantly better characteristics in these...Show more →
Both the D7000 and 5D Mark II demonstrated that these attributes aren't compromised by going to smaller pixels. In fact the D7000 proved the opposite.
Nov 20, 2011 at 04:43 PM
p.4 #7Nikon D800 pictures..almost here??
Mark_L wrote:
You seriously never have an out of focus shot with the D700?
your D700 may need a tuneup with nikon. i shoot the 24/1.4, 50/1.8g, 70-200vr1 and previously the 85/1.4g. wide open most of the time. not a pro, just follow my kids and their classmates. playgrounds, museums, aquariums, theme parks, sports, horseback riding. mostly candids, no posed portraiture. i say easily a 90% keeper rate, if not better. and this is with me rarely using the center af point. if i toss a shot, its because of bad composition. if i miss AF, its typically me being too slow on the draw, or just being too near MFD (big reason why i sold the 85/1.4).
FWIW, i am an AF-ON user. more than any camera body, the switch to the backbutton focusing method gave me the biggest jump in keeper rates...
Mark_L wrote:
I'm sure we will see af points nearer the edges of the frame or pretty soon we will have m4/3rds cameras with better af than slrs. At the very least we need to have more cross sensors, even the new canon has that
i too hope for a wider spread of AF points. however, i will say that having used both canon and nikon cameras, i have achieved better accuracy out of nikon's regular af point over canon's cross-type af points. nikon may have slower AF, but its pretty darn accurate.
You seriously never have an out of focus shot with the D700?
your D700 may need a tuneup with nikon. i shoot the 24/1.4, 50/1.8g, 70-200vr1 and previously the 85/1.4g. wide open most of the time. not a pro, just follow my kids and their classmates. playgrounds, museums, aquariums, theme parks, sports, horseback riding. mostly candids, no posed portraiture. i say easily a 90% keeper rate, if not better. and this is with me rarely using the center af point. if i toss a shot, its because of bad composition. if i miss AF, its typically me being too slow on the draw, or just being too near MFD (big reason why i sold the 85/1.4).
FWIW, i am an AF-ON user. more than any camera body, the switch to the backbutton focusing method gave me the biggest jump in keeper rates...
Mark_L wrote:
I'm sure we will see af points nearer the edges of the frame or pretty soon we will have m4/3rds cameras with better af than slrs. At the very least we need to have more cross sensors, even the new canon has that
i too hope for a wider spread of AF points. however, i will say that having used both canon and nikon cameras, i have achieved better accuracy out of nikon's regular af point over canon's cross-type af points. nikon may have slower AF, but its pretty darn accurate.
Nov 20, 2011 at 05:24 PM
p.4 #8Nikon D800 pictures..almost here??
snapsy wrote:
Both the D7000 and 5D Mark II demonstrated that these attributes aren't compromised by going to smaller pixels. In fact the D7000 proved the opposite.
Well, I agree with you if these are compared at their respective base ISOs. But as DXOmark shows, at high ISOs the bigger pixels of the D700 win out over the smaller pixels of the D7000 in the areas of color sensitivity and tonal range. When you really start moving exposure sliders a lot for challenging images this is where the larger area pixels start to show their stuff.
Alas, I've been putting off getting a D7K, but perhaps I should and then I'll be able to do direct comparisons. What I've said above has also been born out by those who have both cameras - at base ISO the D7K wins, but at ISO 3200 or so the D700 is much more pliable. And it's already very pliable with 3+ year old sensor technology. I would have to be far more pliable if today's sensor technology were used at the same pixel density.
John
jhinkey wrote
I'm with you Jammy - but some people are saying we are dinosaurs for not wanting 36MP for whatever we do. I'd rather have the same process and electronic technology let loose on a 16MP or 18MP FX chip for even better color sensitivity and tonal range. I really like to stretch my shots regarding exposure in the shadows or in extremely low light situations and many times the color info is just not there - so a sensor sharing the same technology as a 36MP one, but with half the pixels ought to have significantly better characteristics in these areas. Unless binning can do this as well.Well, I agree with you if these are compared at their respective base ISOs. But as DXOmark shows, at high ISOs the bigger pixels of the D700 win out over the smaller pixels of the D7000 in the areas of color sensitivity and tonal range. When you really start moving exposure sliders a lot for challenging images this is where the larger area pixels start to show their stuff.
Alas, I've been putting off getting a D7K, but perhaps I should and then I'll be able to do direct comparisons. What I've said above has also been born out by those who have both cameras - at base ISO the D7K wins, but at ISO 3200 or so the D700 is much more pliable. And it's already very pliable with 3+ year old sensor technology. I would have to be far more pliable if today's sensor technology were used at the same pixel density.
John
Nov 20, 2011 at 05:29 PM
p.4 #9Nikon D800 pictures..almost here??
Now if they came up with a new AF system that was more light sensitive so that it worked at lower light levels that would be fantastic!
- John
- John
Nov 20, 2011 at 05:30 PM
p.4 #10Nikon D800 pictures..almost here??
patriot wrote:
The D700 was is a great general purpose camera that excels in high ISO's. Would it be progress to accept less? Especially if this can be done with a more reasonable increase in MP's?
You have obviously not understood that smaller pixels doesn't give more noise per image height.
In fact, you can get lower noise with more pixels because of lower readout noise. The only concern would be DR, but as you probably know, the current sensors with highest DR are smaller than FF.
jhinkey wrote:
Well, I agree with you if these are compared at their respective base ISOs. But as DXOmark shows, at high ISOs the bigger pixels of the D700 win out over the smaller pixels of the D7000 in the areas of color sensitivity and tonal range. When you really start moving exposure sliders a lot for challenging images this is where the larger area pixels start to show their stuff.
Having 2-3 times as many of those D7000 pixels, will definitely more than make up for the higher noise per pixel.
The D700 was is a great general purpose camera that excels in high ISO's. Would it be progress to accept less? Especially if this can be done with a more reasonable increase in MP's?
You have obviously not understood that smaller pixels doesn't give more noise per image height.
In fact, you can get lower noise with more pixels because of lower readout noise. The only concern would be DR, but as you probably know, the current sensors with highest DR are smaller than FF.
jhinkey wrote:
Well, I agree with you if these are compared at their respective base ISOs. But as DXOmark shows, at high ISOs the bigger pixels of the D700 win out over the smaller pixels of the D7000 in the areas of color sensitivity and tonal range. When you really start moving exposure sliders a lot for challenging images this is where the larger area pixels start to show their stuff.
Having 2-3 times as many of those D7000 pixels, will definitely more than make up for the higher noise per pixel.
Nov 20, 2011 at 05:38 PM
p.4 #11Nikon D800 pictures..almost here??
jhinkey wrote:
If they haven't got any sort of binning option... well... that sucks. I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.I'm with you Jammy - but some people are saying we are dinosaurs for not wanting 36MP for whatever we do. I'd rather have the same process and electronic technology let loose on a 16MP or 18MP FX chip for even better color sensitivity and tonal range. I really like to stretch my shots regarding exposure in the shadows or in extremely low light situations and many times the color info is just not there - so a sensor sharing the same technology as a 36MP one, but with half the pixels ought to have significantly better characteristics in these...Show more →
what you need is exactly a canon 1dx
Jammy Straub wrote
If Nikon has implemented PIXEL BINNING this could be an amazing all around camera and not so much a niche landscape & studio camera.If they haven't got any sort of binning option... well... that sucks.
I'd not have any interest in processing 1000+ 36mp RAW images from a wedding or event.what you need is exactly a canon 1dx
Nov 20, 2011 at 06:24 PM
p.4 #12Nikon D800 pictures..almost here??
Makten wrote:
You have obviously not understood that smaller pixels doesn't give more noise per image height.
In fact, you can get lower noise with more pixels because of lower readout noise. The only concern would be DR, but as you probably know, the current sensors with highest DR are smaller than FF.
Having 2-3 times as many of those D7000 pixels, will definitely more than make up for the higher noise per pixel.
Not talking about noise.
patriot wrote
The D700 was is a great general purpose camera that excels in high ISO's. Would it be progress to accept less? Especially if this can be done with a more reasonable increase in MP's?In fact, you can get lower noise with more pixels because of lower readout noise. The only concern would be DR, but as you probably know, the current sensors with highest DR are smaller than FF.
jhinkey wrote
Well, I agree with you if these are compared at their respective base ISOs. But as DXOmark shows, at high ISOs the bigger pixels of the D700 win out over the smaller pixels of the D7000 in the areas of color sensitivity and tonal range. When you really start moving exposure sliders a lot for challenging images this is where the larger area pixels start to show their stuff.Not talking about noise.
Nov 20, 2011 at 06:36 PM
p.4 #13Nikon D800 pictures..almost here??
patriot wrote:
The D700 was is a great general purpose camera that excels in high ISO's. Would it be progress to accept less? Especially if this can be done with a more reasonable increase in MP's?
Makten wrote:
You have obviously not understood that smaller pixels doesn't give more noise per image height.
In fact, you can get lower noise with more pixels because of lower readout noise. The only concern would be DR, but as you probably know, the current sensors with highest DR are smaller than FF.
Yes you are correct, it is possible, but I think you missed my point:
"If this rumored D800 does not perform better than the D3s in low light, would you call that progress?"
Please note the "If" at the beginning of that statement.
No one knows if there is any validity to this rumor, let alone how it will perform. My point is 36 MP's by itself does not equal progress for a successor to the D700: there needs to be a step forward in high ISO noise better than Nikon's the current best sensor (found in the D3s). Equal to, or slightly better than, the D700 is not progress.
Perhaps Nikon can do this on a 36 MP sensor. BUT if they can't, it would not be a true successor to the D700. It might be a worthy competitor to the 5D MkIII, but the D700 is not and was never intended to compete with the 5 series.
Nov 20, 2011 at 06:50 PM
p.4 #14Nikon D800 pictures..almost here??
static808 wrote:
your D700 may need a tuneup with nikon. i shoot the 24/1.4, 50/1.8g, 70-200vr1 and previously the 85/1.4g. wide open most of the time. not a pro, just follow my kids and their classmates. playgrounds, museums, aquariums, theme parks, sports, horseback riding. mostly candids, no posed portraiture. i say easily a 90% keeper rate, if not better. and this is with me rarely using the center af point. if i toss a shot, its because of bad composition. if i miss AF, its typically me being too slow on the draw, or just being too near MFD (big reason why...Show more →
I shoot musical theater, which includes action and dance. Lots if lighting shifts, and often low light. I select the AF point I need because I want to isolate certain performers, often within groups. It's fairly common on the outer AF's for the AF to perform slowly and hunt, and an expression or dance move missed.
I had a 1D MkIII, and the AF lock on any AF point was instant, rarely hunted. BUT on that model, lock did not always mean accurate. I missed many potentially good images due to this. On the other hand, I've missed images because the D700 could not lock because the outer AF's are not cross types. So it's got the shot, but unusable with the 1D, or never got the shot with the D700. I get more keepers with the D700, which is why I kept it and sold the 1D.
I'm hoping to see the next generation AF in the D700 successor have more cross type AF's, and would like to see the spread out a little more.
Mark_L wrote
You seriously never have an out of focus shot with the D700?Mark_L wrote
I'm sure we will see af points nearer the edges of the frame or pretty soon we will have m4/3rds cameras with better af than slrs. At the very least we need to have more cross sensors, even the new canon has thatI shoot musical theater, which includes action and dance. Lots if lighting shifts, and often low light. I select the AF point I need because I want to isolate certain performers, often within groups. It's fairly common on the outer AF's for the AF to perform slowly and hunt, and an expression or dance move missed.
I had a 1D MkIII, and the AF lock on any AF point was instant, rarely hunted. BUT on that model, lock did not always mean accurate. I missed many potentially good images due to this. On the other hand, I've missed images because the D700 could not lock because the outer AF's are not cross types. So it's got the shot, but unusable with the 1D, or never got the shot with the D700. I get more keepers with the D700, which is why I kept it and sold the 1D.
I'm hoping to see the next generation AF in the D700 successor have more cross type AF's, and would like to see the spread out a little more.
Nov 20, 2011 at 07:10 PM
p.4 #15Nikon D800 pictures..almost here??
jhinkey wrote:
Not talking about noise.
The "higher-density sensor is equal to or better" than lower density sensor applies to all the IQ metrics at the image level, be it SNR, dynamic range, tonal range, etc.. These elements are not a function of pixel size (at least at the image level) but instead a function of sensor technology, and each generation of sensor gets better in this regard at the pixel level so at the image level you'll see IQ equal to or better than the previous low-density generation.
Not talking about noise.
The "higher-density sensor is equal to or better" than lower density sensor applies to all the IQ metrics at the image level, be it SNR, dynamic range, tonal range, etc.. These elements are not a function of pixel size (at least at the image level) but instead a function of sensor technology, and each generation of sensor gets better in this regard at the pixel level so at the image level you'll see IQ equal to or better than the previous low-density generation.
Nov 20, 2011 at 07:32 PM
p.4 #16Nikon D800 pictures..almost here??
*sigh*
Can we wait until the friggin' thing announces to argue about it?
For the record, I want to see the (hypothetical) D400 as much or more as the D800...
Also, unless I'm misreading every review and the technical comparison, whoever said the D3s color is in some way inferior to the D700/D3 is on drugs. If I'm mis-informed/wrong, please enlighten me.
Can we wait until the friggin' thing announces to argue about it?
For the record, I want to see the (hypothetical) D400 as much or more as the D800...
Also, unless I'm misreading every review and the technical comparison, whoever said the D3s color is in some way inferior to the D700/D3 is on drugs. If I'm mis-informed/wrong, please enlighten me.
Nov 20, 2011 at 07:36 PM
p.4 #17Nikon D800 pictures..almost here??
Makten wrote:
You have obviously not understood that smaller pixels doesn't give more noise per image height.
In fact, you can get lower noise with more pixels because of lower readout noise. The only concern would be DR, but as you probably know, the current sensors with highest DR are smaller than FF.
I bet most random image noise is quantization noise anyway, and this can be overcome by stuffing the image processing chip with more ADCs (assuming staggered round-robin sampling). More ADCs => lower clock => log drop in noise. Also, more real estate => greater cost. I don't know what the economy would be, but clearly it's quite possible to improve image quality by throwing silicon at the problem. There may also be various incremental improvements in noise and power usage. (Ground noise is a significant contributor to quantization noise, and a DSP core that uses lots of power produces lots of noise.)
You have obviously not understood that smaller pixels doesn't give more noise per image height.
In fact, you can get lower noise with more pixels because of lower readout noise. The only concern would be DR, but as you probably know, the current sensors with highest DR are smaller than FF.
I bet most random image noise is quantization noise anyway, and this can be overcome by stuffing the image processing chip with more ADCs (assuming staggered round-robin sampling). More ADCs => lower clock => log drop in noise. Also, more real estate => greater cost. I don't know what the economy would be, but clearly it's quite possible to improve image quality by throwing silicon at the problem. There may also be various incremental improvements in noise and power usage. (Ground noise is a significant contributor to quantization noise, and a DSP core that uses lots of power produces lots of noise.)
Nov 20, 2011 at 07:45 PM
p.4 #18Nikon D800 pictures..almost here??
Nikon had a winner with the D3/D700 combo, the D700 being the baby D3, same sensor, inferior body (but in some eyes superior). Did Nikon introduce the D700 simply to try to lure photographers back to Nikon after so many had jumped ship to Canon? Who knows, only Nikon does, all we can do is guess. The D3s was introduced and many D700 owners have been waiting for the day the D700s will be announced, same D700 but with the D3s sensor... but now everything is pointing at the D800 and the possibility of a 36MP sensor? Seriously a 3x jump in megapixel? My only thought is Nikon wanted something to compete with the 5D Mark III and they assumed it would have monster MP as well, but where is the D700s? My hope is that since it would be just an upgrade to the sensor it is hiding in the shadows and will be introduced after the D800 and D4. Canon got it right with the 1D X and 18MP, I hope Nikon didn't get sucked back in to the MP war. For years I've been wanting low MP and better ISO. Would be nice if we had a D700s at 12MP with the D3s sensor, a D4 with 18MP or so to compete with the 1D X and a D800 with 36MP for those wanting uber high MP, but if Nikon kills the "700" series I would be very sad. Right now I'm waiting to see what Nikon/Canon will do with their replacements.
Nov 21, 2011 at 12:05 AM
p.4 #19Nikon D800 pictures..almost here??
patriot wrote:
My point is 36 MP's by itself does not equal progress for a successor to the D700: there needs to be a step forward in high ISO noise better than Nikon's the current best sensor (found in the D3s). Equal to, or slightly better than, the D700 is not progress.
This is a bit ridiculous, since that would then be the successor of the D3s, not the D700. Higher resolution is of course progress. If it suits your personal taste or not has nothing to do with it.
My point is 36 MP's by itself does not equal progress for a successor to the D700: there needs to be a step forward in high ISO noise better than Nikon's the current best sensor (found in the D3s). Equal to, or slightly better than, the D700 is not progress.
This is a bit ridiculous, since that would then be the successor of the D3s, not the D700. Higher resolution is of course progress. If it suits your personal taste or not has nothing to do with it.
Nov 21, 2011 at 12:55 AM
p.4 #20Nikon D800 pictures..almost here??
Jan Brittenson wrote:
I bet most random image noise is quantization noise anyway, and this can be overcome by stuffing the image processing chip with more ADCs (assuming staggered round-robin sampling). More ADCs => lower clock => log drop in noise. Also, more real estate => greater cost. I don't know what the economy would be, but clearly it's quite possible to improve image quality by throwing silicon at the problem. There may also be various incremental improvements in noise and power usage. (Ground noise is a significant contributor to quantization noise, and a DSP core that uses lots of power produces...Show more →
You'd have to secure that bet pretty heavily, since 99% of all noise contaminating a capture of any kind, of any pixel size, is photon shot noise. This is governed by the laws of physics as we know them, not by the electronics interpreting the measurement results we get.
Lower ADC clock and more accurate readouts gives you more DR, which in no way should be mistaken for "lower noise" over the whole picture frame as such. Yes, a better DR means that you lower the point where the capture induced noise gets stronger than the shot noise, so in theory you also get a very slightly better S/N ratio at 18% gray too. But this is hardly even noticeable until you get up to about ISO25600. The point where you can see it is right at the limit where the noise gets stronger than the signal - and that's at the blackpoint.
18% gray at ISO6400 in a D7000 is about 130 photons per pixel, giving a shot noise of about 12e-. At the same point of measurement, the total sum of added electronic noise is about 3e-. That's four times lower than the shot noise, or 2Ev down if you want to see it that way. An Rntot
of 3 gives that you need to get down to 3^2 = 9 photons before the electronics do as much "damage" to the signal as the inherent properties of light itself.
At the 18% gray level, the electronics only add (sqrt(12^2 + 3^2) - 12) = 0.37e- to the 12e- that we can't get away from. That's an increase of about 3%, or 1/25Ev. Not a significant deterioration of the signal.
So, in a D7000 size pixel, the noise inherent in mother nature outweighs the added synthetic noise until you get down to about -7Ev at ISO6400 (this is the base definition of the term "DR"...). This is of course assuming the electronic noise is of a "good and random nature", and not in a fixed pattern structure, like they are in banding effects and so on.
What you need to get good low-light behavior is a combination of high efficiency and good DR. And if the D7000 and the D5100 could be taken as indicators, Nikon are already getting there. They are - if you look at say a 10x10mm cutout from the sensor - both a lot closer to D3s performance than to D3/D700 performance at high ISOs - even though their pixels are more than three times smaller in area...
I bet most random image noise is quantization noise anyway, and this can be overcome by stuffing the image processing chip with more ADCs (assuming staggered round-robin sampling). More ADCs => lower clock => log drop in noise. Also, more real estate => greater cost. I don't know what the economy would be, but clearly it's quite possible to improve image quality by throwing silicon at the problem. There may also be various incremental improvements in noise and power usage. (Ground noise is a significant contributor to quantization noise, and a DSP core that uses lots of power produces...Show more →
You'd have to secure that bet pretty heavily, since 99% of all noise contaminating a capture of any kind, of any pixel size, is photon shot noise. This is governed by the laws of physics as we know them, not by the electronics interpreting the measurement results we get.
Lower ADC clock and more accurate readouts gives you more DR, which in no way should be mistaken for "lower noise" over the whole picture frame as such. Yes, a better DR means that you lower the point where the capture induced noise gets stronger than the shot noise, so in theory you also get a very slightly better S/N ratio at 18% gray too. But this is hardly even noticeable until you get up to about ISO25600. The point where you can see it is right at the limit where the noise gets stronger than the signal - and that's at the blackpoint.
18% gray at ISO6400 in a D7000 is about 130 photons per pixel, giving a shot noise of about 12e-. At the same point of measurement, the total sum of added electronic noise is about 3e-. That's four times lower than the shot noise, or 2Ev down if you want to see it that way. An Rntot
of 3 gives that you need to get down to 3^2 = 9 photons before the electronics do as much "damage" to the signal as the inherent properties of light itself.
At the 18% gray level, the electronics only add (sqrt(12^2 + 3^2) - 12) = 0.37e- to the 12e- that we can't get away from. That's an increase of about 3%, or 1/25Ev. Not a significant deterioration of the signal.
So, in a D7000 size pixel, the noise inherent in mother nature outweighs the added synthetic noise until you get down to about -7Ev at ISO6400 (this is the base definition of the term "DR"...). This is of course assuming the electronic noise is of a "good and random nature", and not in a fixed pattern structure, like they are in banding effects and so on.
What you need to get good low-light behavior is a combination of high efficiency and good DR. And if the D7000 and the D5100 could be taken as indicators, Nikon are already getting there. They are - if you look at say a 10x10mm cutout from the sensor - both a lot closer to D3s performance than to D3/D700 performance at high ISOs - even though their pixels are more than three times smaller in area...
Nov 21, 2011 at 02:16 AM