Any thoughts on how the 5d3 will compare on dxo mark to the Nikon D800?

[jrista]

@LetTheRightLensIn:

Your still missing the point. So lets use an example to demonstrate. Say you shoot a scene with 14.4 stops of DR, and blow the highlights, since the camera is only capable of 13.23 and your trying to preserve as much shadow detail as possible. You THINK the camera is capable of capturing all of the DR in the scene, because you believe what DXO claims about the D800's DR. However, once you get to it, you realize no amount of POST-PROCESS SPATIAL AVERAGING is going to RECOVER those blown highlights. They were blown well before it ever got to the point of averaging them down...they were blown in the photodiode, amplified in the sensor, converted at maximum level by the ADC. Those pixels hit their maximum saturation and then some...by 1.17 stops (2.25 times more than the sensor is capable of.)

Do you agree or disagree with that point?



Lets agree that downsampling can produce a result that is a perceptual improvement over a similar image from another camera at the same dimensions (because on that point, I do agree!) When you downsample, you average noise across pixels and therefor lose noise, you sample detail from multiple pixels into fewer pixels and lose resolution and detail, you collapse more information into less total area to produce an image of smaller width and/or height so you lose pixel density at a similar print size. Etc. etc.

Downsampling incurs a LOSS OF INFORMATION, not a gain of information. The only improvement, the only gain, is on a perceptual basis, in comparison with other images that started out at smaller sizes. Since a downsampled image is based on a similar image of larger size, every pixel in the output image has more information to work with than the source image...however the grand total amount of information remains the same! There is no increase of data in the output image, it simply makes more effective use of the information that was available. An increase in DR cannot actually occur, because if the original source information contains blown highlights, no amount of multisampling or averaging can change that.

 

[LetTheRightLensIn]

1. If your highlights are blown it means you over-exposed, simple as that. No sensor will ever bring back details blown to nothing even if it has 100 stops. You always expose to not blow the brightest part of a scene that you care about and then the darkest parts fall where they do, either above or below a usable noise floor.

2. It's not important to think of the exact print DxO numbers in absolute terms, they will only be correct for viewing an image at once particular scale. The more generally meaningful part is their relative values between various camera bodies.

3. Canon potentially (assuming this is all true and there were no legit hidden technical reasons) does deserve to get blasted for this, because, from what I've heard whispered about, while they have gone around trumpeting how infinitely far ahead in sensor design they were and how they could just sit around and rule the roost doing nothing or react, if actually need ever be, the real story is that they have been doing things like brushing aside IN-HOUSE (although in the DSLR or camera division) developed patents that would have increased dynamic range by 2 stops because they didn't care to bother, they just want to milk their old designs and fabs and charge more for it, their goal for the 5D3 was to first, insure increased profit margin per copy.

And they probably could've had a true 1920x1080 res video, at least in 1.6x crop mode, out of the 5D3 as well, but either didn't think of it in time or chose to leave it out in the end to protect the C300 and cinema line, never mind that their revolution was with the 5 series for video.

Yes, word is they had the tech to make it 2 stops better dynamic range, developed by an internal, but outside of Japan, division of Canon, but Canon DSLR Japan division wanted nothing to do with it and blew off their other division and here we are with Nikon/Sony looking in their rear view mirror at Canon DSLR sensor tech. One group was willing to put in the effort to build a better sensor an done group couldn't have cared less apparently.

I don't know if they still have access to those patents, whether they will look into them again now that they are getting blasted for being left behind in DR or how long it would take to get them into a working sensor at this point or if something was lost in translation and there was some sneaky little techincal thing that made the thing infeasible and they actually turned it down because it wasn't realistically workable.

4. If they made use of the internal patents and given it the cropped true 1920x1080 I bet this thing would be getting nearly universal praise even at $3500 and been one of their most universally praised releases. A few might still gripe a bit about MP, but I think the better part of that crowd would've been made happy enough with the 2 extra stops of dynamic range.

@LetTheRightLensIn:

Your still missing the point. So lets use an example to demonstrate. Say you shoot a scene with 14.4 stops of DR, and blow the highlights, since the camera is only capable of 13.23 and your trying to preserve as much shadow detail as possible. You THINK the camera is capable of capturing all of the DR in the scene, because you believe what DXO claims about the D800's DR. However, once you get to it, you realize no amount of POST-PROCESS SPATIAL AVERAGING is going to RECOVER those blown highlights. They were blown well before it ever got to the point of averaging them down...they were blown in the photodiode, amplified in the sensor, converted at maximum level by the ADC. Those pixels hit their maximum saturation and then some...by 1.17 stops (2.25 times more than the sensor is capable of.)

Do you agree or disagree with that point?



Lets agree that downsampling can produce a result that is a perceptual improvement over a similar image from another camera at the same dimensions (because on that point, I do agree!) When you downsample, you average noise across pixels and therefor lose noise, you sample detail from multiple pixels into fewer pixels and lose resolution and detail, you collapse more information into less total area to produce an image of smaller width and/or height so you lose pixel density at a similar print size. Etc. etc.

Downsampling incurs a LOSS OF INFORMATION, not a gain of information. The only improvement, the only gain, is on a perceptual basis, in comparison with other images that started out at smaller sizes. Since a downsampled image is based on a similar image of larger size, every pixel in the output image has more information to work with than the source image...however the grand total amount of information remains the same! There is no increase of data in the output image, it simply makes more effective use of the information that was available. An increase in DR cannot actually occur, because if the original source information contains blown highlights, no amount of multisampling or averaging can change that.

 

[LetTheRightLensIn]

Apparently the tech engineers at Canon DSLR were never even let look at it (some method to get 2 stops better DR), the marketing guys and lawyers saw it, more or less said they had no clue what it meant and it seemed dumb more DR and told their employee in the other division to not bother to file a patent on it! Apparently it would have not required any fancy new setups or anything to implement. Maybe it had some flaw, but the marketing droids just tossed it away not even letting the DSLR engineers get a look at it and didn't want to pay a token for a patent filing!

 

[jrista]

Apparently the tech engineers at Canon DSLR were never even let look at it (some method to get 2 stops better DR), the marketing guys and lawyers saw it, more or less said they had no clue what it meant and it seemed dumb more DR and told their employee in the other division to not bother to file a patent on it! Apparently it would have not required any fancy new setups or anything to implement. Maybe it had some flaw, but the marketing droids just tossed it away not even letting the DSLR engineers get a look at it and didn't want to pay a token for a patent filing!

Where the hell do you get this shit, man? That sounds like the biggest bunch of crazy bull you've spouted yet, and there isn't a single reference to back a scrap of it up (Canon Marketing/Legal interfering with Canon Engineering, preventing them from doing their job? WTF kind of inane shit is that?). I keep trying to get you back to facts and science (your last two posts have completely skirted my direct questions). You just seem to want to rag on Canon, and your starting to really hit the deep end now. If you want to rag on Canon such that they will hear you...call them.



No, Canon's DR is NOT as good as Nikons...but Nikon's is not nearly as good as DXO makes it sound. There aren't 3 stops of difference between the two...at a hardware capability level (derived from native, unprocessed RAW data), there is a difference of 1.77 stops, or about 1 2/3rds of a stop, of DR between the 1D IV and the D800 (and I expect teh 5D III and 1D X to outperform the 1D IV at a hardware level). Ok, so Nikon wins by a third less than two stops...wow? At ISO 400 onward, its all neck and neck, and when you get to really high ISO's, the 5D III seems to be doing considerably better than any SoNikon in the area of SNR and noise quality.

You complain that many Canon users in a Canon form are happy about the improvements Canon HAS MADE, making exclamations about how its all pure fanboyism devoid of any meaningful data to back up their ilbegotten exuberance. You seem to think you have the high road here as well with your complaining about a single issue that exhibits at a single camera setting, and think its the only valid purpose from now until the day Canon finally declares you king of the squeaky wheels and oils the one issue you personally want to complain about.

Flip it all around for a moment dude, and try to see it from the other side. Despite the fact that the 5D III is not some revolutionary headliner in the annals of camera technology designed to wow and jerk tears from camera technotites the world round...despite the fact that they did not fix EVERY SINGLE TINY THING THAT EVERYONE ON EARTH WANTED FIXED...despite the fact that Canon most likely gimped its video to help sell more C300's....despite all that, it is probably the best camera Canon has ever produced. It's relatively reasonably priced (something the 1D X is not, which is why I don't proclaim that as the best), it has Canon's best AF system ever (and potentially the best AF system across brands period once people learn how to use it), it has what is potentially Canon's best sensor (yes, beyond even that of the 1D X, given that it is a very nice rounding of resolution AND high ISO performance), along with an army of ergonomic, functional, and software improvements to make this camera potentially the most well-rounded, well-balanced, broad-reaching camera money can buy. Its not the absolute best at any one thing...but it could very well be the best at just about everything. (Including answering many very long-term, major, and explicitly valid complaints from existing 5D II users...such as its AF system, the greater desire for better ISO and noise quality than continued increases in megapixels, and better metering.)

Nope, its not the DR king. Nope, its not the ISO king. Nope, its not the FPS king. But it'll DO EVERYTHING, SERVICE EVERY NEED, and do it all quite well.

 

[edknuff]

This is all very interesting, seriously, but y'all are making my brain hurt!
;-)

 

[peederj]

Since brains can't feel pain, let's muse...

Let's say your lens and subject can throw 7.5 million photons on a 1mm x 1mm sensor area while your shutter is open. (The real numbers may be orders of magnitude off from these, just for sake of discussion.)

Let's also assume a photocell can count from 0 to 16,000 photons per the time period you have your shutter set for. At the 0 end, 0 photons won't be counted as zero every time, instead, there is a standard deviation of 10 photons of noise on all readings. Fine. Our dynamic range, per photocell, is established at 0-16,000 +/- 10.

Now we can't do much about the noise factor per photocell other than try to develop better photocells. But what if we can squeeze more of those photocells, without altering their dynamic range, in that 1 square mm area?

The noise per-native-pixel performance will remain constant at +/- 10 photons. But what about the headroom from clipping? Since we have now 500 photocells in that 1x1sqmm area instead of 300, the number of photons per photocell on our brightest subject areas will be smaller. What would have saturated out 16,000 photons per photocell is now divided into several other photocells, each of which is now able to count more photons prior to clipping. Yes? Follow me?

And afterward, as we have I hope established, downscaling from the higher resolution is no longer limited to the 16,000 photons per pixel. When we are downscaling, we can work in a deeper numerical space, allowing the equivalent of, say, 64,000 photons per pixel without incident. We can normalize the result of the downscaling to the noise floor of 10 photons.

And as a result, should this analysis apply (and please tell me how it doesn't): greater sensor photocell density, while maintaining a relatively constant dynamic range per photocell, will increase the dynamic range deliverable in a given lower-than-native resolution output simply by virtue of increasing the effective headroom by dividing the load onto more photocells.

No?

 

[LetTheRightLensIn]

Apparently the tech engineers at Canon DSLR were never even let look at it (some method to get 2 stops better DR), the marketing guys and lawyers saw it, more or less said they had no clue what it meant and it seemed dumb more DR and told their employee in the other division to not bother to file a patent on it! Apparently it would have not required any fancy new setups or anything to implement. Maybe it had some flaw, but the marketing droids just tossed it away not even letting the DSLR engineers get a look at it and didn't want to pay a token for a patent filing!

Where the hell do you get this S___, man? That sounds like the biggest bunch of crazy bull you've spouted yet, and there isn't a single reference to back a scrap of it up (Canon Marketing/Legal interfering with Canon Engineering, preventing them from doing their job? WTF kind of inane S___ is that?). I keep trying to get you back to facts and science (your last two posts have completely skirted my direct questions).

Where? from .
Check your PMs.

No, Canon's DR is NOT as good as Nikons...but Nikon's is not nearly as good as DXO makes it sound. There aren't 3 stops of difference between the two...at a hardware capability level (derived from native, unprocessed RAW data), there is a difference of 1.77 stops, or about 1 2/3rds of a stop, of DR between the 1D IV and the D800 (and I expect teh 5D III and 1D X to outperform the 1D IV at a hardware level).

The 5D3 has so far tested same as the 5D2 and that is a good 2.5 stops less than D800. Whether you care or anyone else cares, who knows, but that doesn't mean it's only 0.5 stop of only 1-2/3rd stops of what not (and, BTW, 1.77 stops would be pretty huge difference actually).

Ok, so Nikon wins by a third less than two stops...wow?

How did 2.5 become 1.77 become suddenly less than .666?

At ISO 400 onward, its all neck and neck, and when you get to really high ISO's,

More like ISO1600 and on, maybe more like ISO3200 and on due to the higher MP count allowing for better noise adjustment than simple resizing and/or more detail until, if you wanted to get really picky. Anyway above ISO1600 the difference should quickly become less and less.

the 5D III seems to be doing considerably better than any SoNikon in the area of SNR and noise quality.

It does? From what I've seen so far, it looks equal, at best, at the mid high isos. The only place it might become considerably better, from what I've seen so far, is at the 50,000/100,000 type ISOs. I'm still waiting for DxO on the SNR stuff though. Hopefully it will more or less tie them for SNR and DR.

You complain that many Canon users in a Canon form are happy about the improvements Canon HAS MADE, making exclamations about how its all pure fanboyism devoid of any meaningful data to back up their ilbegotten exuberance.

No I didn't. I praised them for putting the 1DX AF into it. And I've said it is too bad but the D800 video seems to be riddled with color moire still.

Flip it all around for a moment dude, and try to see it from the other side. Despite the fact that the 5D III is not some revolutionary headliner in the annals of camera technology designed to wow and jerk tears from camera technotites the world round...despite the fact that they did not fix EVERY SINGLE TINY THING THAT EVERYONE ON EARTH WANTED FIXED...despite the fact that Canon most likely gimped its video to help sell more C300's....despite all that, it is probably the best camera Canon has ever produced. It's relatively reasonably priced (something the 1D X is not, which is why I don't proclaim that as the best), it has Canon's best AF system ever (and potentially the best AF system across brands period once people learn how to use it), it has what is potentially Canon's best sensor (yes, beyond even that of the 1D X, given that it is a very nice rounding of resolution AND high ISO performance), along with an army of ergonomic, functional, and software improvements to make this camera potentially the most well-rounded, well-balanced, broad-reaching camera money can buy. Its not the absolute best at any one thing...but it could very well be the best at just about everything. (Including answering many very long-term, major, and explicitly valid complaints from existing 5D II users...such as its AF system, the greater desire for better ISO and noise quality than continued increases in megapixels, and better metering.)

Nope, its not the DR king. Nope, its not the ISO king. Nope, its not the FPS king. But it'll DO EVERYTHING, SERVICE EVERY NEED, and do it all quite well.

 

[briansquibb]

I wonder how many shots will deliver the full dr avalable at iso 100?

I have been deliberately trying to get max dr on my 1d4 and 1ds3 - and if DPP is to believed I seem to max out at about 10, whereas dxo say about 12.

How important is that extra from the D800 at iso100/200 - when in fact the main place it is needed is in low light when the iso is high and the dr is low.

I would love to acheive a DR of 10 (dpp measurement) at iso 3200 - so the important measure to me would be the dr at iso 1600-6400 which is where it seems the gap between the D800 and the 5DIII is almost negligable.

 

[jrista]

And afterward, as we have I hope established, downscaling from the higher resolution is no longer limited to the 16,000 photons per pixel. When we are downscaling, we can work in a deeper numerical space, allowing the equivalent of, say, 64,000 photons per pixel without incident. We can normalize the result of the downscaling to the noise floor of 10 photons.

And as a result, should this analysis apply (and please tell me how it doesn't): greater sensor photocell density, while maintaining a relatively constant dynamic range per photocell, will increase the dynamic range deliverable in a given lower-than-native resolution output simply by virtue of increasing the effective headroom by dividing the load onto more photocells.

No?

NO. Ignoring electron gain for the moment, your hardware is still limited to 16,000 photons per well.

Lets say we have an ideal monochrome 32x32 pixel sensor (1024 pixels), with a maximum saturation of 16,384 electrons per well, an electron gain of 1 (one photon = on electron) and a standard deviation of noise of +/-5. If you capture 17,384 photons in the center 16x16 pixels, those 256 wells are fully saturated, registering 16,384 electrons per well. The excess 1000 photons is waste. The diode cannot hold any more electrons, so any additional photon strikes are either reflected, or converted to heat (which, itself, could end up becoming an electron in a neighboring pixel or elsewhere in the electronics, contributing to noise). Now you digitize with a 14-bit ADC. Lets just assume you have a gain of 1, so you get 1024 pixels, the center 256 of which have a maximum value of 16,384 (pure white). The center white pixels are bordered by a basic grayscale gradient faloff, with the outer edge containing pixel values 0 +/-5 (which obviously leads to pixels with values between 0-5.)

Now you want to downsample. Your digitized image has 256 pixels in the center with BAKED IN MAXIMUM VALUES. Lets say we downscale considerably, such that we have a 3x3 pixel image. We average the center 256 pixels into the single center pixel of our 3x3 pixel image. The center pixel is...still pure white. The border of pixels around it become...well, roughly an even gray (the average of pixel values between +/-5 to 15360 differentiated by 1024 levels), with the corner pixels possibly approaching a darker gray more so than the rest, but none actually reaching the +/-5 floor for noise.

Two key points here:

1. Averaging pixels cannot increase highlight range, so if highlights are blown, they are blown regardless of how big the image is.
- Assuming you have a 9 pixel 3x3 area full of maximum saturation pixels, sampling them all to produce a single output pixel will always result in a pixel of maximum value.
- (16384 + 16384 + 16384 + 16384 + 16384 + 16384 + 16384 + 16384 + 16384) / 9 = 147456/9 = 16384
- The only thing that changes is the physical dimensions of the blown area...which would logically shrink in a downscaled image.
2. Averaging pixels also cannot increase shadow range, at best it can maintain it, and on average it will likely reduce shadow DR.
- If you have a 9 pixel 3x3, ranging in value from 0 to 5, sampling them all to produce a single output pixel will rarely result in an output pixel that is zero unless the majority of source pixels are zero.
- (0 + 2 + 1 + 5 + 2 + 4 + 0 + 0 + 3) / 9 = 17/9 = 1.8889 = 2

Mathematically, once you start averaging pixels with baked-in representations of an analog signal, you cannot gain dynamic range by downsampling (averaging multiple source pixel values together into single output pixels), even with more complex algorithms like bilinear or bicubic filtering (which does more to smooth averaging out over a greater area than anything, incurring an additional cost in terms of detail.) Mathematically, when it comes to dynamic range, you are more likely to LOSE DR (as demonstrated in point 2 above, where you started out with a dynamic range of 0-5, and ended up with a range of 2-2), and at best, you might keep it the same. You will never be able to recover what you lost by not having a deep enough pixel well to start with.

Relating this to DXO:

1. Screen DR is representative of post-ACD baked-in RAW representations of the analog sensor signal (divided by a possible gain factor).
2. Print DR is representative of a downsampled RGB image generated from the RAW representation that is itself a representation of the analog sensor signal.

If you want to know what your sensor is capable of, the DXO "Screen DR" measurement is the closest you are going to get to a hardware reading. As for Print DR...mathematically with simple downscaling, you can't gain DR. I am not really sure what print dr represents (I used to think I did, however DXO showing that the D800 gains 1.17 stops of DR simply by downsampling, putting the "Print DR" ABOVE the physical limitation of 14 stops the hardware imposes...has me utterly baffled and finding the measurement rather useless as a result), however it is not actually indicative of what the actual camera HARDWARE is capable of. It is far more indicative of what digital computer algorithms can do with an image once its been converted from analog to digital and turned into a RAW or RGB file, and that really has nothing to do with the camera, and everything to do with software and your desktop computer.

 

[Tcapp]

This is all very interesting, seriously, but y'all are making my brain hurt!
;-)

My thoughts exactly! I don't care about numbers. I care about making art. Sure, image quality is important. If it looks good to my eyes, i'm happy. And if it looks good to my clients eyes, i'm even happier. Nuff said. 8)

 

[psolberg]

I wonder how many shots will deliver the full dr avalable at iso 100?

I have been deliberately trying to get max dr on my 1d4 and 1ds3 - and if DPP is to believed I seem to max out at about 10, whereas dxo say about 12.

How important is that extra from the D800 at iso100/200 - when in fact the main place it is needed is in low light when the iso is high and the dr is low.

I would love to acheive a DR of 10 (dpp measurement) at iso 3200 - so the important measure to me would be the dr at iso 1600-6400 which is where it seems the gap between the D800 and the 5DIII is almost negligable.

why would dynamic range not matter any more than it does at other ISOs? You can't just make blanket statements like that and we still don't have the DXO results from the 5DIII to measure how much of a nosedive that camera takes as ISO increases. DR is important at all ISOs. Surely landscape and studio artists (the main audience of the D800 love the ISO100 performance). But even wedding photogs which are often asked to shoot at 1pm with blazing clear sky and big apertures will not be pushing the ISO often and will be more concerned about highlights. There are plenty more cases for the low ISO DR just as there are for the higher ISO DR. Neither is more important and both are very key.

No, Canon's DR is NOT as good as Nikons...but Nikon's is not nearly as good as DXO makes it sound. There aren't 3 stops of difference between the two...at a hardware capability level (derived from native, unprocessed RAW data), there is a difference of 1.77 stops, or about 1 2/3rds of a stop, of DR between the 1D IV and the D800 (and I expect teh 5D III and 1D X to outperform the 1D IV at a hardware level). Ok, so Nikon wins by a third less than two stops...wow? At ISO 400 onward, its all neck and neck, and when you get to really high ISO's, the 5D III seems to be doing considerably better than any SoNikon in the area of SNR and noise quality

as "LetTheRightLensIn" points out, you're being very incosistent with your numbers and I have no idea where you got those numbers and how you're backing them up. regardless of you accepting the DXO results, the fact remains the DR difference is both noticeable and relevant even if it isn't exactly their quoted figures. Still the 5DmkIII and 1DX haven't been tested that much so I'm open to the possibility they may be better than the standard canon low scores, but as far as your argument to bring down the Nikon's obvious advantage, it simply holds no water. Let's just accept that fact and move on. 12 stops is still quite good even if it is behind everybody else.

Nope, its not the DR king. Nope, its not the ISO king. Nope, its not the FPS king. But it'll DO EVERYTHING, SERVICE EVERY NEED, and do it all quite well.

I agree, and most cameras are like that so clearly it isn't all that fair to judge the 5DmkIII purely on the shortcomings it has. Nikon had the more exciting camera this time as illustrated by all the D800 madness and let them have the glory since they really hit it out of the park when everybody doubted it. There is always another year and newer models coming out better than the last one so it is pointless to focus on a single head to head comparison for too long.

 

[PhilDrinkwater]

I must be honest - I find it mildly sad that a d800 vs. 5diii sensor ratings thread makes it to 11 pages It suggests that many people are looking in the wrong place for their photography. I know this is a gear forum, but still... photography is about capturing photographs, not drawing graphs.

As soon as a camera reaches a certain level, I think the body becomes vastly less relevant to photographers *on average*. The minute detail differences between the cameras will *for most photographers* make very little difference.

My view is that I could take my photographs (see links below) with pretty much any cameras from the last 5 years and it would be impossible for most people to tell the difference. I know there are always some who push the envelope in terms of dynamic range or ISO or AF, but the average photographer (pro, semi pro or amateur) won't notice a difference and their work will be the same as it was before.

It's all mildly interesting but it certainly won't make me change my mind about what I'm buying and why...

"Photographer > Lens > Body - for most photography"

 

[briansquibb]

why would dynamic range not matter any more than it does at other ISOs? You can't just make blanket statements like that and we still don't have the DXO results from the 5DIII to measure how much of a nosedive that camera takes as ISO increases. DR is important at all ISOs. Surely landscape and studio artists (the main audience of the D800 love the ISO100 performance). But even wedding photogs which are often asked to shoot at 1pm with blazing clear sky and big apertures will not be pushing the ISO often and will be more concerned about highlights. There are plenty more cases for the low ISO DR just as there are for the higher ISO DR. Neither is more important and both are very key.

Whilst iso at 100/200 is important, the point I am trying to make is that improvements there are unlikely to be significant in terms of printed output - printed output is the most critical. No we dont know the DxO measurements for the 5D3 - however we do have the 5D2 figure, and my thoughts are that the 5D3 results will if anything be better.

The most important dr is when it drops off as this leave most headroom and flatter colour.

There are numerous effective techniques for handling blazing sun and high contrast - high contrast and strong shadow makes horrible photos so you will need to work on them rather than rely on high DR to avoid blowing the highlights. Numerous videos on youTube are available for the old fashioned but effective techniques.

 

[Northstar]

Just my opinion, but something doesn't seem right regarding DXO:

1. They give the D800 a very high "all time record" score (that they knew would have the internet buzzing)
2. THEN....THE DON'T REVIEW/SCORE THE CANON 5D3(it's been two weeks since the d800 review) EVEN THOUGH BOTH CAMERAS WERE RELEASED AT THE SAME TIME. How long does it take to do a review?

I find this interesting because they're partly responsible for creating a situation where the D800 is in extremely high demand and is getting very positive "word of mouth"....while the 5d3 plays second fiddle. It's almost as if they've INTENTIONALLY delayed their 5d3 review.

At this point, even if they come out and give the 5d3 a very good score, they've already sold a bunch of D800's for Nikon(and therefore a bunch of lenses), and given them a clear boost during the very important initial launch period for both cameras.

makes me wonder.....

 

[LetTheRightLensIn]

Just my opinion, but something doesn't seem right regarding DXO:

1. They give the D800 a very high "all time record" score (that they knew would have the internet buzzing)
2. THEN....THE DON'T REVIEW/SCORE THE CANON 5D3(it's been two weeks since the d800 review) EVEN THOUGH BOTH CAMERAS WERE RELEASED AT THE SAME TIME. How long does it take to do a review?

I find this interesting because they're partly responsible for creating a situation where the D800 is in extremely high demand and is getting very positive "word of mouth"....while the 5d3 plays second fiddle. It's almost as if they've INTENTIONALLY delayed their 5d3 review.

At this point, even if they come out and give the 5d3 a very good score, they've already sold a bunch of D800's for Nikon(and therefore a bunch of lenses), and given them a clear boost during the very important initial launch period for both cameras.

makes me wonder.....

On another forum, someone claimed that Canon France delayed getting them a release copy of the 5D3. Who knows. (and the thing is the 5D3 probably will test worse, so, if anything, they may be helping Canon )

 

[LetTheRightLensIn]

Interesting stuff on DPR where they figured out how D800 video works. They read all pixels in a scanline, but only read every third scanline. They then downsample it from a bit over 1920x1080 down to 1920x1080. That explains why compared to 5D3 video:

D800 video is a bit sharper (not by much once 5D3 is sharpened in post)

D800 video has considerably worse SNR (5D3 appears to 3x3 bin the entire sensor while Nikon appears to skip 2/3rds of the lines and thus tosses away 2/3rds of the collected photons, it collects even a bit less light than an APS-C sensor of the same technology would collect) and yet the same is not true for stills (where both make use of the full sensor). The D800 tosses away 1.8 stops of SNR and dynamic range for video compared to what it can do for stills apparently (but having a large base ISO DR advantage the ISO100 video on the D800 still has better DR than the 5D3, I'd guess they are even for DR by ISO400 for video and by ISO800 it should already have worse DR than 5D3 video and by ISO1600+ it should have a lot worse DR than the 5D3 video probably like 2 stops worse). The worse SNR should hold across all ISOs.

So D800 video probably has something along the lines of 1.5 to 2 stops worse SNR across the ISO range and ranging from perhaps something like 2/3rd of a stop better DR at ISO 100 to even at ISO400, worse at ISO800 and probably 1.5-2 stops worse from ISO1250 or so and up.

Most importantly D800 video has more aliasing and a lot more color moire.


I think you have to give the video win to the 5D3, a little less sharp and maybe 2/3 stop less DR at ISO100 (but likely already even by ISO400) but none of the moire mess, little of the aliasing and 1.5-2 stops better SNR across the board and at ISO800 and up it has better DR, especially ISO1600+.

(that said they still BADLY need to update 5D3 firmware to add zebra stripes, video focus peaking and a 1.6x crop 2x2 sampled video mode with top clarity and better reach, then they will get the videographers singing its praises and they will fly off the shelf to movie guys)

5D3 also has better liveview sensor output since for some odd reason nikon uses some sort of video feed o the D800 liveview instead of scaled mode like on D7000 so their liview 100% look on the D800 has every other line skipped, apparently it makes it tricky to use to focus unless you try to look at vertical details only.

So it seems that 5D3 gets the win for video and liveview quality.

 

[peederj]

Very interesting and potentially relevant. Link please.

Update: I found this:

http://falklumo.blogspot.de/2012/04/lumolabs-nikon-d800-video-function.html

Interesting stuff on DPR where they figured out how D800 video works. They read all pixels in a scanline, but only read every third scanline. They then downsample it from a bit over 1920x1080 down to 1920x1080. That explains why compared to 5D3 video:

D800 video is a bit sharper

D800 video has considerably worse SNR (5D3 appears to 3x3 bin the entire sensor while Nikon appears to skip using 2/3rds of collected photons, it puts even a bit less light than an APS-C sensor would collect) and yet not so much difference for stills (where both make use of the full sensor). D800 tosses away 1.8 stops of SNR and dynamic range for video compared to what it can do for stills apparently (but having a large base ISO DR advantage the ISO100 video on the D800 still has better DR than the 5D3, I'd guess they are even for DR by ISO400 for video and by ISO800 it should already have worse DR than 5D3 video and by ISO1600+ it should have a lot worse DR than the 5D3 video probably like 2 stops worse).

D800 video has more aliasing and a lot more color moire.

I think you have to give the video win to the 5D3, a little less sharp and DR at ISO100 but none of the moire mess. At ISO800 and up it has noticeably better SNR and DR, especially ISO1600+.

(that said they still BADLY need to update 5D3 firmware to add zebra stripes, video focus peaking and a 1.6x crop 2x2 sampled video mode with top clarity and better reach, then they will get the videographers singing its praises and they will fly off the shelf to movie guys)

5D3 also has better liveview sensor output since for some odd reason nikon uses some sort of video feed instead of scaled mode like on D7000 so their liview 100% look has every other line skipped.