Concerns over the Nikon D5 DR Strategy

[Josh-H]

On the other hand, many images sell well precisely because they have so much visible detail,

Interesting.. This has not been my experience and the images I am seeing that sell in Galleries are not purchased because they have 'so much visible detail'. They are purchased for the emotional connection the viewer has with the image. Im not sure I have ever heard anyone say 'I bought this because it had so much detail'.

In my own experience, it is on the whole, photographers who care about 'visible detail'. And again, on the whole, photographers are not the ones purchasing prints in my experience. Art buyers care about emotional connection and not about detail (assuming they are not an investment purchase). YMMV.. but thats my experience

BTW: We are now way off the original topic.. sorry about that.

[ErikKaffehr]

Hi,

It is an interesting discussion, thanks for taking on that…

I am not selling pictures, so I have no experience with that. My standard print size is A2, 16"x23". That size is the largest you can print on a normal desktop printer using cut sheet paper. For me it works nicely with 50x70 cm framing with a mat.

At that size I essentially found that there is no great advantage of say 24 MP over 12 MP. Most of my best pictures were made 12 MP because I was doing some nice travel. When I got into 24 MP I did compare 12 MP to 24MP and could observe a small advantage of 24 MP, but not in all cases.

For some reasons, most not very good, I bought a used MFD equipment. So for the last three years I was shooting 39 MP with an old Hasselblad system with the P45+ back. The P45+ gives 39 MP. Again, I made some comparisons and found that I and a few other observers could not tell my A2-prints apart. Must tell, in one of the cases three observers picked the MFD image, although I could not see a technical difference.

So, my conclusion is that 24 MP is very good for A2 and 12 MP is reasonably good for A2. On the other hand, I was told by a printer I respect that 36 MP has advantages over 16 MP when printing on glossy papers even at A2 size.

I seldom print larger than A2, in part due to wall space limitation, next size up is either 70x100 cm or 50x100 cm. At 70x100 mm I think it is possible to make stunning prints from 24 MP, but both the photographer and the printer need to do their homework. I think a 24 MP images will be OK but not great in 70x100 cm without proper sharpening workflow.

My best friend uses Canon 5DIII and I don't lack DR in his images. I always had cameras having a decent amount of DR, at least since the Sony A-100 came out, but I seldom felt DR was a limitation. The Sony A99 had a significant advantage in DR over my Sony Alpha 900, but it took a long time to observe.

I have been on two Dolomites workshops with Hans Kruse. Hans shoots both Canon and Nikon, with Nikon having the DR advantage and before the 5DsR also the resolution advantage. Hans fully acknowledged the DR advantage of the Nikon, but you cannot miss that he likes shooting with the Canon a bit more.

Getting back to Megapixels. My clear understanding is that human vision is dominated by low to medium frequency detail. So an image that has good MTF for low and medium frequencies will look better than an image processed for actual pixel viewing on screen.

I started looking at sharpening for prints a while ago, using SQF (Subjective Quality Factor) as a measure for print quality, using MTF data measured on my A7rII, what I found that SQF can be essentially constant for increasing image sizes even at short viewing distance (50 cm / 20"). What I found worked best was no sharpening in LR, followed by 75% at radius r=2 in FocusMagic.

Check the corresponding SQF plot, below. You can see that MTF is almost constant up to 0.3cy/pixel, that is 60% of Nyquist, than drops rapidly to Nyquist. Even at large print sizes the contrast sensivity of human vision will be within the straight part of the sharpened MTF curve. Sharpness drops at Nyquist so aliasing/fake detail near Nyquist is not enhanced.

Technically, this image is a bit oversharpened. MTF goes over 100% between 0.1 and 0.3 cy/pixel. I have seen some "museum standard" demanding between 90 and 105% in the relevant range. But, this sharpening would give optimal quality for large prints. At least for that camera used with that lens.



I have of course made prints, using small cropped areas, and I have tested both LR and C1. C1 produces a tiny bit cleaner detail. The crops I viewed corresponded to something like 40"x60". The crops were A4 size, which allowed for side by or one on top over the other viewing at close.

Best regards
Erik

BTW: We are now way off the original topic.. sorry about that.

[shadowblade]

Interesting.. This has not been my experience and the images I am seeing that sell in Galleries are not purchased because they have 'so much visible detail'. They are purchased for the emotional connection the viewer has with the image. Im not sure I have ever heard anyone say 'I bought this because it had so much detail'.

Being aesthetically pleasing, having an interesting subject and fitting in with the decor is a baseline prerequisite - if an image doesn't fit those criteria, it's not even in the race.

But, often, many images will fit the criteria. Then people will take a second, third or fourth look at them, get closer to the image, and, assuming the print size is large enough to see a difference, the ones they will usually comment on and show further interest in are the ones where, as they look more closely, they are able to see more and more things which weren't apparent on the first viewing that grabbed their attention, maintaining interest for longer - things which aren't apparent on lower-resolution images because the resolution wasn't great enough to show them.

Sure, if you show them a low-resolution image printed at a large size in isolation, they'll usually be impressed (again, provided the image meets the basic criteria), but, put a low-resolution and high-resolution image with similar aesthetic qualities (sometimes even shot at the same place), printed on smooth paper or high-gloss aluminium, side-by-side, and the difference is startling - all of a sudden, the lower-resolution image doesn't look so good, all else being equal.

I guess what I'm saying is that high resolution doesn't replace aesthetic considerations. But it certainly helps. The aesthetics of your shot are not going to change depending on which camera you're shooting with - that part's up to you. But what the camera can do - and what no amount of skill on your part can do - is give you greater resolution, better colour and better dynamic range.

[BernardLanguillier]

Back to the original topic... or sort of...

http://cameradecision.com/compare/DxO-One-vs-Nikon-D5

This was the 5th link proposed by Google when I searched for DxO D5 support...

Cheers,
Bernard

[BernardLanguillier]

Dear all,

I have come accross this interesting site enabling easy comparison of photographic DR per ISO.

http://www.photonstophotos.net/Charts/PDR.htm

The D5 has already been measured.

It puts things in pespective a bit, as it shows that the catastrophically poor DR of the D5 at base ISO is in fact ahead of the Canon 5DIII for instance and within 0.5 stop of the 5DS until ISO 800~.

Cheers,
Bernard

[shadowblade]

Dear all,

I have come accross this interesting site enabling easy comparison of photographic DR per ISO.

http://www.photonstophotos.net/Charts/PDR.htm

The D5 has already been measured.

It puts things in pespective a bit, as it shows that the catastrophically poor DR of the D5 at base ISO is in fact ahead of the Canon 5DIII for instance and within 0.5 stop of the 5DS until ISO 800~.

Cheers,
Bernard

I've been looking at that.

The 5D3 and 5Ds are not exactly great at low ISO. They still use the old off-sensor ADCs and have significantly poorer low-ISO DR/noise than the Exmor sensors or the new 80D (when adjusted for sensor size). Congratulations - at low ISO, the new sensor slightly beats an 8-year-old 5D2 sensor and 4-year-old 5D3 sensor, and loses slightly to a new 5Ds sensor based on twelve-year-old technology that every other sensor company (and now Canon itself) has abandoned. Obviously, at high ISO, it's no contest.

Compared with the D4s, the D5 is a stop behind at low ISO. Compared to the D810 and A7r, it's about two stops behind at low ISO, and no better than the A7r2 at high ISO.

And look at the D5 in DX mode vs the Canon 80D, which uses the new on-sensor ADC (as does the 1Dx2). DX mode is 1.5x crop whereas the 80D is 1.6x crop, so the Nikon even has the sensor size advantage here. Yet, at base ISO, it's almost one-and-a-half stops behind.

I suspect the D5 will lose badly to the 1Dx2 on any objective level. The 1Dx2 is likely to have a better sensor - much better at low-ISO, and comparable at high-ISO. It shoots 2fps faster. It's meant to do better video. The AF systems are both top-of-the-line for their respective companies, so should be comparable. And Canon's line of superteles is superior to Nikon's. Of course, none of this matters if you're already heavily invested in Nikon glass...

[BernardLanguillier]

I suspect the D5 will lose badly to the 1Dx2 on any objective level. The 1Dx2 is likely to have a better sensor - much better at low-ISO, and comparable at high-ISO. It shoots 2fps faster. It's meant to do better video. The AF systems are both top-of-the-line for their respective companies, so should be comparable. And Canon's line of superteles is superior to Nikon's. Of course, none of this matters if you're already heavily invested in Nikon glass...

Realistically speaking AF is by far the most important aspect of these cameras. The rest doesn't matter as far as it is good enough, and both cameras are good enough in terms of sensor, handling, video,...

I'd love the 1DXII to have an AF performing even better than that of the D5, this would open even more possibilities for photographers. Per my first hand experience, the AF of the D5 is a significant progress on difficult subjects compared to anything I have used prior to it. It is anyone's guess whether Canon has been able to match that or not. Those 2 cameras being on top of their respective line up doesn't mean that they will perform the same (sensors would be identical too according to this very wicked logic)...

As far as super teles is concerned, both brand have an amazing line up with +/- on both sides with no real weak points. Any of those lenses can serve well top pros, the absolute better one will depend on the lens you actually need for your applications. If you need a 200-400 f4, Canon is clearly ahead, if you need any of the classical long teles (400, 500, 600, 800), Nikon is a bit ahead at the moment. If you need a DO lens, Canon is the way to go,...

Cheers,
Bernard

[shadowblade]

Realistically speaking AF is by far the most important aspect of these cameras. The rest doesn't matter as far as it is good enough, and both cameras are good enough in terms of sensor, handling, video,...

AF may be the most important aspect to these cameras, being action cameras, but is probably not the biggest distinguishing feature between them. After all, both will have AF that's more than good enough for 99.99% of purposes - in fact, the 1Dx and D4s are probably good enough for 99.99% of real photographic use, and the 5D3 and D810 good enough for 99.9% of use.

Video will likely be the biggest distinguishing feature. These are photojournalist's cameras, and, increasingly, photojournalists need to capture not only stills, but also video for web and TV (with newspapers and magazines increasingly moving online rather than being printed). The 1Dx2 essentially fulfils the role of both the 1Dx and 1Dc and, supposedly, has a lot of video functionality. From what I can tell, the D5 is less optimised for video, although its stills may be good. A sensor with better DR also helps a lot in this respect, since videographers don't have the same ability to use filters and other methods to deal with wide dynamic range.

I'd love the 1DXII to have an AF performing even better than that of the D5, this would open even more possibilities for photographers. Per my first hand experience, the AF of the D5 is a significant progress on difficult subjects compared to anything I have used prior to it. It is anyone's guess whether Canon has been able to match that or not. Those 2 cameras being on top of their respective line up doesn't mean that they will perform the same (sensors would be identical too according to this very wicked logic)...

I don't know - was there anything you really couldn't capture effectively with the D4s and 1Dx? This may be like comparing 720ppi vs 600ppi input between Epson and Canon/HP printers - both are far beyond what anyone actually needs.

As far as super teles is concerned, both brand have an amazing line up with +/- on both sides with no real weak points. Any of those lenses can serve well top pros, the absolute better one will depend on the lens you actually need for your applications. If you need a 200-400 f4, Canon is clearly ahead, if you need any of the classical long teles (400, 500, 600, 800), Nikon is a bit ahead at the moment. If you need a DO lens, Canon is the way to go,...

The long teles of both companies are probably good enough that there's little to choose between them - all of them are super-sharp and focus quickly and accurately. The big difference is the 200-400 - if you're not currently invested in either system, the Canon's inbuilt 1.4x TC would probably be enough to swing you in that direction, since it's so much more convenient (and also sharper) than Nikon's option.

[dwswager]

Resolution:

Given a 20MP image, double the resolution is 80MP, not 40MP.  Megapixels is just a proxy for resolution which for an image is 2 dimensional.  Hence, unless you print a 1 dimensional string of pixels, to double the resolution is not 2x, but 2^2x!  This is why the difference from 20MP or 24MP to 36MP is not as significant as MP count would lead you to believe. 

DR:

Nikon could have released a D5 with equal or better DR and similar curve characteristics to the D810 or D750, but chose not to.  So my point is trying to determine if they made a good choice.  Is the trade they made worth what they are giving up for the intended users of the D5.  I think the D500 people would make that trade.  That is give up base ISO to get better images at higher ISOs, but even then it will depend how much one gives up at the low end and how much one gets at the higher ISOs.

[BernardLanguillier]

AF may be the most important aspect to these cameras, being action cameras, but is probably not the biggest distinguishing feature between them. After all, both will have AF that's more than good enough for 99.99% of purposes - in fact, the 1Dx and D4s are probably good enough for 99.99% of real photographic use, and the 5D3 and D810 good enough for 99.9% of use.

Frankly, I don't think so.

When you talk about success ratio of perfectly focused images on randomly moving subjects in low light, we are very very far from being 99% good.

Cheers,
Bernard

[dwswager]

Frankly, I don't think so.

When you talk about success ratio of perfectly focused images on randomly moving subjects in low light, we are very very far from being 99% good.

Cheers,
Bernard

I concur.  I shot a soccer game 2 nights ago and the one shot I know I wanted on the card was a header goal.  It was serviceable, but not exactly sharp.  This was shot with a D810, 1.2x crop, 70-200mm f/2.8, at ISO 2500.  With action shooting, it is a very minor change in focus or release that takes a shot from great to merely serviceable.

[shadowblade]

Resolution:

Given a 20MP image, double the resolution is 80MP, not 40MP.  Megapixels is just a proxy for resolution which for an image is 2 dimensional.  Hence, unless you print a 1 dimensional string of pixels, to double the resolution is not 2x, but 2^2x!  This is why the difference from 20MP or 24MP to 36MP is not as significant as MP count would lead you to believe. 

That's obvious enough to anyone with a basic grasp of mathematics.

Even a 1.4x increase in linear dimensions is a huge increase in the size of a print. And, in practice, it's more than that, since larger prints tend not to need the same ppi as smaller ones.

DR:

Nikon could have released a D5 with equal or better DR and similar curve characteristics to the D810 or D750, but chose not to.  So my point is trying to determine if they made a good choice.  Is the trade they made worth what they are giving up for the intended users of the D5.  I think the D500 people would make that trade.  That is give up base ISO to get better images at higher ISOs, but even then it will depend how much one gives up at the low end and how much one gets at the higher ISOs.

They sacrificed low-ISO performance. But did they actually gain high-ISO performance from that, or was the gain in high-ISO performance completely unrelated? After all, the A7r2 sensor appears to have the same SNR at high ISO as the D5 sensor, without sacrificing low-ISO DR.

The only reason DR plateaus out at low ISO is due to the contribution of read noise - usually from electronics between the photosites and A/D conversion. Without the contribution of read noise, sensors are effectively ISO-less - that is, pushing an ISO 100 shot by six stops gives you the same result as taking it at ISO 6400 natively.

[shadowblade]

Frankly, I don't think so.

When you talk about success ratio of perfectly focused images on randomly moving subjects in low light, we are very very far from being 99% good.

Cheers,
Bernard

I'm not talking about 99.99% of shots of randomly-moving, fast subjects in very low light. I'm talking about 99.99% of all images.

The new AF system gives you an advantage in a subset of a subset of shots. If the subject is not both fast and random, or the light level is not low, the D4s/1Dx also keep up perfectly.

[ErikKaffehr]

Hi,

Don't you think there will be a D820 or a D900 soon, using the AF-technology from the D5 utilising a Sony sensor?

I am a bit surprised we have not seen the A7rII sensor in a Nikon.

If you check out Bill Claff's data it looks like a real champ at both low and high ISO. Keep also in mind that Nikon often makes a bit better use of Sony's sensors than Sony themselves. http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Nikon%20D810,Sony%20ILCE-7RII

Regarding resolution, I would say it is a good thing to have. It seems that the AA-filter has been dropped on most high res sensors, I am not sure that is a good thing but it increases perceived sharpness a bit.

If I can choose between a 24MP sensor and a 42 MP sensor and both have similar performance I would prefer the higher res one, any time!

Best regards
Erik

Best regards
Erik

[BernardLanguillier]

Yes, most probably, but don't forget that the D810 uses the same AF module as the D4s, yet is pretty far in actual AF performance.

Cheers,
Bernard

[BernardLanguillier]

I'm not talking about 99.99% of shots of randomly-moving, fast subjects in very low light. I'm talking about 99.99% of all images.

The new AF system gives you an advantage in a subset of a subset of shots. If the subject is not both fast and random, or the light level is not low, the D4s/1Dx also keep up perfectly.

For the applications targeted by these cameras the improvements at hand are very significant.

http://www.fredmiranda.com/forum/topic/1425827/0#lastmessage

Cheers,
Bernard

[shadowblade]

Yes, most probably, but don't forget that the D810 uses the same AF module as the D4s, yet is pretty far in actual AF performance.

Cheers,
Bernard

Not that far. For 99% of shots, both will keep up just fine, and you won't see a difference between the two.

If you're an action photographer, of course, you probably live in that other 1%, so you will see a difference in a larger number of your shots.

It's the same with resolution. 99% of the time, people will be printing small enough that there's no perceptible difference between a 20MP camera and a 50MP one. But if you live in that other 1% and a significant number of your shots end up as large prints, you will see a noticeable difference.

That's why these are specialist tools, not general-purpose cameras.

As for the AF module, remember that the D810 needs significantly more precision than the D4s to take advantage of all its pixels, due to the greater resolution. The AF is slower to allow this.

[shadowblade]

For the applications targeted by these cameras the improvements at hand are very significant.

http://www.fredmiranda.com/forum/topic/1425827/0#lastmessage

Cheers,
Bernard

Yet the applications where these improvements will be noticeable are only a tiny proportion of photographs.

If you're shooting fast action in good light, a D4s or 1Dx will also track perfectly (so will a 5D3 or D810). If things are dark but the subject isn't moving so fast towards or away from the camera, or is far away enough that the lens doesn't have to move much, a D4s or 1Dx will also track perfectly. You can't track better than perfectly - the D5's AF won't help where the tracking is already working well.

It's only in situations where the D4s/1Dx shoot somewhat less than perfectly that you'd see a significant improvement through the D5's AF - fast action in very dark environments.

Only a small number of users will actually gain a benefit over the previous model. That's to be expected, and there's nothing wrong with that - it's the nature of specialised cameras.

[hjulenissen]

Not likely. Current, even very dense sensors can handle live view and video (which still reads every pixel on the sensor) and can be used for hours-long exposures without overheating.

I doubt it, since current 42MP and 50MP sensors can sample video at 25 or 50fps, utilising every pixel on the sensor (although downsampling to 1080p or 4k for storage). The sampling rate of that is much faster than even reading 80MP at 14fps.

Are there any cameras on the market that reads out 50 million pixels at 50 fps?

This is probably the biggest issue, actually. Burst rates at full resolution would be very slow, unless you used a large, fast buffer. But it can be alleviated by downsampling to mRAW or sRAW for the faster speeds.

Not sure that you understood my point here. As long as you are writing to a large, fast internal buffer, then (obviously), the write speed to flash card does not matter.

The moment you need to drain that buffer (by shooting long series), the flash card writing starts to matter.

-h

[dwswager]

I started this thread specifically because the Nikon D5 is considered, by Nikon, as their preeminent DSLR.  That is, it is it is the one stop shop, generalist camera.  It certainly also is the best specific fast action based on the shutter box and buffer.  So the real question turned out to be is what they were getting on the high ISO end worth what they were giving in trade for it at the low end.  So far the general consensus seems to be yes...in a qualified manner.  The qualification seems to be most people commenting with thumbs up seem to be shooting it for action/low light. 

The real fault line here is the difference between how Nikon views their line up and how users view that same line-up.  The easy one is the D750.  Most users would have moved that camera to the professional line leaving the D610 as the enthusiast 24MP full frame camera.  And when I say move, I mean in a body of comparable durability and in the same accessory line as the D810 and D500.  Easy way to spot is the round viewfinder eyepiece.  Hence, the majority that would be effected by the loss of low ISO performance, probably are already using something else like a D810 or D750 or DF.