Nikon D5 - first impressions

[Frank Fremerey]

Generally I like very much what you say, got me thinking a lot.

The D810 converts 2.8 Billion Photons to Electrons at ISO64 and below?
The D750 converts 2.1 Billion Photons to Electrons at ISO150 and below?
The D3 converts 0.6 Billion Photons to Electrons at ISO200?

What I do not understand yet is how base ISO affects the picture.

In post #55 I collected a comparsion of your quoted side, here is another one

(Also: what does the wave like pattern at low ISO mean (D3 / D5?):

[Øivind Tøien]

...
Another wish: Larger pixels in theory mean better DR but this rule seems not to hold true here.

Did the real base ISO go up so far?

When looking at Bill Claff's Photographic Dynamic Range curve one need to keep in mind that the data are standardized to a certain print size/viewing distance. Thus if two sensors of different resolution have the same Photographic Dynamic Range, the lower resolution sensor will be better when compared at pixel level (100% crop). So the low res sensors will inherently seem to have a disadvantage here which is not reflected at pixel level, but is so when viewing a print at the same same size and distance (similar magnification). Another consequence of this is when an Fx sensor image is cropped to DX format, the photographic dynamic range will be lower, as the image will be magnified more.

It will be interesting to see if the data can be reproduced. I supplied Bill data for the Nikon 1 AW1 sensor, and that changed some suspicious data he had received before which did not make sense as it differed markedly from the other 1 sensor curves.

[Frank Fremerey]

Thank you, Øivind Tøien. A lot to investigate here...

[Andy]

What I do not understand yet is how base ISO affects the picture.
(Also: what does the wave like pattern at low ISO mean (D3 / D5?):

Frank, the individual sensel is usually ISO agnostic. It just captures light. As much as it possible can.
The "ISO setting" happens later in the processing as the output of the sensor is calibrated towards the selected sensitivity level.

rgds, Andy

[Frank Fremerey]

Frank, the individual sensel is usually ISO agnostic. It just captures light. As much as it possible can.
The "ISO setting" happens later in the processing as the output of the sensor is calibrated towards the selected sensitivity level.

OK. And is there something like a "real ISO" or a "natural ISO" or a "sweet spot of performance" in the ISO range?

Why can I not dial in ISO 100 or 50 or 25 in my D3?

[Andy]

OK. And is there something like a "real ISO" or a "natural ISO" or a "sweet spot of performance" in the ISO range?

To my memory (I looked at this back in 2012). No at the individual sensel level, yes at the chip level.

If you are interested in further information:
The description of the D800 Sensor readout system in Sony's semiconductor news #47.
A patent for Panasonic on Ramp Wave based CDS systems.
A discussion we had in a german foto forum about these matters back in 2012 (sorry, in german language)

Why can I not dial in ISO 100 or 50 or 25 in my D3

2 quick comments on this:
1) The "native" ISO of the D3 is 200. The reason Nikon is labelling this lower value "Low 1" and not "ISO 100", is this setting doesn't fit the official ISO standard. Like the higher "Hi 1, etc ..." settings aren't compliant to the ISO definition (ISO standard 12232). We usually refer to a Hi 1 as (for example) ISO 12.800, but the sensor is beyond its ability to comply to the standard (esp noise floor levels). As it is technically easy to crank up the sensitivity of the sensor and the marketing department enjoys these super high values, the industry started a while ago with this "Hi" race. That'S why we have native ISO (compliant to the ISO Standard) and low/high settings (non or somewhat compliant to the ISO standard)
2) Setting a lower setting like 100 on the D3 wouldn't get you anything. The sweetspot of the D3 is closer to ISO 200 (max FWC and lowest read noise). Any lower setting reduces DR. So why offer it then? (One use case could be for a photographer, who would like to do panning of i.e. cars in motion during the day and need as low an ISO Setting as possible to stretch exposure time. A ND filter would impact his AF system, so a low ISO would be better)


As we talk about all the "flexbility" manufacturers have in "re"defining common terms, you might also pay attention about the change Nikon introduced with the D2X. Look up the technical data in the manual and you will find in the chapter on ISO settings, the term REI (Recommended Exposure Index). Another vehicle for manufacturer to be a bit more flexible what their cameras will do when a user set for instance ISO 100. ISO 100 on the outside is <> ISO 100 in the inside. Please read Douglas Kerr's paper on this subject.

Wouldn't it be easy, if it was easy?
I've spent some time in the past trying to understand all of this and its implications on photography. The consequence for me was, that I take now a much more relaxed view about settings, metrics and appropriate values and rather enjoy the easier side of photography

rgds, Andy

[Andy]

I was ready to use the superior capabilities of the D5 as a sports and action camera and travelled to a place to visit a horse race last weekend. Kind of.
All my gear was ready, the battery of the D5 was fully charged, AF was set to AF-C, CH was set, a new and empty card in the camera. A super fast lens on the camera. Ready, set, go!

Please find below the interpretation of the term "go" the horse had in mind instead. So, here we "go". This is then my "action shot" with the D5 on this occasion. The camera nailed it perfectly. No other camera could have done that (with the probable exception of the D40)

next subject please ....

[Akira]

When looking at Bill Claff's Photographic Dynamic Range curve one need to keep in mind that the data are standardized to a certain print size/viewing distance. Thus if two sensors of different resolution have the same Photographic Dynamic Range, the lower resolution sensor will be better when compared at pixel level (100% crop). So the low res sensors will inherently seem to have a disadvantage here which is not reflected at pixel level, but is so when viewing a print at the same same size and distance (similar magnification). Another consequence of this is when an Fx sensor image is cropped to DX format, the photographic dynamic range will be lower, as the image will be magnified more.

It will be interesting to see if the data can be reproduced. I supplied Bill data for the Nikon 1 AW1 sensor, and that changed some suspicious data he had received before which did not make sense as it differed markedly from the other 1 sensor curves.

Øivind, if I understand correctly, the dynamic range (DR) is the range between the highlight clipping point and the noise floor.  So, the low ISO DR may be more influenced (limited) by the clipping point, and the low ISO DR, noise floor.

[Frode]

CNX- D support for D5: http://downloadcenter.nikonimglib.com/en/download/sw/58.html

[Hermann]

CNX- D support for D5: http://downloadcenter.nikonimglib.com/en/download/sw/58.html

Thanks. And not just support for the D5, there are also some bug fixes and they finally added a retouch brush tool. CNX-D seems to be getting there, albeit slowly.

No support for the D500 yet though.

Hermann

[Tersn]

Thanks again, Andy, for all your images and your rather systematic approach!
As Frode noted, the NEF files can now be processed in CNX-D:

CNX- D support for D5: http://downloadcenter.nikonimglib.com/en/download/sw/58.html

Thanks, Frode. It could also be convenient to download the  new version of ViewNX-i as it contains several operations not (readily) available in CNX-D :
S-NXD___-010400MF-ALLIN-ALL___.dmg

[Andy]

CNX- D support for D5: http://downloadcenter.nikonimglib.com/en/download/sw/58.html

Thank you Frode. This was the moment I was waiting for
In the end, it is Tuesday ......

rgds, Andy

[Frode]

Thank you Frode. This was the moment I was waiting for
In the end, it is Tuesday ......

rgds, Andy

Thank YOU, Andy, for your "tutorial"!

 

[Ilkka Nissilä]

If I understand it correctly, Bill Claff uses mainly  real world NEF images sent by early owners of the cameras to estimate the dynamic range. I think the oscillation is probably not real, and may be a result of insufficient samples of images or the methodology used. But this is just my guess.

Maybe the D5 is optimized for high fps and uses fast readout of data from the sensor, leading to a situation where the dynamic range at low ISO is not as high as with some other cameras (Nikon has used slower readouts with limited fps rate to increase low ISO DR in the past, e.g. D300, D3X). It is a little disappointing that they couldn't meet D4s DR at base ISO. Of course we should wait for more testing before drawing conclusions.

I guess if the D500 uses similar sensor technology as the D5, it would likely fall behind the D7200 in low ISO DR, by some margin. But no data so far.

[Andy]

Ilkka,
the data Bill used for the D5 were coming from the one production camera I had access to over the weekend. As I have more than 30 different Nikon D-SLRs I've done the RAW data collection for Bill in the past accross the whole range and quite a few of the other Nikon DSLR measurements in his tool are based on these files.

For this camera, I sent Bill 3 data sets.
1) D5, set to 14bit lossless NEF
2) D5, set to 12bit lossless NEF (Bill has different values for 12bit, but this also served as cross check to the 14bit data set)
3) D4, set to 14bit lossless NEF (for crossreference, quality check of the setup used for this run)

While he is reasonable confident that the data is what it is, we agreed to re-run this exercise when I get my own copy later this week. Just to have more data points.

wrt to the emerging discussion about the "low" DR at low ISO levels the D5 provides:
1) We have to confirm with more datasets that this is indeed a pattern of the D5 camera type, and not only this particular sample
2) As many have stated previously, we are approaching fast physical limits (i.e. growth in quantum efficiency), so there will be trade-offs to be made in optimizing for a particular use case. Like the Nikon D810 is optimized for very high resolution and high DR at low ISO, it is not the tool for fast, low light circumstances. The sweet spot of the D5 might differ, but it can't be the champ in all corners of the potential usage spectrum a large customer base has.
3) The "merit" of the digital world is, that Nikon doesn't only has to rely on interviews and customer surveys what their users do, they can analyze millions of images of their target audience and actually see what people do and how their tools are being used for. What if (hypothetically), most D5 target users used the current generation cameras in JPEG mode only? (because speed of the workflow is of paramount importance, like in sports).
4) Or the professionals who work for Reuters have to comply with the JPEG only policy for photo journalists? Other news channel might mirror this as well. And Nikon basically optimizes the design for this world, where their users have to accomodate these emerging requirements?
5) It has to bee seen, what the addition of 4k video functionality contributed to the seemingly lower ISO DR performance. What if video has a much higher priority on the list of requirements Nikon heard from their core customer base, vs. superior DR at low ISO?
6) We don't know yet what Nikon has in their sleeves for a D810 successor. It might be the greatest camera for landscape shooters anybody could wish for. Nikon would probably not object if people would buy 2 cameras - it's called business strategy
7) We hadn't had a chance to compare the D5, D4 and D810 in such a particular scoped test. Will need to be done anyway to measure visible impact and outcomes. Would not be the first time, if internet folklore treats some individual measurements as the biggest issue on earth and most wouldn't even see anything in their results.
8.) Starting with the D3 and then proceeded with the D4, the D5 shows again a different pattern of (higher frequency) noise stored in the NEF files. Need to be checked, what the Nikon engineers decided to design in this regards and how it impacts different scenarios and workflow options.
9) An example how others see it: a quick comment by Iliah Borg

My personal approach:
Like you, stay "calm", look at it with curiosity when more info is available. But this is no different as with the other technical factors were the D5 behaves a bit differently, like WB consistency, exposure consistency, color fidelty, etc ...
In the end, a camera consists of so many interdependent components that a holistic view might be the most appropriate. It's the total "package" what counts.

rgds, Andy