Nikon D5 - first impressions

[Andy]

Hi Andy
Many thanks for this wonderful review.
My D5 is available from the 24th, but as I'm away, I'll have to wait until the 31st, 
I'm wondering how you found the AF speed/acquisition in very low light given the -4EV,  and whether you had time to test the Auto AF fine tune?
Cheers
Marc

Marc, thanks for your kind remarks. Based on the short weekend experience, I assume you will enjoy your D5 when you will get it.

wrt to AF performance - Please consider that I didn't do any of the heavy duty AF tests (AF lock-on a football star and all the other player crossing constantly line of sight, Capturing an approaching bear while running away, etc ...)

For low light performance, the AF system can't provide magic. The D5 "suffers" the same problem like its immediate predecessors (D4, D3) that the low light sensitivity of the sensor improved much more (and faster) than the low light performance of the AF system. If it is really dark (below -2 EV) the AF struggles relatively often to find contrast edges, while the sensor is still able render a somehow useful image at ISO 50k or 100k. What makes significant difference is the AF setting in operation. If the AF is set to AF-S and a single AF field, then the limitation is much more severe, vs setting AF to "group AF" with multiple AF fields active. This improves the likelyhood that any of the active AF fields catches on and the AF CPU seems to be fast enough to use all active AF fields. Group AF should get you going until 50k or 100k.

In a nutshell: The darker it becomes, the more AF fields should be active.

Having said that: For my type of photography, the sweet spot of this camera is in the ISO 3200 to ISO 12800 range. A range which used to be an area of "under exception only" becomes now a kind of standard range to include in the set of choices if needed. No big worries in most circumstances. In this range, the AF is fast and precise, adding to my former comment that the camera comes around as a very "round package".

rgds, Andy

[Frank Fremerey]

Thank you Andy.

I still want D600 innards in a D5 style body and feature set.

The area below 3200 is most important to my kind of photography.


Hey. I know this motherboard! I am runnig is at this very second.

[Andy]

Frank,
see how important those motherboards are

wrt to D5:
I would be more cautious on asserting the < ISO 3200  performance as the weak spot. The D5 nailed the exposure and WB so often, that it not only took my attention how pleasing they looked, but it also reduces the effort for required post processing. It could very well be, that Nikon found that their typical D3/D4 audience didn't leverage the DR below ISO 1600 to the full extend and used it as an opportunity to improve the design in some other areas.

Having said that, I don't think the D5 should be the preferred camera for still and studio photography - here the D750/D810 are the better (and more cost effective) cameras. Used as a sidekick for whatever unexpected might come around in your days or weeks, I would assume that you might appreciate then the versatility of the D5.

As always, it depends on the use case, what the most appropriate tool is.

rgds, Andy

[Marc]

Marc, thanks for your kind remarks. Based on the short weekend experience, I assume you will enjoy your D5 when you will get it.

wrt to AF performance - Please consider that I didn't do any of the heavy duty AF tests (AF lock-on a football star and all the other player crossing constantly line of sight, Capturing an approaching bear while running away, etc ...)

For low light performance, the AF system can't provide magic. The D5 "suffers" the same problem like its immediate predecessors (D4, D3) that the low light sensitivity of the sensor improved much more (and faster) than the low light performance of the AF system. If it is really dark (below -2 EV) the AF struggles relatively often to find contrast edges, while the sensor is still able render a somehow useful image at ISO 50k or 100k. What makes significant difference is the AF setting in operation. If the AF is set to AF-S and a single AF field, then the limitation is much more severe, vs setting AF to "group AF" with multiple AF fields active. This improves the likelyhood that any of the active AF fields catches on and the AF CPU seems to be fast enough to use all active AF fields. Group AF should get you going until 50k or 100k.

In a nutshell: The darker it becomes, the more AF fields should be active.

Having said that: For my type of photography, the sweet spot of this camera is in the ISO 3200 to ISO 12800 range. A range which used to be an area of "under exception only" becomes now a kind of standard range to include in the set of choices if needed. No big worries in most circumstances. In this range, the AF is fast and precise, adding to my former comment that the camera comes around as a very "round package".

rgds, Andy

Thanks for the detailed reply,

I should have clarified my question on the AF -4EV,

it was in comparison to my D4s, perfectly understanding that we're talking about very low light here, just was interested in seeing if AF pickup was improved over the D4/D4s.
Like you my primary interest is in the ISO range of 3200-12800, whereby I shoot wildlife at the hours of sunrise/sunset where this range is really needed.
I'm hoping that can now stretch to 25600 when really needed, your results seem to indicate that possibility. 

Cheers
Marc

[Andy]

Marc,
the advantage you have against the city shooter is that dawn, even if it is dim, has a very balanced spectral distribution. This helps the sensor enormously. While in city- and incandescent light, quality of light becomes fast a challenge. ISO 25600 at dawn might be very well within the quality envelope you need. Unfortunately (or rather fortunately) you need to check for yourself if ok or not, too many factors need to be considered (i.e. how contrasty is the fur of the animal you want to photograph).

I can't claim that the D5 AF was always rock solid, it was sometimes too dark, but from a gut feeling perspective, it was a bit better than the D4 catching contrast lines. As said, just a gut feeling.

Not really a challenge for the D4 or D5, just a confirmation that the D5 AF works (as intended ). Each one out of a series of approx 20 shots. Slight crops, otherwise unchanged ooc JPEG. Resized.

[Bjørn Rørslett]

Andy's comment on spectral quality of light of different locations is spot on. Various artificial light sources are spectrally deficient as a rule.

Even my D3S can do clean ISO 51200 captures in broad daylight.

The problem with these ultra high ISO speeds usually is you either run out of aperture or shutter speed settings, or both. The experiment with D3S referred to above was only possible using 1/8000 sec and a lens capable of going to f/64. ISO 25600 would be more rational to use and the test pictures printed in the magazine article I wrote very virtually indistinguishable from those taken at much lower ISO values. Had I shot night scenes in a city, the outcome would be completely different.

[Andy]

Looking at my pictures and remembering under which conditions I shot them, I am still reflecting on the color reproduction of the D5.

Something is different vs. other cams.
The colors of shots with different lighting conditions are much more "consistent" - seen across the series and within the individual series.
There is less than expected deviation between the shots. Independent if it was shot with mixed light, or large sky parts were in an image, or cloud and overcast, or sunny, or different articial light sources. Hmmm.

Don't know yet where it comes from, but a few sources might contribute:
1) A new CFA with a different spectral response?
2) The high resolution exposure meter providing better detail and information what is actually in the image (i.e. a face)
3) The 24-70mm/2.8G VR lens with much better contrast in cross light situations (and lower flare/ghost). The "E" aperture with more precision?
4) new SW algorithms
5) a combination of 1-4 (most likely)

Anyway, most are pleasing (for me) from a color balance. I don't have any outlier with a completely wrong color reproduction. This might save some time in the future with post processing when the WB correction step isn't needed anymore.

Similar observation with exposure: There is more consistency across the different series I shot.

Interesting. This will be one thing I'd like to better understand. Need to wait for my camera ....

rgds,
Andy

[Frank Fremerey]

The first picture in #64 has such a lovely clarity and liveliness. I have witnessed a new quality of shots in the sports part of my newspaper and I ask myself if this has to do with new lenses or new cameras on the market.

I would describe it as: "relaxed vivid pop out", meaning that live action shots show qualites usually seen only in thoroughly crafted studio work with tons of post production.

Concerning the consistency of White Balance I said that in post #22 and I am happy that you see it similarly.

[Marc]

Marc,
the advantage you have against the city shooter is that dawn, even if it is dim, has a very balanced spectral distribution. This helps the sensor enormously. While in city- and incandescent light, quality of light becomes fast a challenge. ISO 25600 at dawn might be very well within the quality envelope you need. Unfortunately (or rather fortunately) you need to check for yourself if ok or not, too many factors need to be considered (i.e. how contrasty is the fur of the animal you want to photograph).

I can't claim that the D5 AF was always rock solid, it was sometimes too dark, but from a gut feeling perspective, it was a bit better than the D4 catching contrast lines. As said, just a gut feeling.

Not really a challenge for the D4 or D5, just a confirmation that the D5 AF works (as intended ). Each one out of a series of approx 20 shots. Slight crops, otherwise unchanged ooc JPEG. Resized.

Thanks again Andy

The quality of light is indeed a pertinent point.

Did you have any time to check out the Auto AF fine tune by any chance?

Cheers
Marc

[Andy]

Concerning the consistency of White Balance I said that in post #22 and I am happy that you see it similarly.

Frank,
you were ahead of me. I was hiding behind the D5 and was then still focussed to take photos

Now, sitting in front of the monitor, the WB pattern over hundreds of large thumbnails is clearly visible.

[Jan Anne]

Thanks for making NG a great and valuable place Andy, much appreciating the shared information.

And of course congrats on your latest acquisition

[Frank Fremerey]

Better AF with glacial speed primes like the 1.8/50G and the 1.8/85G ?

Please comment on that.

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?

[Erik Lund]

The D5 looks more and more like a pure Sports and PJ camrea - Would be amazing if they crunch out a D5X base ISO 64 or 32,,, 

[Jan Anne]

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

The opposite is true, the higher pixel cameras tend to have a wider DR range. Very simple, smaller pixels capture more detail at different levels of light so there's more data to work with.

When I moved from the D3s to the D810E I couldn't believe how much I could push the files in CNX2 both ways, HDR from one file was easily done. With the 12MP a7S I lost that capability but gained high ISO performance, now the a7RII with it's 42MP sensor gives me that stunning detail and DR range again but to my surprise also with a very good high ISO performance.

[Andy]

Jan Anne: Thank you. I thought that some info about the D5 might be of interest to the community (being curious by myself )

Frank:
Better AF with slow lenses?
Why not. It is not always speed what counts. Finding a contrast pattern in low light is more dependent on the AF system than the speed of the slow lens. What is more important is
a) that the lens is not decentralized, especially if it is a fast lens - this introduces a long list of issues on AF accuracy with cross-type AF sensor fields
b) the AF system is not spectral sensitive (i.e. the AF of the D3 and others from this period is impacted by different spectral distributions). The D5 seems to be fine in this sense what I've seen initially.

wrt to larger pixel and better DR:
I am sorry to say, that this saying (coming from the CCD days) was carried over to the CMOS world and seems to never die. Good DR is dependent on 2 major things: High Full well capacity and low read-out noise. The reason the D800 became such a success, because it excelled in both vs. the D3 and D4 (not surpassing everywhere, but getting very close). Both, D3 and D4 had a sensor design with an external (fast) A/D converter. This converter has to transform the analog values for all pixels serially. Given the high speed, high precision is very hard to achieve in an economical way. The Sony sensor Nikon used for the D7000 and the D800, has a lot of parallel A/D converters on chip. 2 benefits: 1) each one has a lot of time for the conversion (relatively speaking) and 2) what is integrated on a chip is easier to be calibrated. If interested, look up Sony's whitepapers about ramp wave conversion for more info. This provides excellent (=low) read noise levels for these type of sensors.

On the full well capacity, it is true that larger pixel can register more photons/electrons in the larger pixels. But there are fewer of them on a given sensor size. Let me suggest a different and simple metric, which served me well in the past to understand sensor performance. Full well capacity per full sensor area. According to sensorgen.info, the D4 has a full well capacity per pixel of 118.339, the D800 of 48818 and the D810 of 78803. Multiplied by the number of pixel on the FX chip gives you the following "light budget":
D4 = 1.912.812.661.760 registered photons FWC for the whole FX sensor
D800 = 1.764.883.957.760
D4s = 2.076.875.637.760
D810 = 2.848.911.272.960
D750 = 1.971.670.171.648

Given that equal output size should be considered in such comparison, this simple number helped me in the past to get a rough estimate how a sensor will perform relatively to others. Just an initial estimate.

For historic comparison:
D2H = 84.277.315.584
D3 = 610.194.772.992

(This is one contributing factor why you can pull so much out of the shadows of an D800/D810/D750/D600 image vs. their ancestors)

Don't know if the real ISO base went up for the D5 (vs. indicated ISO Setting). Bill Claff uses set ISO levels, DXOmark uses measured ISO levels and DXOmark numbers aren't out yet.


rgds, Andy