Nikon D500 - first impressions

[Roland Vink]

You are right of course Andy, perhaps I should have written: any improvements in this area will be small using current technologies ...

[stenrasmussen]

I was lucky enough to get my hands on the D500 for a few hours.
As any respectable nerd does I went straight to comparing it to my D5300 and Df. All cameras set to same WB, no NR during PP, etc. The D500 is sparingly good at ISO 6400.
The viewfinder is big (almost feels FX), the shutter sound is very behaved, the responsiveness and not to forget, the AF is an absolutely killer! I will need this camera later this year.
LV rendering on the rear screen in dim/poor light is very (Canon) good. It is better than the Df's and buckets better than the D5300's.
Auto AF Finetune is a really cool and swift feature (although care must be taken when performing the task). I can see this as a very helpful way to get lenses tuned when I shoot stage performances under difficult lighting.
For you Øivind, I've also included a shot showing the focusing screen (via the mirror).

[simsurace]

Try 1640K


As for sensor performance, I'm don't know how efficient modern sensors are at converting photons into an image signal, but I believe it's in the region of 50% or so. Improving that to 100% gains only one stop ... any improvements in this area will be small, perhaps more will be gained from dynamic range as the signal:noise ratio improves etc

The quantum efficiency off almost all current cameras is around 50%. See http://www.sensorgen.info
Increasing this to 100% would lead to a 1stop increase, but since the dynamic range (using the SNR=1 definition, which is different from the one Bill Claff is using) is so close to 14 stops, one would need to increase the encoding to 16bit in order to see the full extent of the dynamic range increase.
I don't understand how the CFA could make up a 2 stop loss, wouldn't that show up in the QE?

[Andy]

Agree Roland.

Left Brussels airport earlier tonight. Sorry, didn't have any time to record more interesting images. Just a few, rushing to the gate.
(all NEF versions of the images can be accessed <here>. Just use the filenumber as an index)

1) ISO 1400 (you might be interested to play around with pushing the shadows)
2) ISO 8000
3) ISO 3600, had to correct for -1.3EV to avoid blown highlights. (He might have missed his flight
4) ISO 8000 and tele (yesterday, we had ISO 3200 and ww)
5) ISO 7200 and tele (yesterday, we had ISO 3200 and ww)

rgds, Andy

[stenrasmussen]

Here are the D500 and Df LV screens seen via my iPhone 6s.

[stenrasmussen]

and finally before I bed, a 5 stop push of an ISO 100 image.

[Akira]

Sten, I figure D5300 is doing a great job!   

[Roland Vink]

I don't understand how the CFA could make up a 2 stop loss, wouldn't that show up in the QE?

QE (Quantum Efficiency) is measured after all filtration. It simply measures how well the photosite (after filtration) converts the photons that reach it  into an imaging signal.

The CFA (colour filter array - usually Bayer RGBG) by definition must filter out a lot of light, at least one stop and probably closer to two. For example the red filter removes most green and all blue, so that's less light reaching the photosite.
Black and white sensors don't have the CFA so are usually a stop more sensitive.

Colour aware photosites would not need a CFA so sensitivity in theory could also be improved. Consider a stacked colour sensor -  the top layer reads the red part or the spectrum, the rest passes through, the green layer reads the green spectrum, and the remaining blue passes to the blue layer. The Foveon sensor used by Sigma works on this principle. However these are complicated to make, and efficiency so far is not very good...

[Øivind Tøien]

...
For you Øivind, I've also included a shot showing the focusing screen (via the mirror).

Thanks so much for remembering and posting Sten, it looks very much like they used the same catch and focusing screen as D7100/7200.

I received my Katzeye all matte (non-Optibright) screen for the D7100 before last weekend, and it is generally 1 stop brighter than the Canon S-Screen, at f/4 perhaps 1.3 stops, and overall with a hair more depth of field showing (insignificant) and better contrast. The rub here is that I managed to scratch it in the upper corner when installing, it sort of got stuck when initially got it in the wrong way in, and I must have grabbed to far in (due to to poor light as I had forgot to put on my headlamp, and then it was too late to go and get it),  or it otherwise touched something. I suspect that once a week or two has gone by it will be mostly forgotten though as I sort of have to look for it.

[Frank Fremerey]

This is another way to put it. But the factor is more like 1.5.

23,5 mm x 15,6 mm = 366,6 qmm
36,0 mm x 23,9 mm = 860,4 qmm

860,4 / 366,6 ~ 2,35

[Frank Fremerey]

I think it is the D500 features which are a huge improvement over D750 or D7200 and will be the reason for many photographers to upgrade. Here are a few - mostly speed & AF based because that was why I got the D500.

• 10 frames per second
• XQD card for speed
• buffering of 200 lossless compressed 14-bit raws (79 if uncompressed)

• new AF engine "multi-CAM 20K AF"
• fast live view AF
• 99 cross-type AF points
• 153 AF points stretching from right to left. (Amazing performance with this. Do focus-and-recompose with this.)
• auto auto-focus fine tuning

• new metering engine "180K RGB"
• new image engine EXPEED 5 (really really nice color)

• electronic 1st curtain
• button illumination
• (a huge amount of cool video stuff which I know nothing about)

You are absolutely right. IQ is as good as it gets with all of those Cams or as Simone put it: Very near to the theoretical limit (with current technology).

Reliable AF with good coverage and consitant colors even in mixed lighting, these are things that really lacked in the past.

If all these reports are true, the fifth generation is a real winner in these respects.

PS: "they are waitung for my cam..."

[Frank Fremerey]

Replacing the CFA with a non-filtering approach (like with a prisma) provides additional potential headroom in the future.

The Kodak Patent for rate counting /overflow counting instead of bucket counting can improve DR to anything. Remember (German): http://www.nikon-fotografie.de/vbulletin/off-topic/45885-beliebig-hohe-dynamik-am-chip.html

And here is the "missing link": http://www.google.ch/patents/US6069377

[Lorne]

I didn't know whether I should write 1,640,000 or 1.640.000.
So I just wrote 1640000. 

Sure would be nice it the industry would revert back to DIN in place of ISO. Then you could just write it as 63°.

[MFloyd]

Or 1'640'000, the way I usually spell "big numbers" 😊 And I'm also in favor to revert to DIN numbers.

[simsurace]

23,5 mm x 15,6 mm = 366,6 qmm
36,0 mm x 23,9 mm = 860,4 qmm

860,4 / 366,6 ~ 2,35

I understand, but magnification is a linear measure, so the additional magnification needed to bring the DX image to the same output size as the FX image is 1.5 times higher.