Sadly Leaving Nikon!

[Akira]

I didn't call BSI 42MP sensor "game changer" because of the pixel count.  The BSI construction enables larger photosites without reducing the pixel count.  A sensor of 36MP and higher can go without AA filter, which can be nice for those who like and/or need sharpness.

[bjornthun]

The readout speed, ability to do on sensor PDAF, good low-ligh AF performance, video capabilities DR, the BSI construction makes Sony's 42mp sensor more wide angle friendly. There are many things that sum up to make this sensor a game changer.

[Akira]

Considering the poor low light AF capability, BSI 42MP sensor is not a game changer in terms of AF, regardless of the method (CDAF or PDAF).

[bjornthun]

Imaging resource, http://www.imaging-resource.com/PRODS/sony-a7r-ii/sony-a7r-ii-field-test-part-ii.htm, has tested the A7R II in low light and they say:
"Great low-light autofocus more often than not
And as I reported in my earlier field test, the Sony's autofocus system also handled low light admirably. Only a relatively few times did I hit a scene where it had to rack focus or failed to lock focus properly on the first try or two, and in every one it was a circumstance in which I'd expect most cameras to struggle. Autofocus is definitely a strength of this camera versus its predecessor."

Dpreview is also pleasantly surprised with the low-light AF of the A7R II, https://www.dpreview.com/reviews/sony-alpha-7r-ii/12.

One more from Dpreview, "Sony Alpha 7R II can match or beat DSLR low light AF performance", https://www.dpreview.com/articles/6884391759/sony-alpha-7r-ii-can-match-or-beat-dslr-low-light-af-performance.

[Roland Vink]

I remember Bjørn's review of D1.  Its "lowly" 2.7MP sensor (see the decimal point?) already surpassed the resolution of Fuji Velvia 50.

I have a print which is about 470 x 690mm (18.5 x 27") made from a drum scanned 35mm Velvia 50 slide. Sharpness is superb. I really wonder if a 2.7MP image would match it ...

[Bjørn Rørslett]

I once had an order of an image to be printed 8 by 4.5 m from a cropped D1X file. It was marvellous even up close but then I spent nearly an entire working day in the pre-press stage to get the maximum quality. I started out with only 5.5 MPix for the project.

All depends on the subject,  the light, and the viewing distance.

[Akira]

I don't think comparing AF speeds using a 35/1.4 lens on one and 50/1.4 on the other two is fair.  The focusing elements of 35/1.4 lens is much lighter and shorter travel, which makes easier to focus faster.

D750 (and even Canon 5D MKIII) proved their solid AF performances even slightly darker (the lenses used were 35/1.4, though):

https://www.dpreview.com/reviews/nikon-d750/7

I have a print which is about 470 x 690mm (18.5 x 27") made from a drum scanned 35mm Velvia 50 slide. Sharpness is superb. I really wonder if a 2.7MP image would match it ...

Here's Bjørn's review of D1 I was referring to.  Sorry that the film was not Velvia but Kodak E100VS.  And D1 didn't "outperformed" it.  My memory is not non-volatile.     But it was well understandable that 2.7MP is enough to compete with the finer grain films.

http://www.naturfotograf.com/D1_review.html#top

[Asle F]

Sometimes sharpness isn't all, and print size is not in question, but resolution is everything and the pictures is made for inspection on screen. Like the reindeer picture I posted recently in the pattern-thread. http://nikongear.net/revival/index.php/topic,2148.msg90651.html#msg90651
The picture was taken for counting the reindeers, and D1X was just at the limits for resolve that many animals in one picture. Most of the pictures I have of that herd, is not good enough. But this one is, thanks to even distribiution of the animals, and very good communication with the pilot of the plane. Even 2000 and 1000 reindeers was a challenge for D1X.
Now I use D800e for this, and the pictures are far better, and I can actually feel that on my eyes after hours of counting. The eyes are not that sore anymore.

[Bjørn Rørslett]

I'm not surprised by your observations, Asle.

The D1X had an almost film-like graininess to its files due to the weird layout of its sensor with the unavoidable consequences for their processing. Results could be very pleasing in tonality but sheer resolution suffered in many cases.

The D800 ushered in the 36 MPix class which allowed nearly 90% higher resolution than the D1X and its 10 MPix brothers. Plus, the D1X was just 10 MPix in theory, not in practice.

[richardHaw]

I'm not surprised by your observations, Asle.

The D1X had an almost film-like graininess to its files due to the weird layout of its sensor with the unavoidable consequences for their processing. Results could be very pleasing in tonality but sheer resolution suffered in many cases.

The D800 ushered in the 36 MPix class which allowed nearly 90% higher resolution than the D1X and its 10 MPix brothers. Plus, the D1X was just 10 MPix in theory, not in practice.

"weird layout"? 

I agree with what Asle had to say, designing for the charts isn't the most important thing.

[chambeshi]

A recent comparison of Fuji GFX and Hasselblad X1D

https://photographylife.com/fuji-gfx-50s-vs-hasselblad-x1d-50c/

[Bjørn Rørslett]

"weird layout"?
...

The D1X had rectangular photo sites so software had to interpolate at the very least one dimension. Either downsampling to 6 MPix, or preferably, upsampling to 10 MPix to make the pixels equivalent to a square area.

Its native pixel dimension was ~ 4000 x 1000. Approximate figures as the various RAW programs at the time had a competition how much they could squeeze out of the "dark frame" borders around the central active sensor area. Typical figures would be 4016 x 1020 or thereabout.

[Wannabebetter]

Arrival of a digital S3 would be keeping up the Nikon tradition for the 100 year anniversary

Dreams are for free. But I'd buy it without any hesitation should it materialise.

Nary a day goes by, that I don't muse on the feasibility of cannibalizing someones' digital box for the purpose of creating my own Franken-Nikon's monster at the expense of some hapless, thoroughly innocent, F or FE body. (If I could write code, this nightmare scenario would have surely come to past by now. Oh, the horror!) As though not in possession of my own faculties, I surrender to compulsion and break out in sweats whilst impulsively trawling eBay® parts listings for Nikon F backs or "as-is" FE hulks. The convulsions are maddening as they are debilitating -- and fortunately at that. After the shock treatments I rarely can recall what came over me, until... [shudders; shakes head impulsively; reaches for oatmeal cookie]

[Akira]

"weird layout"? 

Actually the CCD used in D1 series was a 10MP+ sensor made by Sony.  However, Nikon combined four photosites to use them as one pixel, which made the D1 a 2.7MP camera.  Nikon combined two photosites to make D1X a 6MP camera, but the combined two square photosites resulted in one rectangular pixel.  So, the raw data of D1X needed to be interpolated to expand vertically.

I noticed weird rectangular artifacts in larger prints from D1X data.

[Wannabebetter]

Somewhat is not a word to be used in Physics

That, and nor is "black-whole".