My own A99-D600 ISO test

There are lots of renewed discussion regarding A99's ISO since DXO's RX1 & A99 rating published. I invited my Nikon buddy over to my house with his Boxing Day gift D600 (and 24-85 kit lens) for a quick AB test. We shot some static subject handheld (his lens has VR, I used Steady Shot), center point AF (AF locked to "Blue" text on bottle) and manual exposure. Afterwards we "zero" set all the RAW file, overriding WB with same value, then pixel pepping 100% on screen.

Here is the same picture differs in camera ISO:

At ISO 1600:

At ISO 3200:

At ISO 6400:

My observation:

1) D600 produced ever-so-slight brighter image overall, which is consistant to DPR's image data where you will find D600 use slight faster shutter speed.

2) At ISO 1600 there is very little difference, at ISO 3200 D600 is about 1/3-1/2 stop cleaner, at ISO 6400 D600 is about 2/3 to almsot 1 full stop cleaner. So we can see D600 gradually pulling ahead from (and including) ISO 3200 until it produces unacceptable images (we didn't test further).

3) Despite nosier images, A99 seems retain better details and having more "bite". Maybe there is focus  shift or difference in quality of lens, but this is also consistent with DPR's studio images where you can see D600 is noticeably softer.

My overall impression is that there is some truth in DXO's ISO number, but only after ISO 3200. In real life shooting, A99 is close enough to D600 (or any other top FF camera) up to until ISO 3200, after that you need to have good NR software. Therefore setting auto ISO limit to 3200 seems to be a good idea.