I am already having to consider the slightly unpalatable idea of having to get a 1Dx despite hating the huge body size, just to get that low-light performance.
I think the jury is still out on that low-light performance, and will remain out until the RAW files can be examined.
Regarding the 2-stop improvement in ISO noise compared to the 1D IV, Chuck Westfall stated, "To get the full two stops of improved performance youâ€™ll need to shoot in the JPEG format. Thatâ€™s not to say that high ISO performance isnâ€™t significantly better when shooting in RAW, because it absolutely is.
Ok, so two stops compared to the 1D IV. But that's for JPGs only. With current images, DxO promises (and delivers!) a 2-stop improvement in ISO noise compared to in-camera JPGs. So, the 1D X has a better in-camera jpg engine with better NR thanks to Digic5+. Whoop-de-do. I don't really care about improvements in JPG images because I'll be shooting in RAW.
But wait, Chuck said 'it absolutely is' significantly better in RAW. This speaks to the cleverness of the Canon marketing team in using the 1D IV as a baseline for the comparison. Of course, going from an APS-H sensor to a FF sensor means less ISO noise, because of the larger total area. As beaten to death, larger pixels than the 1DsIII/5DII means nothing, although the gapless microlenses will help. The 5DII beats the 1D IV on ISO noise, by about a 1/2-stop (which is slightly less than the improvement predicted by sensor size alone, and speaks to the improved on-sensor NR of the 1D IV). So, my guess is the when comparing the 1D X to the 1D IV RAW files, where it really matters, we'll see somewhere a bit less than 1 stop of ISO noise improvement.
Personally, I'm uninterested in the comparison to the 1D IV. What interests me is the comparison of the noise from the 1D X to the 5DII RAW files. Sadly, my guess is that we see less than 1/2-stop improvement there, probably closer to 1/3-stop. I think the math works out that way - 2-stops compared to 1D IV in JPG, which is 1 stop in RAW and 1 stop in the conversion. Of the RAW, 2/3-stop comes from the larger sensor, leaving 1/3-stop for tech improvement (gapless microlenses, on-sensor NR).
So, ISO 4000 on the 1D X will look ISO 3200 on the 5DII. Is that 'significant'? Sure. But it's nothing to write home about...and Martin, maybe not enough to justify the extra weight (although the low-light AF improvement is certainly also important).