After deBayering, the max luminance resolution you'll get is generally taken to be about 75-80% of the sensor count in that direction. It's hard to give an exact figure since it tends to roll off and it depends at what point you draw the line.Maybe someone can answer this for me. I know the RED At 4K after the bayer process is really like 3.4-3.6k. So if the Arri is essentially an upscale from 3.2K (correct me if this is wrong) after the bayer process is it essentially 2.5k?
For chrominance the resolution is much less, and maybe 50% is a good estimate though it varies depending on the colour. (Think about it. For a 1920x1080 HD Bayer sensor, you have 960x540 red and blue photosites. So point it at a saturated red or blue image, and what do you think is the maximum it will resolve.....?) There's more to it than that, concerning aliasing, and particularly colour aliasing, but we'll leave it at that.
The funny thing really is that it was Arri in particular who always (very rightly) stressed the importance of sensor resolution being much higher than output resolution. As the nofilmschool link says:
That seems to have been brushed under the carpet so they can use the 4k term in their marketing.This is why single-sensor cameras wanting the highest fidelity have always started with recording more resolution for later (like ARRI's original 2.8K mode for 2K/1080p).
As for them saying nobody will see the difference with 4k, I think I read a very good paper from them going into the maths with the physics of the human eye. I think the (theoretical) conclusion was that for an average cinema, the back 50% won't be able to see any difference, sit nearer the front than midway, and 4k will make a difference.