Nice to see my X70 review is still useful and still being read! Thanks for quoting it. I just wanted to clarify on the fine convergence as I found out a bit more since I wrote that review.
I did manage to prove that there is indeed a hit to resolution when using it. But it depends on how much you shift it. For example, the fine convergence has a range of 16 steps to move it a whole pixel. In other words, 8 steps would be equal to 0.5 a pixel. If you change red vertically by 8 steps, and then display 1 pixel red grid test pattern, you will notice that there are now TWO lines of red sub pixels on every vertical row instead of 1! In other words you have just halved the resolution of red in the vertical orientation. Now of course, this is just one sub colour and only in one orientation. This is a worst case scenario, but I wanted to explain it so that you understood it does have an impact. When I was testing it with the X70, I was using smaller adjustments like 3 or 4 steps. As I mentioned in the review, this is a bit of trickery as the panels are not moving. I was searching for a technical explanation which I saw previously for the Sony implementation but I cannot find it now. But In mathematical terms, the less you shift, the less resolution loss you will have. For example 8 steps means halving the resolution in that direction, 4 steps would be a quarter. There are some strange colour distortions you will also see with certain test pattern also.
However, my guidance in my review is still the advice I would give today. Use the zone convergence adjustments at the boundaries of the image only. This way you can prevent visible fringing and is normally where misconvergence is worst. So start by adjusting full pixel adjustments across the whole screen and optimize the centre of the screen in this way. If it is not perfect, live with it! Red/Green convergence is MUCH more important than blue. Blue misconvergence almost had no effect on the perception of sharpness from seated. After optimizing the centre with whole pixel shifts, then use the fine convergence for the boundaries of the image. Trust me, the less you use this, the better. I am glad the feature is there, but it's not a magic bullet.