My experience is that you “had” it working, and then you broke it again…
I run nvidia surround with 3 screens, 7680×1440 resolution. And now it is IMPOSSIBLE to calibrate, as I always get the, “cannot detect eyes”… well the lastest version says somehting else.
If I break the surround setup, and run the 3 independant screens, it calibrates very fast on the center screen. when I then re-enable the surround it works most of the time until I reboot the computer…
I do not understand why it should be so hard have the tracking work on defined area of the screen. Which to me would be sort of expected behavior. Just like when you do your setup, and adjust the marking on the screen to the markings on the sensor, why not have a test to see how much movement you can track, up/down, left/right, and add the possibility to adjust these, this way you should be able to confine the tracking to a single screen, and the the extra screenarea in a double/tripple screen setup would be “untracked”, and then the same would go for those superwide screens. It is not perfect, but it would be ALOT better than what we have today.