Can someone explain to me (mathematically) /provide Links/Papers what erros can occure when calibration distance ≠ validation distance. Although I noticed that the accuracy is often gaze-angle dependent.

Cobe

I think it is because of the Least-Squares fit of reference points and Gaze points during calibration. As the gaze angle of each eye changes with distance - e.g. gaze gets more parallel at far distances the previous calculated mapping at a fixed distance doesn't provide a could fit anymore . am I right ?