It draws a slightly tilted sine wave gradient (i=16y+x so atan(1/16) ≈ 3.6°) whose frequency increases until it starts to alias due ro the limited resolution (cf. Nyquist sampling theorem) and exhibit what’s essentially the wagon wheel effect [1]. Nice illustration of signal processing fundamentals!

[1] https://en.m.wikipedia.org/wiki/Wagon-wheel_effect