I’ve played a bit before with twists in non-round tubes and using them in models. These test folds are my first attempt to combine those with my diagonal shifts, giving two square tubes that are offset from each other. The offsets of the two square tubes are the same in both test models.
In both versions (as in my previous tests), the points on each face of the square tube all fall along a parabola. Because of the offset between the tubes, the vertex of the parabola is aligned with the focus of the twist (the point at which the paper converges).
In the first version, I tried a simpler version of the math that places the focus of the twist at the same height as the vertex of the higher parabolas. In this version, the shortest segments of the twist go through the focus essentially horizontally, giving some vertical separation between the two tubes.
In the second version, the focus is shifted down to eliminate that vertical gap, or at least greatly reduce it. This gives something that looks a lot more like the twists I used in my diagonal shift series.