I recently posted a test model of a cone-based diagonal shift where the cone pointed inward toward the shift. This model is the reverse of that, where the cone instead points outward toward the shift.
The math of this variation is very similar to the inturned cone. For both variations, if the cone angle and the plane angle are the same, the convergence point is exactly the same distance from the top edge of the ellipse. But, when the paper comes back to being a cylinder, the effective distance of the horizontal shift is very different. When the cone is inturned, the top edge of the ellipse is shifted toward the center of the base cylinder, so the shift looks small. Here, since the cone is out-turned, the top edge of the ellipse instead sticks out quite a ways past the edge of the narrow base cylinder, so the amount of total shift horizontally looks much larger.
