The renderer is particularly sensitive to contours, which is what allows it to pick out the edges of figures so cleanly and vividly in depth.

The ground of a scene is constructed from tesselated geometry, which inevitably have seams between them (due to rounding errors). Normally these would be blurred so as to be imperceptible, but we sharpen these contours as well, which become a series of lines receding from the camera — a “mistake” that paradoxically accentuates the illusion of depth.