Well there is no literal curving. By curve I am referring to the entire RPM / degree advance trace.
Each pickup generates an integrated waveform( ramp waveform). These wave forms are compared to determine the advance timing between the 1900 and 3300 (sloped portion of the curve). Above 3300 the comparator will switch immediately causing the top end horizontal part of the curve. At the low end (1900 rpm and lower), the long ramp wave form is reset when the second pickup pulse comes in causing the low end horizontal part of the curve. It is a simple analog computer, where the output is a variable delayed pulse inversely proportional to the rpm.
There is a misconception that the pickup pulses work independently, or one is responsible for high rpms and the other for low. This is incorrect. You need both of them for a comparison.
Think of the pickup pulses as representing the limits of travel with the old point advance system. In this analogy, the sloped part of the curve, is that area where the plate on the point system is able move and advance using centrifugal force and spring tension.
There are ways of constructing an ignitor using only one pickup. Most often these days using a microprocessor or microcontroller. If you are going to use one pickup, you would use the 1st pickup at the 40deg point. You could use the whole pickup pulse, negative and positive signals, establish an approximate rpm based on the width of the pulse, then refine your advance, by measuring the time for one complete revolution. The key is to use the first pickup pulse. The "advance" or sloped part of the curve begins immediately after this pulse. An rpm above 3300 would force you to fire right away. You could use an algorithm or a map from this point on.