In terms of logits (log odds), the probabilities 0.1 and 0.9 are actually quite close together: −2.2 and +2.2, for a logit range of . The infinite range seems appropriate because the difference at the boundaries of probabilities (e.g. between 0.9 and 0.99) should get large, which can’t be captured with a finite range, like the cubic bezier.
In terms of logits (log odds), the probabilities 0.1 and 0.9 are actually quite close together: −2.2 and +2.2, for a logit range of