The anti-aliased Euclidean distance transform is a recent development that redefines the distance transform concept, in particular the concept of precision and correctness, and has been shown to benefit certain applications. This paper presents and evaluates new versions of the anti-aliased distance transform. Our vector-based version simplifies the algorithm while providing a richer output in the form of vector data, with no measurable degradation in quality compared to the algorithm it is based on. Finally, we use a new method for measuring errors based on generating exact ground truth images.