optimal number of cells for interpolation

Hi

I am using cellAlign to compare trajectories derived from kind of large datasets (~11k vs ~4k cells). The default option to perform initial interpolation is using 200 cells. I was wondering if you have tested the same on similarly large datasets, and if so, what is the way to infer the optimal number of cells to pick at that step.
Thanks