The number of points (20 vs 30) just smooths out the curve. You can test this out by generating multiple DLFs, and only changing the number of points. A smooth curve enables better interpolation between the system's natural frequencies when determining the effects of the applied load. I've always suggested 25 to 30 points, more than 50 is over-kill.

As to extending the frequency cut-off, yes this will change the resulting DLF. There is a good explanation in our newsletter article in the "November 1994" issue of M.E.N. If you don't have this, I'll request a link from Tech Support.